Post by: Angle on September 24, 2013, 08:43:56 pm
This isn't a very good explanation, but I'm not sure how else to put it. Any questions?
Update- Now with collaborative Google Doc (https://docs.google.com/document/d/1CjERoM9jefThEU35WAVT9IALjrYhhYgcAHbBvfgjWbs/edit)!
Update: Github (https://github.com/organizations/Agora-Project)Added
Update: We now have a live site for further testing. You can view it here (https://agora-2866.nodechef.com/).
Post by: Urist McScoopbeard on September 24, 2013, 08:58:45 pm
However, despite that, I like the idea, it feels like it could be good, maybe go more in-depth? Or draw a picture describing the layout of a page.
Post by: Angle on September 24, 2013, 09:06:24 pm
So like a debate in text form? Sounds good, But it suffers from the same problems normal forums do, i.e. after a while there are simply too many posts for the USERS to keep track of, unless you can sort them into generalized 'sides' within a thread, even then it seems like a post might not be noticed.The beauty of this idea is that no given user has to keep track of all the posts, just the ones they're commenting on. Let's say you present an idea, and several people all reply to it. I only need to read the given reply that I'm commenting on. There will of course, be problems with duplicated posts, where one person posts something and then another posts the same thing. I figure that can be solved via moderation of some sort.
This isn't for a 1 to 1 conversation, this is for exploring large, complicated ideas with large numbers of people.
Also, how would an active conversation take place? Unless you mean to use some sort of text-chat in conjunction with these posts intricately interwoven and orginized.
However, despite that, I like the idea, it feels like it could be good, maybe go more in-depth? Or draw a picture describing the layout of a page.
Sure, lemme fire up paint.
(http://img.ie/images/jho3n.png) (http://img.ie/)
Post by: Darkmere on September 24, 2013, 09:40:31 pm
Also, it kinda sounds like a forum crossed with a wiki.
Post by: freeformschooler on September 24, 2013, 09:42:20 pm
Post by: Angle on September 24, 2013, 09:47:45 pm
There would still need to be some limiting of how big any single thread got, but if you limit things to the original topic, and split off any large derails as their own threads, it should be workable.
It would take a genius feat of interface design.
Nosense, I'll show you! *Opens paint again*
Spoiler: Interface (click to show/hide)
Obviously, this is just a rough draft. I'd need to actually have this made and test it out for a period of time in order to perfect it.
Post by: Darkmere on September 24, 2013, 10:03:05 pm
It's not the idea I have a problem with, but the actual implementation worries me.
Post by: Angle on September 24, 2013, 10:13:37 pm
Post by: freeformschooler on September 24, 2013, 10:33:34 pm
Might not want to jump on the kickstarter idea immediately - I think it's something that, with some jquery/sql/other web type coding you could probably learn to make yourself!
Post by: Angle on September 24, 2013, 10:39:48 pm
Post by: Anvilfolk on September 24, 2013, 10:44:23 pm
This is actually totally related to my MSc thesis and subsequent avenue of research my university in Portugal's been pursuing for a while. We've got some nice theoretical/mathematical foundations for proper social argumentation. We currently have a simple but efficient algorithm for number crunching the solutions to our debating structures, but we're missing proper implementation - and are actively looking for it, but none of us have time.
Here are the links for a couple of our papers on the subject:
Social Abstract Argumentation (http://centria.di.fct.unl.pt/~jleite/papers/ijcai11.pdf)
Extending social abstract argumentation with votes on attacks (http://centria.di.fct.unl.pt/~jleite/papers/tafa13.pdf)
If you are serious about this, and would like to get involved in this, PM me and we can go from there :)
Post by: Draxis on September 24, 2013, 10:48:46 pm
(http://img.ie/wmum1.gif)
Post by: Angle on September 24, 2013, 10:51:45 pm
Post by: Anvilfolk on September 24, 2013, 11:10:07 pm
What we've proposed is having each point, or argument, be one of those nodes. If one argument undermines another, you draw an arrow between the two, showing an "attack".
For a forum, I imagine that you'd easily highlight what piece of text you consider to be arguments in your post, and then possibly drag them onto other arguments to specify attacks between arguments.
From an interface perspective, you could have multiple zoom levels. One with just nodes, one with a little bit of text, and "full view". Of course, mousing over could show you more text and statistics and all sorts of other cool stuff :)
Post by: Angle on September 24, 2013, 11:19:20 pm
You shouldn't really have one post equalling one node. The reason is simple: one post can make a LOT of points.
For this reason, much of my theorizing has been going towards figuring out a format that would take good advantage of this system. in my format, I'd probably have it so that each argument IS it's own separate post. This keeps things neat and easy to navigate, and keeps things from getting gummed up large walls of impenetrable text.
Post by: Draxis on September 24, 2013, 11:27:11 pm
You shouldn't really have one post equalling one node. The reason is simple: one post can make a LOT of points.I don't see the conflict there; as long as users only make one argument (here meaning an idea trying to be communicated) - which could be based on multiple points (bases of that argument) - per post, the same linking system between posts could be used.
What we've proposed is having each point, or argument, be one of those nodes. If one argument undermines another, you draw an arrow between the two, showing an "attack".
For a forum, I imagine that you'd easily highlight what piece of text you consider to be arguments in your post, and then possibly drag them onto other arguments to specify attacks between arguments.
The zoom levels would be a neat way to help the clutter; the idea also made me realize that this forum could be a good form for all the touchscreens and such that are being used today - something to keep in mind if it ever does come to a kickstarter.
The forum will need a very robust threading system; posts should be able to be linked between threads, or even whole chains copied over if relevant. A good splitting mechanism will also be important, to keep the threads from getting too bloated like a lot of the GD ones do. Whether this is done by thread starters, moderators, or the participants, will be an important decision to make. I lean towards the participants being able to manipulate threads, based on some sort of concensus system, but that may turn out to be to slow or manipulable.
Post by: Anvilfolk on September 24, 2013, 11:38:36 pm
Visualisation and the interface is going to be a HUGE part of it. For instance, how do you represent threads of conversation? A flat structure isn't interesting at all for a forum like Bay12. You need to keep track of that. How do you visualise time? Posts are linear, for instance... meaning there's a nice flow. How is that represented when everything is now nodes?
The forum will need a very robust threading system; posts should be able to be linked between threads, or even whole chains copied over if relevant. A good splitting mechanism will also be important, to keep the threads from getting too bloated like a lot of the GD ones do. Whether this is done by thread starters, moderators, or the participants, will be an important decision to make. I lean towards the participants being able to manipulate threads, based on some sort of concensus system, but that may turn out to be to slow or manipulable.
That's part of the beauty of what we've proposed, for two reasons:
1) Because everything is flat (there's "just" nodes/arguments), you can reference everything from everything. You can keep track of which arguments belong to which thread, and restrict the view to that, like a more traditional forum. Of course, you might have incoming and outgoing attacks from the thread, but they'll be temporarily greyed out so you can follow the thread itself. It should be easier to get a temporal visualisation of arguments within a single thread as well. Because the underlying structure is a simple directed graph, the UI is also not overly complicated to implement and play with. I'm sure there's already tons of graph-based algorithms and stuff that can be used.
2) By allowing people to vote on arguments and attacks, dumb posts or attacks that people specified for some reason but don't really make sense are easy to handle. You just downvote them, and they essentially become irrelevant.
Post by: Draxis on September 24, 2013, 11:54:13 pm
I was imagining the structure to be stored in the server as a list of posts, chronologically, and that the graph program would interpret them into a graph using information in the stored posts like links (attacks, but also supports or references) to others, and maybe a list of threads - each containing any posts put into this thread during the posting process. The graph would draw a thread as all the posts in it, with their references to one another shown, and the references to posts not in this thread shown in a different color or something. The threads would be only an organizational tool used to help follow a given topic. Is something like that what you are talking about?
Post by: Urist McScoopbeard on September 24, 2013, 11:56:48 pm
Especially in this system, just because one person is in the minority puts them at a disadvantage because their post can be downvoted because the majority thinks its stupid, not everyone is loving and open-minded on the internet after all.
Post by: lue on September 25, 2013, 12:02:40 am
I was thinking, one thing that would improve both classic forums (like this one) and your idea would be some sort of special summary page thing, that summarizes the content of the thread. It would, naturally, have to be maintained by a human or a group of them. You could already do this with a first or second post in a thread reserved for updates, but that relies on the thread creator to stick around until the thread is utterly forgotten.
Wait, what do you mean by 'flat' and 'thread'?
I think some usages of the word "flat" here refer to the way the data is stored; putting all the posts into one or many text files would be a flat data structure, while putting them in a database structure (things like MySQL) would not be "flat". I could be wrong though, so perhaps those talking can correct me if I'm wrong.
Post by: Angle on September 25, 2013, 12:06:59 am
In terms of a summary page, I was thinking that some kind of overview that only shows the title's of the posts could be used.
Post by: Anvilfolk on September 25, 2013, 12:12:08 am
I was imagining the structure to be stored in the server as a list of posts, chronologically, and that the graph program would interpret them into a graph using information in the stored posts like links (attacks, but also supports or references) to others, and maybe a list of threads - each containing any posts put into this thread during the posting process. The graph would draw a thread as all the posts in it, with their references to one another shown, and the references to posts not in this thread shown in a different color or something. The threads would be only an organizational tool used to help follow a given topic. Is something like that what you are talking about?
Yeah, that's what I mean. Sorry, it's getting late here :)
I was still imagining a post to be longer, like the ones we are writing here, though that is perhaps counter-productive for proper debates. Either way, eventually, it all becomes nodes in a graph. That's what we ultimately use to compute an output. What you can do is keep meta-data about which thread the post was added to, and so still maintain a forum-structure for organisational purposes.
Up/downvoting is a bad system on the internet, entirely too easy to troll it and abuse it. Especially where a post might only get a few votes.
Especially in this system, just because one person is in the minority puts them at a disadvantage because their post can be downvoted because the majority thinks its stupid, not everyone is loving and open-minded on the internet after all.
Think of it this way. The outcome of a debate is the result of two things: 1) popular opinion and 2) debate structure. As a result of 1), a post in Bay12 might be super upvoted, whereas the same post, say, in Reddit, would get downvoted. It's a matter of context. This is taken into account. As a result of 2), even if a post has lots of crowd support (more upvotes, less downvotes), if it is logically attacked by another post with a similar crowd support, it will be logically weaker.
In your example of a debate between a minority and the majority, the outcome will necessarily be skewed towards the opinion of the majority. Heck, that's how democracy and elections work, just at a larger scale. Having only registered accounts add votes makes it so bots and so forth can't troll it too hard. Also seems to work nicely on reddit, slashdot, etc.
I think some usages of the word "flat" here refer to the way the data is stored; putting all the posts into one or many text files would be a flat data structure, while putting them in a database structure (things like MySQL) would not be "flat". I could be wrong though, so perhaps those talking can correct me if I'm wrong.
Forums typically don't have as much flatness since there's a forum->subforum->thread-post hierarchy. Databases do a good job of representing this. For computing a solution to a debate, we use a flat structure in the sense that there's only arguments. There's no threads or forums or anything. But like I said before, that should definitely exist as metadata. It makes it easier to browse and search and visualise.
But Angle ninja'd me and explained it better! I really should get to bed ;)
Post by: Angle on September 25, 2013, 12:15:08 am
Post by: Urist McScoopbeard on September 25, 2013, 01:58:21 am
Also, in keeping with this debate-centred discussion, I think we need actual different posts, i.e. topic posts = you bring up an issue, no arguement yet only pros and cons listed. Then discussion posts, where you might argue for either side of the topic. Finally 'attacking' posts, basically your responding to other posts, you should be able to 'attack' any post whether it be a topic directly, a discussion post of a topic or any othef 'attacking' post.
Its too late for the pc, but im tempted to design a layout of this, i figure tht each topic post appears as a green dot, which when moused over will display a tree of attacking posts in red and discussion posts in blue as well as the actual post itself. When you click the dot, this information stays on the screen instead of disappearing after your mouse leaves the dot as if you moused over it. Now you can read a post, and see a tree of posts 'attacking' or 'discussing' it. Again, if you mouse over a dot in this tree it expands to show you a post and list of posts 'attacking' it. Etc. ever onwards.
To be clear all of this is contained in one 'thread' that theoretically covers on broad topic, the topic posts are more specific aspects of the thread topic. Large issues might have several dozen topic posts, while smaller ones might only have 2-3, etc.
Furthermore, the OP should be able to control topics, lest things get out of hand or inappropiate, he/she should be able to lock any topic post, as in its still viewable but can long longer be contributed to. I think that the OP should not have too much, if any, control over discussion and 'attacking posts'.
Post by: Eagleon on September 25, 2013, 02:22:32 am
Post by: Draxis on September 25, 2013, 06:00:29 am
Think of it this way. The outcome of a debate is the result of two things: 1) popular opinion and 2) debate structure. As a result of 1), a post in Bay12 might be super upvoted, whereas the same post, say, in Reddit, would get downvoted. It's a matter of context. This is taken into account. As a result of 2), even if a post has lots of crowd support (more upvotes, less downvotes), if it is logically attacked by another post with a similar crowd support, it will be logically weaker.I'm not comfortable with an up/down voting system, because it will lead to people pursuing votes rather than effective arguments, and groups who share an opinion downvoting anyone who opposes them. You can see it in places like Youtube or Reddit, although those are extreme examples - the community is a large factor here, but it just opens the possibility for abuse.
In your example of a debate between a minority and the majority, the outcome will necessarily be skewed towards the opinion of the majority. Heck, that's how democracy and elections work, just at a larger scale. Having only registered accounts add votes makes it so bots and so forth can't troll it too hard. Also seems to work nicely on reddit, slashdot, etc.
A system with a number of different 'ratings' you could use, such as "Agree","Disagree", "Good points", "Spam", "Need Source" and so forth would be better, as people who have problems with the post could be more specific about them without just attacking the whole post, and people could just vote if they agree or disagree in general - which, while just about useless, is far better than them making "+1" posts, or downvoting the whole post because it makes different points than them.
Also, in keeping with this debate-centred discussion, I think we need actual different posts, i.e. topic posts = you bring up an issue, no arguement yet only pros and cons listed. Then discussion posts, where you might argue for either side of the topic. Finally 'attacking' posts, basically your responding to other posts, you should be able to 'attack' any post whether it be a topic directly, a discussion post of a topic or any othef 'attacking' post.The idea about multiple post types seems promising, but that could just end up like a less structured version of a normal forum, if everyone used only discussion posts - any they would, because it would be easier that argument posts, unless the community was strongly against it. Since the whole point of the model is to facilitate proper arguments, that is undesirable; discussion posts will need to be implemented carefully. Similarly, topic posts are a hazard: unless they can only start threads, people will just post their points in them and wander off, without bothering to really participate in the arguments. I would say that having them would cause more trouble than they would be worth, and it woud be better to just let people use argument posts for that.
...
Furthermore, the OP should be able to control topics, lest things get out of hand or inappropiate, he/she should be able to lock any topic post, as in its still viewable but can long longer be contributed to. I think that the OP should not have too much, if any, control over discussion and 'attacking posts'.
Post by: Angle on September 25, 2013, 10:13:09 am
The idea about multiple post types seems promising, but that could just end up like a less structured version of a normal forum, if everyone used only discussion posts - any they would, because it would be easier that argument posts, unless the community was strongly against it. Since the whole point of the model is to facilitate proper arguments, that is undesirable; discussion posts will need to be implemented carefully. Similarly, topic posts are a hazard: unless they can only start threads, people will just post their points in them and wander off, without bothering to really participate in the arguments. I would say that having them would cause more trouble than they would be worth, and it woud be better to just let people use argument posts for that.
Again, this is why I'm putting a lot of effort into thinking up a good set of rules for said forum. If it's against the rules to have your arguments in a discussion post, and the mods go and delete them when you do that, then people will make goo and proper use of arguments. This does leave the moderators with a lot of power - for that purpose, I was considering making a system that would log anything the moderators do, so that people could keep track of them and point out if they did anything wrong.
Post by: Anvilfolk on September 25, 2013, 10:18:23 am
If you guys don't mind I'll just start mentioning about Social Abstract Argumentation (SAA), which is what we developed in my research group. Makes it easier to point out what I'm referring to :) The theory behind SAA can very easily be changed so that we accept slightly more complex crowd voting. What does need to happen is that all the votes eventually get you a value in [0, 1]. For instance, an "agree" vote could be value 1, "disagree" could be value 0, "spam" could be value 0, "need source" could be value 0.5, and then you'd average all those votes out.
How you go from plain text to a formal argumentation context (as a directed graph) is still very much an open question. I don't think we will ever be at a point where you can automatically mine that information, since natural language is so hard to make sense of with computers. This means that ultimately it'll be some form of user that has to say what is an argument and what is an attack.
The Abstract in SAA refers to the fact that we distance ourselves from that problem. We work exclusively with nodes and arrows (and votes). It is not clear to us that you should have different argument types at all - what would be their purpose? How do they differ? SAA is nice because it is super simple and to the point. An argument is some sort of statement that may or may not undermine other statements, and that may or may not be undermined by other statements. This notion of "undermining" induces the attack relation.
The Social part of SAA means that the community can self-regulate. In other words, if someone goes into a nice, mature community and starts spewing nonsense as arguments, the community has the power to downvote that opinion into irrelevance.
I think it's also important to realise that outcomes to all debates are ultimately always dependent on the context, whether in "real-life" or online. I mean, talk about gay marriage in a southern US state or in a metropolitan area, and the result will be reversed. Talk about it in a US wide context, and it might be 50/50. An atheist in a religious forum and a believer in an atheist forum WOULD likely be downvoted simply because people don't agree with their opinions. That's the way the world works.
If you accept that the formal outcome of a debate depends on its context and participating community, your complaints aren't really bugs of the system, they're features ;)
To respond to Angle's more recent post. You *could* simply create an argument that states another argument is Non-sequitur, ad hominem, or strawman, and have that attack the original flawed argument. If people agree with that, using !!LOGIC!!, SAA would reduce the strength of the flawed argument until it became too weak to have any real impact. Heck, if people realise that the argument is flawed, they can simply downvote it.
Overly complicated systems restrict usage. We'd really like to have as many people as possible using this, right? Arguments, attacks and voting can pretty much represent any situation that we've been able to think of for quite a while now. That doesn't mean extensions aren't possible, but before making the system more complex and less usable, those changes should be strongly justified.
Post by: Eagleon on September 25, 2013, 01:07:27 pm
How you go from plain text to a formal argumentation context (as a directed graph) is still very much an open question. I don't think we will ever be at a point where you can automatically mine that information, since natural language is so hard to make sense of with computers. This means that ultimately it'll be some form of user that has to say what is an argument and what is an attack.This is actually why I'm very interested in this concept - it pushes the user to formalize the relations between what they're talking about and what it is in response to, and allows a much nicer picture of who is talking to who and why than you can get with a conventional forum, while still allowing as many people to participate as possible. It would highlight redundancy nicely if you did include some machine learning techniques into the mix, if you made the relationships inferred visible to the user.
More broadly for my purposes, I'd like to combine it with a research/citation database to also allow someone to visualize the relations between a conversation and the technical details it might be relying on, based on some sort of learned confidence rating, so that you can generate new links to conversation outside what the user is expecting to see. All to maybe spark some interesting insights outside their comfort zone as they use the system.
Post by: Killjoy on September 25, 2013, 02:05:36 pm
It is actually not that difficult to set up something like this.
However, presenting it in a way that facilitates these digraph discussion features is much harder.
You could show the parent threads appended on top. But I don't really see any reason not just to use quoting, unless you want to preserve context, which is not an issue on modern boards anyway since you can simple link the post where the quote comes from.
This does however give me an idea. Why not make sort of a discussion graph or web. You can't start a new discussion topic per se, but you can always derive discussions from other discussions. Meaning any topic discussion thread could be traversed back to the origin. The origin would be sort of a hello world, opting to discuss this type of fora or something.
What would be even more valuable would be if any new topic had to actually discuss the parent topics in some way or form. I doubt however that it is really possible to moderate and enforce such rules, even if a dataset with this rule would be extremely valuable.
Farming this graph would make for some very interesting data.
I don't mind setting up a little web service like this if anyone is interested. If anything it gives me something to do today.
Post by: GalenEvil on September 25, 2013, 11:39:48 pm
Post by: Angle on September 26, 2013, 12:57:10 am
I don't mind setting up a little web service like this if anyone is interested. If anything it gives me something to do today.
That'd be cool, although we don't have any of it programmed yet.
Post by: Anvilfolk on September 26, 2013, 09:36:34 am
Eagleon: yeah, what if we could take research papers from known websites, and each paper would be a node? I doubt it would be possible to effectively mine each individual paper for arguments, but even having a paper represented as a node would be enough to denote that a research paper contradicts some proposed point. That could work out really nicely as a basic system!
Post by: Virex on September 26, 2013, 11:31:19 am
This does however give me an idea. Why not make sort of a discussion graph or web. You can't start a new discussion topic per se, but you can always derive discussions from other discussions. Meaning any topic discussion thread could be traversed back to the origin. The origin would be sort of a hello world, opting to discuss this type of fora or something.
What would be even more valuable would be if any new topic had to actually discuss the parent topics in some way or form. I doubt however that it is really possible to moderate and enforce such rules, even if a dataset with this rule would be extremely valuable.
Farming this graph would make for some very interesting data.
I don't mind setting up a little web service like this if anyone is interested. If anything it gives me something to do today.
You mean something that looks like a collaborative mind map (http://etjune.brochure.examtime.com/files/2012/11/How-to-create-a-mind-map-mindmap.jpg)?
Eagleon: yeah, what if we could take research papers from known websites, and each paper would be a node? I doubt it would be possible to effectively mine each individual paper for arguments, but even having a paper represented as a node would be enough to denote that a research paper contradicts some proposed point. That could work out really nicely as a basic system!
Web of Science already has an option to create a map of how papers are linked to each other. Remember however that a reference in a paper doesn't really mean that paper #2 actually addresses paper #1's points. For example, references are also used to show the state of the art (Bla Blason et al. synthesized poly(randomene) via a 6 step route and Bla Blabla et al. synthesized iso-poly(randomdiene) using a similar method).
Post by: Eagleon on September 26, 2013, 12:19:19 pm
Web of Science
Spoiler: derail (click to show/hide)
I'm willing to bet they had the start of the same idea, only with big goofy money symbols over their eyes - I'd like to get the same thing going for free, and give tools/information to make the transition of diving into the papers and reference system easier for inventors, engineers, independent researchers, highschool students, etc. Expanding it to patents would be the next obvious step, and then integrating it with research/engineering software or making our own. And if we had that, and a discussion medium more suited to education, collaborative projects between researchers and enthusiasts like EyeWire might become much more common. I think that just having that option and support available would give more incentive to open up publications in the future, as well.Post by: Angle on September 26, 2013, 01:06:08 pm
Post by: Eagleon on September 26, 2013, 01:30:21 pm
I'm Planning on making all my stuff free and open source.Hey, if you can find a way to make money off of it (maybe by looking for some capital to hire a professional designer and web developer, and then licensing the framework you've made to the people that want to use it) I would encourage you to do so. Even as a non-profit, organizing this way can actually act to send an idea to the top if it's really good - example, Wikipedia, which relies entirely on donations, still has a well-paid staff because what they're doing is so appealing to everyone.
For my idea I plan to be focusing on rather specific engineering goals to generate things we can license, and that still leaves plenty of room to bring scientific literacy to people like myself. You still deserve to make a living off of what you do if you can do it well, the question is only do you want to do this (or anything else) for a living.
Post by: Angle on September 26, 2013, 07:22:33 pm
1) Details on exactly what I'm trying to accomplish.
2) Background research. People who have done similar things, and the differences between what they've done and what I'll do.
3) Details on how I'll spend the money. Who I'll hire, what they'll do, etc.
You still deserve to make a living off of what you do if you can do it well, the question is only do you want to do this (or anything else) for a living.
I don't see any way to make a living off of just this project without undermining the purposes for which I do it.
Post by: Eagleon on September 26, 2013, 11:12:35 pm
I don't see any way to make a living off of just this project without undermining the purposes for which I do it.Wikipedia isn't the only wiki out there now. Once upon a time there was no Wikipedia. They developed their idea as a charity splitting off from a business, seeking donations, grants, sponsorship, merchandising, etc, and they make enough from that to support their admittedly small staff with competitive salaries (or the appeal of maintaining it for less keeps their employees there - also possible) It's sparked hundreds of other similar systems, shaping and boosting interest for the internet as a source of information on a global scale.
(It's a little ironic I got all of this information from Wikipedia)
Not saying you have a Wiki-level concept on your hands with this project alone (I don't know either way, I think it's awesome), but it can only do you good to think about how you can feed yourself and help save the world, since those are the ideas that typically run the furthest.
---
Still working on my implementation. Are there libraries anyone else is familiar with that I could base the node/connection model on? Mostly to avoid confusion as it's inevitably ported away from Java.
Right now I'm working on the assumption that the data contained in each node is expected to be asked to split at any time into new nodes that are composed of more refined bits of content (for instance, part of a conversation splitting off because someone has specified that particular portion of it as a target for their argument), so I've made an abstract 'content' class that expects this to be implemented for whatever data it contains.
Nodes contain Content-type objects, the Content class is subclassed into things like Message and Hyperlink and Youtube objects, and those would be converted into references pointing to parts of a database or assets off-site. Later we can store in the connection objects how the parent and its child node are related, so that we can work that information into the UI.
I could introduce something strange from one of my own projects and handle connections as Content. That way you could point an argument at an argument in your interface, and it wouldn't break anything - just generate a new node between all three. If you're sufficiently boggled, think of each connection as a special type of node containing data pointing to and from two other nodes - now add one more connection pointing at it. You now have something like a transistor junction. I don't know how useful that would be for this, but regardless, since everything is very abstract right now, it'd be easy to work out two or three versions with different features and topology.
Also, if all of this is alternately gibberish or clueless, I apologize - school is expensive :P
Post by: Sensei on September 27, 2013, 03:26:02 am
Post by: Anvilfolk on September 27, 2013, 09:08:49 am
I've seen a forum that worked this way... while it didn't help that it was all in one narrow column of posts, putting replies to specific posts next to those posts made it nigh impossible to tell in what order the conversation happened without carefully scrutinizing each and every post date. Furthermore, it made it difficult to discern new posts from old posts without carefully reading each.
Sounds like a kind of mailing list. I've seen those systems for tech support stuff and older websites. Not a big fan of it, and definitely an example of what to try to avoid, although a debate structure IS going to be very similar.
---
Eagleon: I wouldn't bother too much with content. I think people can warm up to the idea that they're no longer writing posts, but simply proposing small, indivisible arguments for a specific purpose (to add an attack to an argument they disagree with).
---
Alright, so I need to put together a proposal for a kickstarter. The way I see it, I need
1) Details on exactly what I'm trying to accomplish.
2) Background research. People who have done similar things, and the differences between what they've done and what I'll do.
Angle: I believe whether my research group ends up participating more directly or not, !!SCIENCE!! is meant to be used (for !!FUN!! ?), so I would again suggest you read our papers on the subject because they seem to be so close to what you want to achieve. Some of the theory can be easily adapted and fine-tuned to different purposes, but some of it can't. The articles also mention quite a few other similar systems, developed within academia or just plain websites, like www.debate.org. Either way, it's probably worth a read to help with both these questions.
I would just say this: most systems essentially let people vote on who they want to win. That's it. It's simple, and it uses the notion of crowd support, but it does not take into account the logic behind the debate. The system you're proposing has an underlying structure that represents logic that you can exploit to obtain results to debates. SAA gives you outcomes of debates that tell you the degree to which each and every argument is accepted. It's not just "side A wins and side B loses". Some arguments from A might be accepted, some might not, some might be accepted so-so... and this notion of acceptance is based on both crowd support and from applying logic to the debate's structure. It's not some random idea that comes from a random algorithm - we've studied it, understood how it behaves and proved a series of desirable properties (like there is always one solution, and there's at most one solution to the debate). No point in debating if everyone can always be right because there are infinite solutions.
Also, quite a few of the suggestions in this thread propose dividing arguments into types, attacks into types, and essentially trying to classify everything into its own little niche that interacts differently with all the other little niches. While this could definitely be interesting, and might ultimately be what you want, I would warn again that simple systems work best and are easier to use. Unless you can absolutely justify the need for them, by having a specific debate instance where that feature is needed to make sense of it, they will just add confusion to the mix.
3) Details on how I'll spend the money. Who I'll hire, what they'll do, etc.This is very important. You need someone with qualifications that he has demonstrated. I would say that something as big as what you're envisioning can hardly be done by someone who is learning to code or has some basic notions. We're talking about something that, if done properly, COULD revolutionise the way people interact over the Internet. We're talking about a system that needs to support most platforms, from webforums, social networks, all the way down to apps on most mobile devices. Ideally, arguments could be shared between most of these systems.
You still deserve to make a living off of what you do if you can do it well, the question is only do you want to do this (or anything else) for a living.
I don't see any way to make a living off of just this project without undermining the purposes for which I do it.
I would say you'd want monetisation to be a side-effect of whatever this ends up, like Google making money off ads, but not really having things you can buy off them. I would really love to see something like this be free for use. If it's pay-to-use, you'll massively limit the potential userbase (almost the entire interwebs?).
Post by: Eagleon on September 27, 2013, 10:12:44 am
It doesn't even store text - that would be done by a BoxNode with a TextContent subclass, or if there were a message title or something, a PostContent. Then you'd ask the PostContent to give your UI nicely formatted chunks of data. Did I mention that I hate UI work? I'm not the one you want to work on UI, haha. At the same time, it allows the advanced UI presentation this project needs rather painlessly (I think).
I needed to name the class something, the nodes in the system contain it, therefore Content. Data sounds just as arbitrary to me, particularly if Connections are going to be based off of that type to take advantage of the splitting behavior it implements, but maybe it's better because they also need to be outside of Nodes. Semantics, blah.
I need to make a simple demo project that uses it for full effect, but if anyone is interested in this weird little stub library that does nothing, I can upload what I've completed tonight. I'm not exactly a newbie, I've worked on my own projects doing my own thing for six years off and on (mostly off), but I haven't actually shown my work to anyone but my roommate outside of an intro-Java class way back when I started. I'm a bit nervous about that fact, hehe.
Post by: Anvilfolk on September 27, 2013, 10:23:59 am
Post by: Angle on September 27, 2013, 11:34:18 am
Also, what do you all think a good name for this project would be? I was thinking something Greek about discussion and debate.
Post by: Eagleon on September 27, 2013, 12:10:22 pm
(it's what I've been using for lack of something better)
Rather than double posting - was distracted by family and a potential business opportunity, haven't lost interest, just time crunch :/
Post by: Angle on October 04, 2013, 03:00:08 pm
Post by: tompliss on October 04, 2013, 03:55:07 pm
I am currently looking for a "real" job, but got my diploma last month, and searching the web and writing cover letters all day long is boring ;)
Post by: Angle on October 05, 2013, 08:56:39 pm
Angle, if you need a mock-up or an testing prototype, tell me.
I am currently looking for a "real" job, but got my diploma last month, and searching the web and writing cover letters all day long is boring ;)
Yes, that would be excellent. In other news, however, I'm having second thoughts about managing the kickstarter. I'm afraid I have little to no experience with that sort of thing. I was going to do it myself mainly because I figured there was no one else who would, but now I wonder if maybe I should seek someone else out to do it.
Post by: tompliss on October 06, 2013, 04:03:08 am
I have a good deal of success consulting on Kickstarters as well as traditional financing.But yeah, you shouldn't try a kickstarter without reading about it, and seeking advice first.
Post by: BoboJack on October 06, 2013, 03:47:15 pm
(https://dl.dropboxusercontent.com/u/86860124/Selection_024.png)
Reference: https://dl.dropboxusercontent.com/u/86860124/enviromental.png (https://dl.dropboxusercontent.com/u/86860124/enviromental.png)
The argument of this picture is completely random and I just wrote down whatever I came up with, it doesn't reflect my opinion.
Oh yeah right, the links are directional, but I've swapped the arrows with the plus and minus signs some days ago and forgot about the directions. oops.
There is actually the possibility to make connections of the kind "only A and B together, imply C"/ A AND B => C
They are supposed to be reflected by the colors of the links, but I haven't really put time into that because I'm unsure about the design.
Right now everything is represented in a big mind map you can wander through, clicking on nodes moves the camera, you can also decide which parts of the mind map you want to currently see.
The back-end is written in Haskell and the front-end in coffee-script. PostgreSQL is the database.
I'd like to license everything under a free software license, BSD3 for example, but I'm not sure yet.
I'm struggling a lot with the design, which forced me to slow down on this project.
Everything should be easy to change and flexible, except the stupid SQL database and the queries. I plan on replacing these with combinator based sql some day. I've already tried mongoDB, it didn't fit this project, because the data is inherently relational.
I've not yet read the whole thread, but I will later.
I'd be happy to throw my stuff together with your ideas and I don't have any problem with giving away authority over my work.
I acknowledge that our ideas have very noticeable differences but I believe we can overcome these and work together.
Post by: Angle on October 06, 2013, 04:07:20 pm
Post by: BoboJack on October 06, 2013, 04:40:31 pm
I don't want to pay 6€/month. :(
But its a real webserver running on my computer.
You can look at it today, as long as I'm awake.
The website is very confusing since I have not spend any time on design, but you should be able to type "climate" into the search bar and hit enter. You should see the map then. (The search capabilities are similar to Wikipedia, but somehow the suggestions don't pop up correctly in firefox for me.)
You can also try this link: it'll link you to the "climate argument"
My ip changes every day.
edit: I'm going to bed now. I'll probably start the server,after waking up, 7. October 12 CEST and let it run for the day, so you people can have a look.
89.246.207.161:8000?id=39 (http://89.246.207.161:8000?id=39)
Post by: Angle on October 06, 2013, 06:53:49 pm
Post by: BoboJack on October 07, 2013, 04:19:38 am
You can not make the view more directional if all posts can refer to all other posts.
Its not a tree anymore if all the leaves are directly connected to the root.
You can take a look at some generic layouts here:
https://github.com/mbostock/d3/wiki/Layouts (https://github.com/mbostock/d3/wiki/Layouts)
Quote
and when someone clicks on a post, they get a view like thisI'll take a look at that. I'm not sure if/how one would switch back to the node overview.
There is a total lack of "how I imagine the user experience to be like" on my side.
Of course I've got a basic idea, but the actual design of which button does what and how is very unclear to me.
But the little pictures already help!
Also why exactly?
Services like iPage and "Quality Web Programming" are probably not what you search for.
iPage doesn't offer the possibility to run your own server software. Take a look at this https://www.heroku.com/ (https://www.heroku.com/)
Quality Web Programming, they offer prices based on how many "pages" you have and only do php programming.
You don't want a regular website. You want a webapp, were everything could be on one page, stuff being loaded dynamically as needed.
I'd like to point out that I pretty much want it to be possible to use the website without reading any long text.
There could be additional information for every node and for the connections (I played around with that, the code is already able to)
but large texts don't work on this kinda mindmap and the whole thing seems to become senseless if the texts are so long that you can not have an overview about an opinion without reading alot, you could just read wikipedia instead.
I also started with the idea of helping with discussions, but right now I feel like there isn't even a place to even see the connection between propositions. There are a lot of ways to discuss, but there is no collective database on how people believe opinions should be linked.
Kinda totally different maybe, idk.
With the stuff I've done so far, we could also just do 2 websites. or so. you know, one being focused on disucssing(but that kinda already exists: all the stuff anvilfolk points out in his paper + http://www.discourse.org/ (http://www.discourse.org/))
the other one being focused on knowledge representation.
edit:
oh and I'm also totally in for everything Anvilfolk said.
Post by: Eagleon on October 07, 2013, 11:27:34 am
All posts can already refer to all other posts.What you can do is decide on a portion of the graph that is a tree. You do this by marching outwards from a root node (whatever your camera is focused on) along edges until you detect recursion (build up references to nodes in a collection, compare each to the next deeper set of connections) The problem is more that this becomes computationally expensive when you're handling multiple users in a forum with hundreds, if not hundreds of thousands of chains and loops of arguments in memory at once. Meaning we might end up needing a more complicated but optimized client-server interaction so that the user can crunch some of this himself.
You can not make the view more directional if all posts can refer to all other posts.
Its not a tree anymore if all the leaves are directly connected to the root.
You can take a look at some generic layouts here:
On the plus side, that leads to the obvious benefits of letting the userbase improve the client interface themselves to their liking.
The more I think about this, the more I think it could resemble a true neural network in complexity. Particularly if it could be interlinked with unrelated forums, it would probably become quite complicated very quickly if people were willing to use it. It's actually a little creepy when I think "Well, what if this is just a natural extension of that underlying thought pattern into our social interactions?" I get a little zoomy, thinking about what it might build up to.
I'd like to point out that I pretty much want it to be possible to use the website without reading any long text.These are both reasons why I think we should leave that up to the people responding - let them dissect opening posts any way they like, people instinctually recognize arguments that they perceive in text (or video, audio, whatever), so if we let them split a conversation off in this way with a clear representation of what's being done and drawn from in the individual post (let the forum-like conversation go normally, and allow users to decide when to send out these special branched arguments) I think things will naturally begin to head off into more specific and pointed conversation as it's used.
There could be additional information for every node and for the connections (I played around with that, the code is already able to)
but large texts don't work on this kinda mindmap and the whole thing seems to become senseless if the texts are so long that you can not have an overview about an opinion without reading alot, you could just read wikipedia instead.
I also started with the idea of helping with discussions, but right now I feel like there isn't even a place to even see the connection between propositions. There are a lot of ways to discuss, but there is no collective database on how people believe opinions should be linked.
Kinda totally different maybe, idk.
This kind of points at another option for regulation of discussion - a currency system, obtained through replying to posts and climbing a 'ladder' of sorts in the network, where one only gains the ability to post a root topic in the forum by participating in discussion. Other user's votes would influence this in a bell curve of relevance, where votes at the bottom would mean quite a bit less for the user's currency rating than the ones in the middle, but the same as the ones on the top. This might eliminate troll-votes, make spam nearly irrelevant by allowing the full usergroup to slam their accounts with poor ratings, and reduce forum bias from both those that have gained the ability to start a discussion and the people that only post one thing and then leave.
And yes, I am thinking microtransactions :P The idea is probably incredibly dangerous for the bias it introduces, so I'm not genuinely thrilled by the idea, but it might be inevitable if for-profit industry adopts the model.
Post by: BoboJack on October 07, 2013, 02:01:21 pm
Quote
What you can do is decide on a portion of the graph that is a tree. You do this by marching outwards from a root node (whatever your camera is focused on) along edges until you detect recursion (build up references to nodes in a collection, compare each to the next deeper set of connections)I don't understand that at all.
There doesn't even have to exist a portion of the graph that is a tree. The only way to display the graph as a tree is by leaving out connections, but why would I do that?
Quote
currency systemBecause of these mechanics a lot of stack overflows legit questions get closed.
trolls/spam get downvoted and maybe flagged as spam, and bots can be kept out by captchas
Quote
root topicSounds complicated and restrictive.
Have you looked at the papers Anvilfolk supplied?
The idea described there is simpler and more general.
People could connect everything to everything, just like they do in real life.
And through their voting, the best assertions can be shown to the user.
No annoying long search for the user, no need to read a thread.
Quote
microtransactionsI wouldn't want to be a part of that.
Post by: Eagleon on October 07, 2013, 03:15:22 pm
To enable a UI that isn't a mind-map. That sort of visualization isn't exactly ideal to display larger blocks of text, as you've discovered, but it's inevitable that people are going to want to make more complicated initial posts as their ideas are not yet refined (like we're doing now - you're supplying some question about the clarity of my writing) so you do need one. You could probably do it by clustering the conversation together in box form and fitting argument lines around them, but then you have to deal with the nodes that are more distant possibly overlapping.QuoteWhat you can do is decide on a portion of the graph that is a tree. You do this by marching outwards from a root node (whatever your camera is focused on) along edges until you detect recursion (build up references to nodes in a collection, compare each to the next deeper set of connections)I don't understand that at all.
There doesn't even have to exist a portion of the graph that is a tree. The only way to display the graph as a tree is by leaving out connections, but why would I do that?
Human language presents multiple arguments and nested levels of meaning, because we're good at picking that sort of thing apart and focusing on one piece of perceived information, so it's strange to expect users to be perfect at refining each post they make into individual, atomic attacks, rather than letting others with different views and experiences do that for them.
No doubt, there are potential problems. It's just brainstorming. What I think might be a problem, specifically, with a random argument graph vs a restricted opinionbase like I described, is that users that have no basis or understanding to question an argument will do so whenever they want, vs. whenever it's warranted by their level of contribution. Of course, other users will downvote what they see as destructive to the conversation, but only if they're motivated to spend the time doing so - if you have a great, devoted memberbase this might be common, but the idea of fair moderation on a network of arguments implies that moderators have to read each link to form a less biased opinion, which would become increasingly difficult as it got more complex - which means artificial limitations on the complexity of an argument are sure (in my mind) to arise.Quotecurrency systemBecause of these mechanics a lot of stack overflows legit questions get closed.
trolls/spam get downvoted and maybe flagged as spam, and bots can be kept out by captchas
Coming from the neuroscience angle, our brains also function off of a currency system. Each cell requires an intake of energy that biases the whole system towards useful behavior in the overall network. If they aren't useful, they eventually become starved of neurotransmitter rewards, and either repurpose towards a more useful function or atrophy.
To be clear I'm not proposing any restrictions on the specific content of arguments, and off-topic ramblings would form their own hierarchial 'root' nodes. Additionally, as long as the information about the structure exists, it's fairly wasteful programmatically to discard it, as additional mathematics can be performed on hierarchial networks that are impossible on random ones. All it requires to keep this information around is a scaling variable for the entire network that increases as it grows new unique levels, and a single integer per node representing what level in the hierarchy they belong to and which levels in the hierarchy are their 'superiors'.Quoteroot topicSounds complicated and restrictive.
Have you looked at the papers Anvilfolk supplied?
The idea described there is simpler and more general.
People could connect everything to everything, just like they do in real life.
And through their voting, the best assertions can be shown to the user.
No annoying long search for the user, no need to read a thread.
Actually I'm not proposing any changes to the core of this project, just branches that I'd like to explore in the future - obviously I can't tell you what to develop, I'm just sharing some excitement over this from a network/information theory nerd's point of view. But if offtopic posts were valuable and extensively explored, they would probably be upvoted to a higher level anyway, as they'd receive more traffic from users exploring and posting in that particular part of the graph.
I'll go and read Anvil's papers, but I can't promise that I'd understand them. I'm informally and incompletely trained, and I acknowledge that openly. Mostly my point is that we can consider other aspects and divisions of network theory to apply to this system, and that's the appeal it has to me.
You don't have to be :P I'm not really interested either, or if I were, I'd want it to be in the position of a non-profit with ethical oversight determining what it got spent on. But devil's advocate: If this is developed into a codebase to be used in multiple forums, you'd need to patent the ideas here to prevent such a use, and defend it vigorously against imitators.QuotemicrotransactionsI wouldn't want to be a part of that.
Post by: Anvilfolk on October 07, 2013, 03:54:25 pm
That implementation is pretty kickass! Like I said through PM, it has a ton of potential! It's also really interesting that it has a backend with the data, and then a separate front-end. This should be fairly critical to the system we've been thinking of. You only implement the system once, and then implement lightweight interfaces for, say, Facebook, different forums, stand-alone applications, mobile applications, etc.
Eagleon, there's no worry in creating the Singularity using this, it eventually just becomes a system of equations :) Nodes don't really do any computation!
I guess the proposal here is, in general, give the ability to formally debate things to communities that are open to it. If the community isn't open to it, because it lets in too many trolls, or because people don't really care about it, it's not going to work. The community is its own moderator, through the means of voting.
Either way, it should be fairly straightforward theory-wise to make sense of having people with heavier weights in their votes. This allows for the system of currency, reputation, credibility, reliability, or positive/negative feedback on users. That could be an interesting feature for more closed communities, although it could put newer members of a community at a disadvantage... And it adds overhead.
Spoiler: Overhead (click to show/hide)
Moderators can simply be added application-side (as opposed to theory-side) by allowing them to add or remove nodes/attacks, rather than just vote.
I still feel that you don't need to lose the forum structure, or whatever underlying, typically hierarchical structure. It should be just as easy (or easier) to navigate as a forum like this one. That information simply isn't used in computation of debate outcomes... but it's there, as meta-data. Certain posts could be identified as being "first posts" or "debate starting posts" within a subforum of a forum. There's no problem with that. Once you get to within a thread, however, you're in graph country. You can still try to display things with a sense of time, by highlighting new posts since your last visit, or by having older posts in the centre, newer posts in the periphery, or try to maintain similarity with a thread by having newer posts nearer the bottom of the screen, if at all possible.
But again, that becomes meta-data, maybe secondary to the fact that you can now reason with other people in a more formal way, in a system that tells you who ends up being right or wrong. Heck, it goes deeper than that. Each node/argument is given a strength, so that you can have any number of participants proposing any number of arguments, and each of those arguments, not sides or people, has a degree of acceptability, so to speak, calculated from votes and from attacks (and votes on the attackers, and so on almost recursively).
The fact that we are writing huge amounts of text here is due to the fact that we only have the forum structure. We are forced to, in text, reply to everything that's been said before, and, in text, explain why it attacks or doesn't attack another point somebody else made. It makes everything more verbose. Hopefully people will get used to using smaller, to-the-point sentences. Instead of replying with one post, they'd reply with maybe half a dozen arguments.
Post by: Eagleon on October 07, 2013, 04:24:56 pm
Eagleon, there's no worry in creating the Singularity using this, it eventually just becomes a system of equations :) Nodes don't really do any computation!Ah, but people do ;) And if they're using this network properly, they'd be extending their knowledge into a system that is readable and processable, and useable by people. I've always said that neuroprosthetics were a better solution than trying to understand our own motivations and behavior without bias to create a new intelligence, and that the internet could be considered one, albeit quite a bit less efficient than we'd like. If you made this hierarchial (which the internet and the brain are both theorized to be) they might fill that role quite nicely.
I guess the proposal here is, in general, give the ability to formally debate things to communities that are open to it. If the community isn't open to it, because it lets in too many trolls, or because people don't really care about it, it's not going to work. The community is its own moderator, through the means of voting.Absolutely, but in the meantime you have apathy and lack of free time by users threatening the stability when trolling and spam happens, and no one is there to downvote it. I know I don't actively use 'thumbs up' or whatnot often - I'd rather post a response, because it lets me present new points of interest in full visibility to the topic at hand. Giving a motivation of some sort would improve that and get more people voting, vs spewing nonsense *coughEagleoncough*
(Edited for more explanation) -This is the purpose of the hierarchy in the implementation I suggested - stagnant debates are, by the structure of this forum, going to be the ones in the middle. They get trimmed the most, so the ones that survive are going to be very strong, and pretty much universal. As you near the top, you get into novelization land, where people want to figure out what they're talking about before delving into more detailed debate. You -don't- want those to disappear, because they represent the basis for the entire discussion. The upper-most ones are going to have lively new discussion branching outwards if anyone sees something new and important to discuss on that level, and will support and reenforce the most amount of nodes, so they should be the rarest and least likely to disappear. The lower levels will be getting the more refined, pointed arguments, which is a good place to demonstrate your core knowledge of a subject and build a reputation anyway. Altogether you get rewarded most for improving the core understanding of the next higher argument. But maybe not.Spoiler: Overhead (click to show/hide)
I still feel that you don't need to lose the forum structure, or whatever underlying, typically hierarchical structure. It should be just as easy (or easier) to navigate as a forum like this one. That information simply isn't used in computation of debate outcomes... but it's there, as meta-data. Certain posts could be identified as being "first posts" or "debate starting posts" within a subforum of a forum. There's no problem with that. Once you get to within a thread, however, you're in graph country. You can still try to display things with a sense of time, by highlighting new posts since your last visit, or by having older posts in the centre, newer posts in the periphery, or try to maintain similarity with a thread by having newer posts nearer the bottom of the screen, if at all possible.Verbosity serves a purpose in communication. We are not naturally inclined to limit it. I appreciate your work (from my very basic understanding of it) on an informational level, I love things like that which break down a system into clearly understandable parts that fit together nicely, but I think it's unwise to ignore the nature of the beast that's using a system like this and expect people to speak in crystal, mechanical clarity (following messy tangles of networked parts that appear completely unorganized), when they have flowers for brains ;)
But again, that becomes meta-data, maybe secondary to the fact that you can now reason with other people in a more formal way, in a system that tells you who ends up being right or wrong. Heck, it goes deeper than that. Each node/argument is given a strength, so that you can have any number of participants proposing any number of arguments, and each of those arguments, not sides or people, has a degree of acceptability, so to speak, calculated from votes and from attacks (and votes on the attackers, and so on almost recursively).
The fact that we are writing huge amounts of text here is due to the fact that we only have the forum structure. We are forced to, in text, reply to everything that's been said before, and, in text, explain why it attacks or doesn't attack another point somebody else made. It makes everything more verbose. Hopefully people will get used to using smaller, to-the-point sentences. Instead of replying with one post, they'd reply with maybe half a dozen arguments.
In any case, we're still at a level where our options for this are not limited by feature fitting towards a specific application, all I can hope to contribute is a nod towards the interrelated aspects of UI work and the gravity of discarding the raw information it could take advantage of down the line.
Post by: Antsan on October 07, 2013, 05:09:33 pm
I have a graph library available and I would like the chance to see which features are missing.
If you use quicklisp, it is available there as "fgraph".
If you just wanna take a look, it is on bitbucket.
https://bitbucket.org/thomas_bartscher/cl-fgraph (https://bitbucket.org/thomas_bartscher/cl-fgraph)
Do a breadth first traversal of the graph and collect the edges and nodes you traverse into a new graph. Viola, you have a portion of the graph that is a tree. Of course it does not include all connections between the found nodes, but for displaying purposes you don't need those available all the time. Display all the edges the node your mouse is over are incident to.QuoteWhat you can do is decide on a portion of the graph that is a tree. You do this by marching outwards from a root node (whatever your camera is focused on) along edges until you detect recursion (build up references to nodes in a collection, compare each to the next deeper set of connections)I don't understand that at all.
There doesn't even have to exist a portion of the graph that is a tree. The only way to display the graph as a tree is by leaving out connections, but why would I do that?
If you want to cut down on computation down, stop the traversal at a specified depth or maybe after you collected a certain number of nodes.
If you wanna get fancy, try to come up with some measures that describe the distance between two nodes and don't do a breadth first traversal but replace the queue in the bft with a priority queue.
Maybe provide different measures so people can have different views on the structure of the discussion.
Either way, it should be fairly straightforward theory-wise to make sense of having people with heavier weights in their votes.Thought to that: Use a network of trust to determine the weight of votes. Not that I know how to do that, and I guess it needs a lot more space and time.
Quote
And it adds overhead.
Quote
I still feel that you don't need to lose the forum structure, or whatever underlying, typically hierarchical structure. It should be just as easy (or easier) to navigate as a forum like this one. That information simply isn't used in computation of debate outcomes... but it's there, as meta-data. Certain posts could be identified as being "first posts" or "debate starting posts" within a subforum of a forum. There's no problem with that.Uhm, I am not quite sure what you are proposing, but it seems backwards in relation to what I thought one would do.
Any post has a time-stamp/version history (yeah, let's get fancy). If you want to see how a certain thread has evolved, choose a starting point, an end point, a heuristic to measure distance between nodes and then run a path-finding algorithm. The output of that algorithm is a thread that can be viewed by the user. It is implicit in the graph and the heuristic, but it needn't be anywhere explicitly.
Post by: Eagleon on October 07, 2013, 07:48:14 pm
Post by: Anvilfolk on October 07, 2013, 10:06:33 pm
On closer examination of centrality algos and what information they require to function I realized that what I'm suggesting would probably only be useful to me. That means I'm the only person that should be putting it in there, haha. This is another reason I'm reluctant to contribute - I tend to go off on half-baked tangents that may or may not pay off. Sorry guys.
Are you kidding?!
You almost just described how research is done! Heck, It happens all the time that people are solving a problem that's already been solved only because it's termed differently. I didn't know about this notion of graph centrality, and but a quick look at the Wikipedia page tells me it'd be a great way to display an argument graph. One of many, perhaps, but absolutely relevant! These could simply be visualisation options: one based on time, another on importance/centrality, perhaps another based on how credible arguments end up being, etc etc.
I will say that speaking over the forum can get totally confusing, and I've been having a hard time figuring out exactly what you mean. It seems I've been similarly confusing, judging by Antsan's post, so we should definitely try to be clearer about what we mean.
------
Regarding the dangers of verbosity in the system we've been talking about: there's always a set of assumptions whenever you start any project. Ours has definitely been that there's an unfulfilled need for more serious, structured debate somewhere on the Internet. People are getting tired of meaningless interactions. I don't care what you ate for breakfast, or what it looks like. I would, however, like to engage in a lively debate about beliefs or whatever technical issues I happen to know. If there's an easy system that allows me to make my points, and at the end tells me whether what I'm saying makes social sense (within the context in which I have spoken/written), then I'd feel that's pretty great. Yes, even if I end up being totally wrong.
Hell, we could be deciding the system we should be building with the system we should be building! Which I guess is part of why it's important that it exists!
------
Big explanation alert
Regarding structure, what I mean is this. As a first step, imagine the Bay12 forums, EXACTLY as they are now. Now, add in another BBCode tag, like:
Code: [Select]
[argument attacks="argument1, argument5, argument115"]I think the existence of a debating system at Bay12 would be really cool[/argument]So you could explicitly define, in your text, what you think is an important point for the discussion, or an argument. You also tell it what other arguments people have previously proposed it attacks. Every point made, like every post, gets its own ID, that people can reference - just like using quoting for posts!
Every time someone posts a new argument, the "outcome" of a debate is calculated. All points are highlighted, or marked, with a colour, anywhere in the spectrum from green to red. Green means fully accepted, i.e., acceptability 1, and red means totally refuted, i.e. acceptability 0. There's the entire spectrum of colour in between, or acceptability in [0, 1], for each argument to take. So at the end of this thread, we've have a fairly obvious clue of what Bay12 prefers in general, and we'd get a pretty clear picture of the kind of system that would work best for Bay12.
You keep the forum-like structure, with subforums and subsubforums and threads and everything. For the algorithm and system that reasons about arguments and tells you whether they end up being acceptable or not, that does not matter. That's why I mean it's a form of "meta-data", if you're thinking in terms of the argumentation system only.
Now get rid of the very strict forum situation. You could still have the forum and subforums, indicating the types of subjects debated. But each thread would be a graph instead of a list of posts. For instance, instead of making a post this long, I'd maybe cut it up, and make each paragraph of so an argument. Since each argument is supposed to be made to counter arguments other people have made, I'd probably declare a bunch of attacks as well. Instead of each argument within a thread being shown one after the other, like here, you'd be able to visualise a thread with a number of different "views", like some of the stuff Eagleon has been suggesting. If you want to see newer arguments, perhaps try to find a way to display the graph where the higher the argument is, the earlier it was posted. Perhaps you're interested in getting the gist of a discussion: then you'd use centrality, like Eagleon said. Or perhaps you're full of yourself and want to see your impact on the debate, and you'd get a view of your arguments in the middle, and the arguments they affect the most around, and so forth. There's many options!
One of the important bits of all this, is that underneath all of this stuff that's got to do with "how do I find a debate I want to engage in" (I usually only visit Other Games and Creative Projects, and avoid certain topics), or "how can I see/visually organise a debate", is a graph. And that graph does not need to be independent from thread to thread, which means one of our arguments could be attacked by something someone said in another thread. Heck, it could be attacked by a point made by someone in a NY Times online article, if they adopted the system!
I don't think we should visualise the entire Bay12 forum as a plain graph. It'd be IMPOSSIBLE to figure anything out. But forums, subforums and threads can become meta-data helping you get to where you want to be, and they can be visual cues for the graph visualisation. Like a nice box around all the arguments proposed in this particular thread. You might get some incoming and outgoing arrows, which is fine. They tell you there's another topic talking about similar issues! You might want to follow that arrow into another box and participate there too!
What you don't want to do is make any arrows disappear because then you can't really make sense of why the debate outcome turned out that way. The only reason you'd want to make arrows/attacks disappear is if they were super duper downvoted and they don't really have an impact, or if they are removed by a moderator.
Sorry for the long forum post, but I hope this gives you a better, more concrete idea of the kind of system we've been thinking about.
Post by: Eagleon on October 07, 2013, 11:17:07 pm
Regarding the dangers of verbosity in the system we've been talking about: there's always a set of assumptions whenever you start any project. Ours has definitely been that there's an unfulfilled need for more serious, structured debate somewhere on the Internet. People are getting tired of meaningless interactions. I don't care what you ate for breakfast, or what it looks like. I would, however, like to engage in a lively debate about beliefs or whatever technical issues I happen to know. If there's an easy system that allows me to make my points, and at the end tells me whether what I'm saying makes social sense (within the context in which I have spoken/written), then I'd feel that's pretty great. Yes, even if I end up being totally wrong.I tend to agree, especially since it's much simpler than pushing for what I want out of it. Still, I wish I had a formal background in programming this stuff so that I could tangent a little better, haha. If nothing else, I still think I'm right about letting users fragment a composite argument themselves, rather than requiring them to make precise ones to begin with and punishing them if they don't. I'll take a crack at the basic model (no fragmentation, simple directed graph, a separate visualization from BoboJack's) again tonight/tomorrow, I've brushed up on an unrelated project and my programming juices are flowing nicely.
Hell, we could be deciding the system we should be building with the system we should be building! Which I guess is part of why it's important that it exists!
What you don't want to do is make any arrows disappear because then you can't really make sense of why the debate outcome turned out that way. The only reason you'd want to make arrows/attacks disappear is if they were super duper downvoted and they don't really have an impact, or if they are removed by a moderator.Question - if you allow for removal of arguments, what do you do about arguments that have been orphaned into their own little isolated graph? If you make attacks disappear this way through user action, it could happen quite frequently, no matter how high you set the threshold to destroy them. Would those nodes just disappear as well? Any time I hear or talk about disappearing nodes/edges through user moderation I kind of cringe for this reason - if there's a faulty argument/attack I think it should be left alone with the arguments against it left intact, or at least simply hidden, so that people don't keep making it and making more orphan networks.
Post by: Antsan on October 08, 2013, 03:44:48 am
I don't think we should visualise the entire Bay12 forum as a plain graph. It'd be IMPOSSIBLE to figure anything out. But forums, subforums and threads can become meta-data helping you get to where you want to be, and they can be visual cues for the graph visualisation. Like a nice box around all the arguments proposed in this particular thread. You might get some incoming and outgoing arrows, which is fine. They tell you there's another topic talking about similar issues! You might want to follow that arrow into another box and participate there too!That makes sense!
Quote
What you don't want to do is make any arrows disappear because then you can't really make sense of why the debate outcome turned out that way. The only reason you'd want to make arrows/attacks disappear is if they were super duper downvoted and they don't really have an impact, or if they are removed by a moderator.I think that views that do not display all edges in a given induced graph could be very useful - maybe you only wanna see how arguments relate to a certain post or you can't figure out what's going on and you need to get rid of some edges, so you want to find a planar subgraph to display or cut down the number of edges per node to some maximum.
If the displayed graph gets recalculated while browsing the missing edges aren't that important - the user will find those while exploring. Missing edges can be indicated in some way.
Question - if you allow for removal of arguments, what do you do about arguments that have been orphaned into their own little isolated graph? If you make attacks disappear this way through user action, it could happen quite frequently, no matter how high you set the threshold to destroy them. Would those nodes just disappear as well? Any time I hear or talk about disappearing nodes/edges through user moderation I kind of cringe for this reason - if there's a faulty argument/attack I think it should be left alone with the arguments against it left intact, or at least simply hidden, so that people don't keep making it and making more orphan networks.I forgot the name, but I think there is a way to detect nodes that are necessary in a component of a graph to keep it connected. If such a node is removed, calculate the new components and make them new topics in the given forum, maybe name them "[topic] - Split [n]", alert the administration and give someone the possibility of choosing a better name for the topics.
I'll start programming an engine for this today.
Post by: Angle on October 08, 2013, 10:33:02 am
Post by: Eagleon on October 08, 2013, 01:11:39 pm
In reality, the change that I depicted ("Your mom misread god" disappearing) would probably result in a reshuffling of all seven nodes based on a few predetermined selection criteria such as high or low rated attacks - you could allow users to pin nodes to prevent this from happening, or prevent it in part, so that you could remain focused on All Hail Grod the Uneasy as an attack, keep Non-Falsifiable in play, and try to see how they relate to the central argument better.
To get this view (as far as I've worked out)
1. Pick an argument.
2. Pick something it's attacking, attach it to a single-axis scaling connection at the top of 1.
3. Collect what attacks 1, decide how to prioritize their display. Attach the two or three best to the other sides with another connection of the same type. Scale all nodes to the screen. Find connections between all nodes and draw them.
4. Collect what attacks 3. Prioritize their display the same or differently (could provide contrast, select for arguments that attack both 1 and 3, etc). Attach them to free sides. Scale, find connections, draw.
And repeat until you're satisfied with the number of nodes on screen. Then you can do the fancy work of figuring out connection paths that don't intersect your OGL objects, preventing overlap of argument blocks, working in floating force-atlas style network sections on the perimeter, allowing hiding and revealing of argument bodies and/or blocks etc. I'm willing to bet you could modify a standard visualization like BoboJack's to do this if the visualization allowed you to single out nodes you don't want animated, and if it doesn't, it could be rewritten to. You could get really fancy and send the nodes back off into the network like a bomb exploding, drawing new ones that have been selected in.
Lots of fiddly! I don't like UI work =P None of that would work for me the first, second, third, n^fanciness times, because there's always some boundry or glitch you aren't going to expect that ends up making it look like garbage all the way through, haha.
Post by: Antsan on October 08, 2013, 01:23:03 pm
So It looks like we have two or three different ptrogrammers- Do we want to coordinate our efforts through github or something?I don't know. BoboJack uses Haskell, I use Common Lisp and whoever else is in on this thing is probably using neither and I guess no one is really willing to switch languages.
If there is anything to coordinate via github besides collaborative programming...
Post by: Eagleon on October 08, 2013, 01:44:20 pm
Post by: Anvilfolk on October 08, 2013, 01:47:52 pm
However, it shouldn't be too hard to go on to www.debate.org, find one of the more refined debates, and use that as a source for choosing what is an argument and what isn't. I've seen a couple of debates with a level of formality that makes it easy to identify what is an argument and what isn't, and what attacks what.
Post by: Eagleon on October 08, 2013, 03:26:10 pm
If nothing else, Gephi can export to CSV, with a fair number of generic datatypes built-in (Data Laboratory is where you want to look if you'd like to specify more columns). Here's the files for this (https://dl.dropboxusercontent.com/u/46645289/Herpdebate.png) - Nodes (https://dl.dropboxusercontent.com/u/46645289/Nodes.csv) (edit: fixed (https://dl.dropboxusercontent.com/u/46645289/Nodes2.csv) small defect in the state military section), Edges (https://dl.dropboxusercontent.com/u/46645289/Edges.csv) - if anyone wants to work that functionality into their program using a non-random data set.
Post by: Anvilfolk on October 08, 2013, 03:29:38 pm
http://wiki.lesswrong.com/wiki/Debate_tools
and http://debategraph.org/ in particular.
Unfortunately, none of these provide any outcome of a debate, let alone a formally justified one, which is what we'd like to have.
Post by: Urist McScoopbeard on October 08, 2013, 05:06:26 pm
that's kinda what i've been envisioning in terms of organization.
Post by: tompliss on October 10, 2013, 05:06:51 am
So It looks like we have two or three different programmers- Do we want to coordinate our efforts through github or something?I don't think that's what is important now... at least for angle's project.
Here is how I see it working :
- First, you (as in every concept designer involved here) need to study the existing tools, as they will give you some more ideas.
- Then, you (Angle, as it is your project) should write some senarii, some user stories, without thinking about an interface. Those should be understood by everybody invloved, as they will be the core of the development.
- Developers take the lead, there, building mock-ups and showing them for tests and approval.
- After a few iterations on the mock-ups, something good should be ready to launch a kickstarter, or "simply" search for fundings.
Post by: Angle on October 10, 2013, 09:58:11 pm
Post by: tompliss on October 11, 2013, 01:39:32 am
They're little scenarii that describe what is done with the solution, with few details on the "how".
They're good at dividing the solutions in a list of features, and seeing the links between them. And of course, they're very good at creating Task-Trees, that are very good at abstracting UI.
Post by: Eagleon on October 11, 2013, 12:38:53 pm
I suggest that as you work out more of what you want from the project, we shift onto a more collaborative medium like Google Docs to share ideas and feedback a little faster. I'm onboard for whatever. I need a little more direction to start work though (I can't run or serve a database, electric bill is an issue), or else I'm just going to end up being redundant and in the way, code-wise.
Talking about this project with my own 'crew' (mostly family, granted, but still capable people) has me excited to continue. Especially since, for the time being, my own plans are on hold while I bunker down until the government starts back up. My mother (who has participated in budget matters) suggested that this might help Congress to present their debate more intelligently. My brother wants to try using something like it to figure out where we need to bone up and what we need to focus on for our university project. It could be applied to detect fault-points in diagnostic reasoning without using a large special-built informatics program, in medicine or just technician work - I can see repair people (and teams of repair people) using it to keep their reasoning and workflow more tightly organized, reducing redundant work without needing to hire a process engineer to point it out. That would be fantastic for small businesses that don't have that luxury or expertise.
Not to say you have to worry about fitting a UI or design to those things, just saying that I think it shows a lot of promise.
Post by: Anvilfolk on October 11, 2013, 01:06:12 pm
Plus, I'd love to see a lot of the applications of these systems you are talking about.
Here's the current situation. So far, none of the proposals we've seen here so far warrant significant involvement from my research group. Please don't take it the wrong way, but like I said before, we're talking about some serious involvement on our part, which could take up a non-trivial amount of what we do for work daily, at the expense of our other research activities.
Most of the proposals I've seen here are either going in a direction that is of little interest to us academically (e.g. they're not interested in calculating a formal outcome to a debate, simply in representing it); or whoever's developing it doesn't really have a lot of time; or its ideas are still a little too abstract.
It's very hard for us to advise an open-source community, since all of us are pretty busy and coordinating with such a fluid entity requires a whole lot more involvement than working with an independent and efficient implementation team/person.
However, if you guys are interested in following up with the idea that we want a formal outcome to a debate that we need to compute and present to the user (i.e. based on the Social Abstract Argumentation that we presented in our papers, open to modification), I'd be happy to help around on my free time. That means I can advise on some technicalities of the project, related to efficiency and good software practices, as well as how the infrastructure is supported, and might even get my hands dirty at some point :)
If you are interested, you should state how you would like to contribute to the project and your qualifications for it. This should help us organise people around better. Also, before starting any implementation, we must figure out the basic architecture of the system. I have a couple of ideas for that, but would like to hear back from you all first.
Since I'm already volunteering with this post, I'll start. Let's meet each other :) Here's what I'd like to help doing:
- - Overall my goal is to have an implementation of the SAA theory.
- Alter the theory as necessary to make the implementation usable and provide what the users want.
- I can be the liaison with my research group as the project becomes more mature, and perhaps involve it officially if this matures enough.
- Make sure that the overall architecture is extensible and as far-reaching as possible. We want it to be technically possible for this to be pervasive.
- Advise overall tech stuff.
- Set up the tech infrastructure for this, like manage GitHub repositories, etc.
My qualifications: I'm a 4th year CS PhD student, and I've got a BSc and an MSc also in CS. I don't know all of these newfangled web technologies, so I probably won't do any of that, but I have a fair knowledge of software architecture, DB and programming languages.
Post by: Angle on October 11, 2013, 01:47:31 pm
Post by: Eagleon on October 11, 2013, 02:20:01 pm
Anvil: Sounds good to me. I'm interested in getting an SAA model first as well, since it's based on published work. The idea of a formal outcome is definitely complimentary to having something that can also function to highlight contradictions and areas that need more backing data or statistical examination for bias. I'm only just poking my head up into collaboration with this, so forgive me if I get distracted by that.
Other areas I'd like to contribute:
- Outline and plan the datastructure for each component, once we figure out what those components are. I love datastructures, finding and eliminating redundancies, etc.
- Make sure there is an interface to the architecture that is as generic as possible. Down the line including things like references to video, HTML, PDF, images, just basically any element we'd like to include in a quickly user-formattable way would permit this to be applied to much more than 'just' debate
- Porting what we have and what I can to Java for general use
- Documentation, organizational and otherwise, with some help for accuracy in terminology and thorough head-kicking for verbosity ;) At the very least I'm enthusiastic with my java-docs, if not entirely lucid
- Debugging, analysis for problem areas, as able
- Research on related projects and hunting down code-snippets that are faster, better, stronger than our own
Post by: Anvilfolk on October 11, 2013, 02:50:49 pm
Eagleon, awesome! From my conversation with BoboJack, I think he's also pretty interested in these kinds of things, and his ideas were already super close to SAA, so hopefully he might contribute a bit.
It's important to realise that this is totally not going to be a full-time job or anything. When you can and feel like it, you contribute. That's all :)
We could use GitHub's wiki system to store decisions we make. In the meantime, we can keep our discussions over here.
Here's my thoughts about architecture. There should be three components:
- Database: it simply stores all the arguments and attacks in an easily accessible format. There can be multiple of these, all over the internet. One for each website, and each website has an ID to identify its own arguments; like A1:D1 means argument 1 from database 1.
- SAA component: it is authoritative over one or more databases. It accesses the databases, crunches the data to provide outcomes, and stores argument acceptabilities in the database. It can do this either periodically or whenever there is an update.
- Interface: there can be as many of these as you want. They can be programmed in any language, and available on any platform. The more the merrier. These gather data from the databases, and just display it for users. They also provide function for adding/removing arguments, etc.
So basically, you've got your arguments all over the place, and some applications crunch the numbers for you, so that you can always display debate outcomes in the interfaces.
Both the interfaces and the SAA components are going to have to query several databases. If I query a database D1, and an argument D1:A1 is attacked by an argument A2 from database D2, then I'm going to have to query D2 about the strength of A2. We can use DNS as an inspiration for how to do these things. We can provide this argument-fetching service as a library with bindings in several languages.
Also, databases are going to have to carry enough metadata for each argument so that you can configure the whole thing like a forum/subforum/thread thing. This needs some thought.
Also, does anyone have a name for the project? That way we could create a GitHub repository (if everyone is ok with that choice, which is free!) and start editing its wiki with information we agree on. I've never done wiki editing before, but it can't be too hard, right? :)
Post by: Angle on October 11, 2013, 02:51:51 pm
Here's a scenarii, albeit with more pictures than words:
Spoiler (click to show/hide)
And here (https://docs.google.com/document/d/1CjERoM9jefThEU35WAVT9IALjrYhhYgcAHbBvfgjWbs/edit?usp=sharing)'s the link to the google doc..
In terms of forum name, I was thinking of something based on ancient democracy, freedom of speech, discussion, discourse, that sort of thing. I had been working under Project Athens, but decided I didn't like it.
Post by: Eagleon on October 12, 2013, 11:22:46 am
We could use some help figuring out what database/server software to work with, and what needs to be done to get a basic interface with one on a remote client up and running. I have very little experience with them - I understand protocol buffers as our best bet for getting readable information to our interface(s) without sending huge XML files or similar back and forth constantly, but I could be completely off-target for all I know.
Lacking that, I could start basing my work on Gephi's server plugin and try to reduce what it provides to an arbitrarily chosen database format, but I don't know how good it'd be, or if there's a better solution already around. Does anyone know any tutorials I could poke through to learn where my efforts are best placed? I'd be more willing to work on a local database if I knew I wasn't wasting my time on things that have already been worked out.
Post by: BoboJack on October 13, 2013, 03:36:12 am
Quote
We could use some help figuring out what database/server software to work with,Database:
The big 3 of Databases I know: MySQL, PostgreSQL and MongoDB.
I've used PostgreSQL and it was ok. I've tried MongoDB first, wasn't so nice.
If someone can come up with a layout for mongoDB upfront, it would maybe be worth it.
Server:
Options I know of:
Apache HTTP server (http://httpd.apache.org/) and write stuff in php (which sounds very ugly to me)
Node.js (http://nodejs.org/) very cool for a community project. Server and web fronted would be in the same, very common, language. +1 from me.
Ruby on Rails (http://rubyonrails.org/) also very popular now, but slow to my knowledge.
I've used a Haskell library, http://snapframework.com/ (http://snapframework.com/), just for the record
Here's a dropbox folder of my stuff (https://www.dropbox.com/sh/jey9bsh88o5m9bl/WBzrHS-Mjr/website)
This is the coffeescript(javascript) front-end I've written. (https://www.dropbox.com/sh/jey9bsh88o5m9bl/WKDBvY7w5t/website/snaplets/coffee/coffee/d3test.coffee)
Quote
and what needs to be done to get a basic interface with one on a remote client up and running.I wholeheartedly recommend D3: http://d3js.org/ (http://d3js.org/)
Quote
I understand protocol buffers as our best bet for getting readable information to our interface(s) without sending huge XML files or similar back and forth constantlyJust send JSON data. JSON is widely supported (native support in javascript, libraries everywhere) and has an efficient binary counterpart called BSON.
MongoDB also uses BSON!
Post by: Antsan on October 13, 2013, 04:16:15 am
Quote
Apache HTTP server and write stuff in php (which sounds very ugly to me)Actually Apache can be configured to accept any language, if I remember that right. At the very least Hunchentoot (Common Lisp) can work with Apache.
Post by: BoboJack on October 13, 2013, 04:50:41 am
Actually Apache can be configured to accept any language, if I remember that right. At the very least Hunchentoot (Common Lisp) can work with Apache.Right. I didn't think about all the cgi stuff.
CGI doesn't sound like a "batteries included" solution though.
Post by: Anvilfolk on October 13, 2013, 09:15:21 am
I'm going through the document and adding comments right now. This is so cool I am itching to get started :D
We need a name to set up the GitHub repository. From there we can start putting up code, put information on a nice wiki, some pretty images of what we envision, and I have a webserver that we can use for testing purposes.
I'm tempted to just start with something simple like SocialDebate or TrueDebate - we can easily change the name at this stage.
Quick note: I think you guys are on the same page as to what you mean by server side stuff, but I think I haven't quite gotten it. The database stores the debate itself. The interfaces grabs information from the DB and shows it to the user. What are you guys expecting the server side stuff to do? I've always imagined that all it's doing is computing and updating debate outcomes.
Also, Angle, could you update the OP with the google doc and other information? We should sig this and make it obvious that it's open-source and open to participation :)
Post by: BoboJack on October 13, 2013, 11:06:18 am
Quote
What are you guys expecting the server side stuff to do?The server could also host the web front-end and manage users and their authentication.
It would also make changes to the DB through an API accessible from the front-end.
+calculations
Quote
The interfaces grabs information from the DB and shows it to the user.How would the interface authenticate with the DB?
If the interface can directly change the DB (voting for example) you could use the login/password from the interface to make any changes you want.
Post by: Antsan on October 13, 2013, 11:31:17 am
Post by: Eagleon on October 13, 2013, 11:58:21 am
BoboJack - thank you! I'll look at those when I get back from my aunt's later. I knew JSON was a thing, but I didn't know it was that kind of thing, heh. So much to learn.
Post by: BoboJack on October 13, 2013, 12:22:03 pm
Also server side computation should be OK, if you don't have to recalculate the whole network on every change.
If the client would do all the computation, the client would have to recalculate everything again on every access, at least for a web front-end.
I also wouldn't want a website to suck all the energy of my smartphone.
Also I encourage you to make the communication between server and client work through a RESTful API (http://en.wikipedia.org/wiki/Representational_state_transfer) (basically the client communicates with the server only through a bunch of HTTP requests)
Post by: Angle on October 13, 2013, 01:03:37 pm
I'm tempted to just start with something simple like SocialDebate or TrueDebate - we can easily change the name at this stage.
Remember that this is intended for more than just debates- I also want this to be useful for things like what we're doing right now - planning projects, discussing ideas, etc. Really, I'd like to call it dscourse, but someone already made off with that name.
Also, Angle, could you update the OP with the google doc and other information? We should sig this and make it obvious that it's open-source and open to participation :)
Done!
Post by: Anvilfolk on October 13, 2013, 02:40:31 pm
The server could also host the web front-end and manage users and their authentication.
It would also make changes to the DB through an API accessible from the front-end.
+calculationsQuoteThe interfaces grabs information from the DB and shows it to the user.How would the interface authenticate with the DB?
If the interface can directly change the DB (voting for example) you could use the login/password from the interface to make any changes you want.
I feel like all of this functionality - authentication, getting information and modification - should be in the libraries we provide to users, including authentication. I don't know too much about this, but it'd be kind of like a PHP session. When you authenticate, you get a "ticket" which is yours alone, and valid while you keep using it. It times out after a little bit, or you store that in a cookie-like situation. Encrypted communications should ensure nobody else gets that ticket.
Having a web frontend makes everybody HAVE to use that particular webserver. I don't see why this can't be handled by whatever interface you decide to use :) If you're using a web interface, you use the web login. If you're using it on your smartphone, you login from your smartphone's application.
I have no experience at all with server-client relations and all of that stuff, but I think computation should not be done by the server but the client. When more and more people start accessing the server you would probably get serious performance problems pretty soon.
You can do client-side computation for things like, "how do I display the debate", but for calculating debate outcomes you need something dedicated. Like BoboJack said, different clients have different architectures, and they give different results. Interfaces are also proned to be hacked (especially if they're open-source...), so that you can't really trust them. At the very least, you'd need server-side verification of the the results, which is actually faster than calculating everything, so that is an option.
It all amounts to the underlying infrastructure. A debate for 3 people can run everything on a mobile phone. A debate on Bay12 needs more infrastructure. It's up to whoever is hosting to handle that.
The algorithm we have right now does not handle incremental changes. There's some minor optimisations you can do, but we still need to recompute almost everything whenever there's an update (including a single vote). This is a limitation that it would be amazing to get over, so I can put up a description of the algorithm at some point for you to review. The idea's actually pretty simple. You can modify the algorithm, but you need to prove that it'll still do what we want.
I think at the moment we can handle a few thousand arguments in a couple of seconds. Like I've told those I've talked to, we've had someone working on optimising these things for a few months, and this is what he came up with, so I believe at the very least it's a good starting point. Eventually people should be able to use higher-end machines with multiple cores (the algorithm is highly parallelizable), which should make things better.
Remember that this is intended for more than just debates- I also want this to be useful for things like what we're doing right now - planning projects, discussing ideas, etc. Really, I'd like to call it dscourse, but someone already made off with that name.
I was using debate in the wider sense - any discussion where points are made and argued over. For instance, I'd say we are engaging in a nice debate about how to organise an application. I suppose people might have some fairly strict notion of what a debate is. We should definitely include the idea that it's "social", because of the voting and the fact that outcomes depend on the context in which discussions are carried over.
Funny how coming up with a name is usually one of the hardest parts :D
Gimmie a few minutes and I'll put up a little infrastructure diagram of how I think most interactions can go.
Post by: Anvilfolk on October 13, 2013, 03:35:37 pm
Spoiler: Overall architecture (click to show/hide)
Let me explain what's happening there.
Lots of users are accessing Server #1. Each of them can be using a different application, or interface, as we have been calling them. To obtain information from or modify a debate (I'm used to using this term, sorry!), the applications/interfaces use a very small client-side library with minimal functionality. You can log-in, ask for debates, and submit modifications. That's should be almost everything.
Once the requests arrive at a server, there is a server-side interface. This parses the requests, and talks to the database. It makes sure logins are correct, that votes are accurate, etc.
In this case, Database #1 contains a small debate that is being affected by a debate stored in Database #2. Upon identifying this, the server-side interface from Server #1 will figure out where the other debate is stored (Server #2), and talk to the interface responsible for it to get the necessary data. ***
There is only one SAA Computation node, who is responsible for computing debate outcomes. He is authoritative over both Server #1 as well as Server #2. That means he will actually talk to both databases and recalculate debates within those databases.
*** An alternative version to this is that the client finds out that there are relevant arguments stored in other databases, and the client goes and grabs them. This removes communication overhead from the server-side interface, of which there can be a LOT if there are many clients logged on. Instead, each client is doing a little bit more work, which is fine. DNS has similar mechanisms. We can figure out which one we want to implement, and possibly both.
The difficult part will be figuring out how SAA authorities work together, and how those edges between databases under different SAA authorities are dealt with. At this point we have our hands full though, and I believe that is something we can deal with in the future.
I hope the diagram makes these ideas a lot clearer to everyone! Let me know if you have any questions! :)
Also, the diagram was made in yEd (http://www.yworks.com/en/products_yed_about.html), and I can provide the base file if you want to play around with it.
Post by: Angle on October 13, 2013, 10:54:43 pm
Post by: Anvilfolk on October 13, 2013, 11:39:41 pm
I'd love some feedback on the architecture. I've put that up along with a description in the online google doc. We are creating a GitHub organisation for this, and from there we can create repositories for all major components!
Post by: BoboJack on October 14, 2013, 03:53:00 am
I'm not so sure about the database set up.
If database #1 and #2 are referencing the same big web, why are they separated?
Every query done by the server side interface now has to query two databases.
Communication/keeping things consistent between the databases seems really hard and would have to be hand coded.
I would rather follow these instructions (http://www.postgresql.org/docs/9.3/static/high-availability.html).
We would have one logical server, to which all the server side interfaces reference to
and the logical server would be made out of many servers which would communicate in one of the ways described in the link.
Post by: tompliss on October 14, 2013, 07:02:26 am
About the computation thing, remember one thing, guys : you don't want the user to download every argument at once in his client for it to filter out what he want to read.
Pretty much the same idea about having pages in a classic forum thread.
Post by: Anvilfolk on October 14, 2013, 08:48:47 am
That's not exactly what I was thinking of here though. What we should make possible, eventually, is for an argument here on Bay12 to reference or be referenced by something on some other website unrelated to Bay12. More specifically, imagine that Agora has been deployed at bay12forums.com and at nytimes.com - there are lots of topics in common in both. I would love to be able to use a nytimes.com argument to attack a bay12forums.com argument, for instance. If you can do this, all of a sudden Agora is infinitely more powerful and interesting. This goes hand in hand with the development of the Semantic Web (http://en.wikipedia.org/wiki/Semantic_Web), which is huge in academia right now (including the guy who made the WWW), and is often said to be the future of the internet.
However this should definitely not be a major concern for initial versions, but we should keep it in mind as we develop the architecture.
What you suggest with PostgreSQL is wicked awesome though. I imagine we can definitely have a package that installs Agora on a PostgreSQL database rather than a mySQL one. As long as there's some install script we can provide with it, and a server-side interface that knows how to communicate with it, it sounds like postgreSQL could be more efficient! That's the beauty of server-side interfaces. They hide whatever is behind! :D
tompliss: yeah, I figure we should definitely be able to keep some forum-like organisation as well. I am guessing that we might use JSON/BSON like BoboJack suggested to store this type of structure. Because JSON is so flexible, it can also be used to store different structures. But again, this needs a whole lot more thought. We need to adapt to forum-like structures, article-like structures (news sites, blogs), etc.
I think the client-side library and the server-side interface should have some sane default values as to how many arguments you can download at once.
Quick notes:
- Angle created a GitHub organisation for the project. Post your GitHub id on the google doc "participants" section and you'll be added.
- If I am being too verbose, let me know and I'll cut down on it.
- Once Eagleon responds, and if he agrees with the architecture, let's go ahead and create the repositories for each component and start a'hackin' :)
Post by: BoboJack on October 14, 2013, 09:56:01 am
Quote
That's not exactly what I was thinking of here though. What we should make possible, eventually, is for an argument here on Bay12 to reference or be referenced by something on some other website unrelated to Bay12.So you want something like the pods from Diaspora? Is that even possible?
How would searches work? Every search by a user makes a request to all pods?
Quote
let's go ahead and create the repositories for each component and start a'hackin'We still need to decide on one language/framework to use.
Post by: Anvilfolk on October 14, 2013, 10:19:08 am
However, Bay12's database might refer to NYTimes arguments. If argument IDs contain location information, this becomes much easier, and kind of like how DNS works. Hence my suggestion of a node ID being D1:A1, for argument ID 1 from database ID 1. D1 could be "nytimes.com", and A1 could be a random number. Thus, when you see an incoming attack from D1:A1, you know you need to contact the nytimes.com authorities to get information about that debate.
Like I said, contacting "secondary" servers when browsing debates could be done either server-side or client-side. I'd go for client-side to avoid overworking servers.
As far as the near-future roadmap, here's my suggestion:
- Develop an independent system, i.e. don't try to integrate it into SMF, phpBB, Blogspot, etc, yet
- Go for the path of least resistance tech wise: mySQL is the most easily accessible database, and for languages we should use our available skillsets
- Before implementing anything, we need the database schema, including how we store the forum-like structure
- Once the basic database structure is up, we need the communication protocol
- With the communication protocol, implement ONE server-side interface and a couple of client-side libraries in different languages
- Implement one or two client interfaces
The SAA stuff I can probably code super quickly, so don't worry about that.
Right now we need to figure out database and comms protocol.
Post by: Eagleon on October 14, 2013, 10:21:38 am
- Once Eagleon responds, and if he agrees with the architecture, let's go ahead and create the repositories for each component and start a'hackin' :)Erp - I didn't realize it hinged on me. Absolutely, I have no complaints as to the organization, but for one thing - in the doc you mention "Server-side interface is the “site layer”" but everything below Database (starting at Site Layer) was me determining what the database needed to send out to the clients. Rather than send out site-level information within each node, it made more sense to encapsulate all nodes and edges from a site with that information. It's a quibble (I'm not disagreeing to a server interface), but it'll help if we end up having multiple different types of 'sites' (could end up accessing offline databases in specific use scenarios) and reduces redundancy in the other objects.
I love "Agora" :DIt should be possible to send out portions of the graph in little sub-networks, by picking a node, traversing a certain distance, and sending one more set of edges past that number of nodes in the graph. The server could also operate on these sub-networks as they're built and packaged, to do fancy math things, but anyway- So if the client can handle having dangling edges to unknown groups of nodes, it can either ignore them or use them to request more. The problem is this adds more for the server to do - processing at the expense of bandwidth. My mind is going to fractal database formats O.o Not helping, brain.
About the computation thing, remember one thing, guys : you don't want the user to download every argument at once in his client for it to filter out what he want to read.
Pretty much the same idea about having pages in a classic forum thread.
You also run into synchronization problems if nodes or edges are destroyed, which should only be confusing to the client, not the server. The server would just have to be telling the client whether the changes were successful, or blocked by an inconsistency.
PostgreSQL talks about load-balancing - if the server load is heavy, maybe we could even do this automatically based on demand, only giving the client as much of the local graph as the server can keep up with and still provide other active users OK service.
Post by: Anvilfolk on October 14, 2013, 11:19:16 am
By site-level information you mean a forum-like structure, or a newsite like structure, or blog-like structure, right? If so, I think the current proposal by BoboJack where that information is stored in a flexible JSON format column in the argument/node table in the database should work for what you mean.
When you access the server-side interface, you can request the "meta-information", which clients should know how to parse easily.
I feel like a generic hierarchical organisation could be enough for most sites, if not all, sites. Let's call these "forums", to keep with intuition. Every Agora-powered website has a main "forum". Any forum can have subforums, which are also forums. Any forum can also have "threads". Threads cannot have sub-threads or sub-forums. Instead, all arguments are proposed within a thread.
This mimics a forum like Bay12 perfectly. It should also be able to capture a blogging website, where the Blog is a forum, and each article is directly a thread (which is used by the poster, and commenters alike).
We can also say that a thread can belong to multiple forums, which allows us to implement a blog post as belonging to several "categories".
Post by: Eagleon on October 14, 2013, 11:27:34 am
Also - the google doc is getting to be quite slow to load, maybe we should split it off into documentation for the different sections (architecture proposals, database, interface, project overview and outline/references to each). That way people without github can quickly see roughly what's going on - or does their wiki system allow anonymous browsing?
Post by: Anvilfolk on October 14, 2013, 11:29:20 am
I'll try to create a couple of empty pages to set up the structure, and we can start migrating there. Since it's open-source, I can't see why it wouldn't be public!
Post by: Anvilfolk on October 14, 2013, 02:36:33 pm
The Google Doc is where we post and discuss proposals more persistently. We can use the really cool comment system to identify and resolve issues as they come along.
Once there is a quorum on a proposal, it gets transferred over to the wiki, and becomes an official decision.
If you guys agree, we can start transferring and removing things from the Google Doc into the wiki. Which, by the way, is becoming really cool :)
Post by: Angle on October 14, 2013, 02:37:40 pm
Post by: Anvilfolk on October 15, 2013, 11:11:52 am
In the meantime I am wondering whether it wouldn't be nice to get more people involved. I've mentioned to my research group that this is going ahead as an open-source project, and they seemed interested. I'm waiting for the right time to let them know officially this is happening.
I'm also thinking of giving a small talk/seminar about the entire project here at uni, from SAA (the academic perspective) to Agora (the application perspective), and inviting participation. I feel there might be some really smart people getting interested about the entire idea and hopping on board. You could probably skype in if you're available at the time.
What do y'all think?
Post by: Eagleon on October 15, 2013, 02:02:22 pm
The last thing I want is for this project to be bogged down, and I'm starting to see it's somewhat of a problem when I'm proposing things to you guys that aren't necessarily even needed or fitting with established CS. Or worse, trying to reinvent the wheel, and succeeding somewhat but with ass-pulled terminology. This list (http://en.wikipedia.org/wiki/Design_pattern_%28computer_science%29#Classification_and_list) is quite humbling, and makes me wish I had gotten into school again for the winter, because at the same time it's exciting to see so many that I've worked out in the past through experimentation on my own.
I think working with a dummy database for a while will solidify what knowledge I do have of the process a bit better, but I'm not about to try to work out the model for the protocol and database on my own. I'll still be able to help debug and implement what you guys come up with. Consider me your pet code-monkey intern, because I believe this project could be fantastic if finished.
Post by: Anvilfolk on October 15, 2013, 02:41:07 pm
I'm for additional expertise. I tried to get my sysadmin friend in on the project, but he hates his job too much to have any motivation to contribute, haha. I've got skype - it could be really productive just to get people even peripherally interested in the project, with the page bookmarked and ready to pass along.
Hahahah, poor sysadmin! :D
We should definitely try using google hangouts at some point to discuss things more easily, especially if we're working together on a proposal. Bouncing ideas off each other is amazing.
The last thing I want is for this project to be bogged down, and I'm starting to see it's somewhat of a problem when I'm proposing things to you guys that aren't necessarily even needed or fitting with established CS. Or worse, trying to reinvent the wheel, and succeeding somewhat but with ass-pulled terminology. This list (http://en.wikipedia.org/wiki/Design_pattern_%28computer_science%29#Classification_and_list) is quite humbling, and makes me wish I had gotten into school again for the winter, because at the same time it's exciting to see so many that I've worked out in the past through experimentation on my own.
I don't know most of those patterns myself! Doing so comes with time, from writing code and from pursuing self-improvement while you do so. Heck, if you pick up a book on the subject and read a bit every night, you'll probably know a lot more than I do in a month's time.
Here's my personal take on "established CS", or any other field for that matter. The more you know of it, the less creative you become. You try to pigeon-hole what you want to do into one of the ways you already know of doing stuff, even though it might not fit perfectly.
What you proposed so far over on google docs *is* being used, including the site-server idea which had absolutely not occurred to me. I really think part of the open-source collaboration experience is learning from everyone else, so you should really propose any ideas you have for the comms protocol, or whatever else.
It's just important to realise that every proposal is meant to be very critically analysed and discussed by everyone. They're not personal attacks, nor personal shortcomings, nor "bogging down". It's just how you should do science, research or any piece of reliable and quality software.
Consider me your pet code-monkey intern, because I believe this project could be fantastic if finished.
I absolutely will not, nor should you.
I'm going to stop here, but I'm going to send you a more personal PM that I'd like you to read!
Post by: Eagleon on October 15, 2013, 03:58:12 pm
It's one of the reasons I'd rather major in informatics, though I'm not sure yet that it's much better. I think that if you put all the systems and techniques and algorithms developed in software engineering into a graph, and started building connections between them based on papers that utilize those elements, and then start drawing new elements out of those papers, you'll see a mind-boggling amount of redundancy that's been missed from lack of complete understanding by its participants. It wouldn't be perfect, but at the very least it would be visible, literally, to anyone that cared to look. This project would allow very rapid collaboration on something like that.
The same goes for, say, chemistry, which has had an explosion of papers as the cost to work on it goes down, but not much work done to unify their relationships. Manual navigation to references that may be completely unavailable. Review of papers requiring the same navigational effort. All kinds of redundant and repeated efforts that bog down the research process to a standstill, until someone, as you said, steps outside this comfort zone and gets a little creative or (re)forges a path on their own.
It's not that this would eliminate the ability or need to be innovative/review, either - it would just break the papers up so that you could see all of the work and data that's been checked by complete accident in distantly related work that researchers are insulated from through this barrier.
I suppose instead of giving up entirely, I could try harder to avoid using words I don't have the background for. It comes from needing an internal set of terminology to work quickly and self-document. At the same time I do know I've been a little more egotistical with flashy nonsense words than is helpful. Trying to stopper that and just communicate is not my strong point yet, heh.
Post by: Anvilfolk on October 15, 2013, 04:11:22 pm
It's even worse in the IT industry though. They have acronyms for EVERYTHING...
Also, research is very distributed in nature, and terminology ends up evolving more organically. It's not an organised process at all, so a lot of it is definitely idiosyncratic. But then again, it's impossible to keep up to date on everything being published, so it's hard to merge names.
We should make an effort to link all the concepts we talk about from now on to wikipedia articles. That could help us slowly develop names for everything we mean!
Speaking of which, the names I used for the architecture are completely random and totally not standard, as far as I know. We should totally come up with names that make sense! I'll try to add some suggestions to sane names to the google doc later.
Post by: Anvilfolk on October 15, 2013, 06:56:53 pm
---
Update: stuff is working after all. Apparently the router won't let me access local network machines by their outside name. FINE, ROUTER, I CAN USE ITS FREAKIN' IP.
Currently upgrading the server from squeeze to wheezy. Hopefully it won't break all the things.
Here's the base setup we should get from this:
- Apache
- php
- mysql
- phpmyadmin
- SSH access
I will distribute passwords to those interested in participating in the project.
---
Server upgrade successful! Getting Oracle JDK 7 there as well now, though it's a bit of a manual process. We might have a basic database up today at some point.
Post by: Anvilfolk on October 15, 2013, 11:37:40 pm
Also, created a new repository for a Java implementation of the outcome computation, and pushed some initial code in. Might be able to finish it up in the next few days.
If anyone feels like they understand the DB enough, we could really use sample debates to start testing. Perhaps something from the SAA papers?
By the way - more collaborators = more better, so if you're interested, it's a great time to start participating before there's a large codebase!
Post by: Solarspot on October 17, 2013, 06:59:16 am
QuoteThat's not exactly what I was thinking of here though. What we should make possible, eventually, is for an argument here on Bay12 to reference or be referenced by something on some other website unrelated to Bay12.So you want something like the pods from Diaspora? Is that even possible?
How would searches work? Every search by a user makes a request to all pods?Quotelet's go ahead and create the repositories for each component and start a'hackin'We still need to decide on one language/framework to use.
While I'm engaging in architecture-astronautism, could a network like this be searched from a client-side index? Something like http://www.yacy.net/en/ (http://www.yacy.net/en/) that records everything any users sees in every client's own index, then users search their local indexes. This avoids querying every server ever a googol times a second in the name of searching the network.
PS: Reading this thread while listening to ambiance music was trippyyyy
Post by: Anvilfolk on October 17, 2013, 07:07:27 am
I guess yeah, you could definitely see the Agora servers as a P2P network!
Here's the cool thing though. Every argument is annotated by its source, meaning its website. So whenever anyone - client, server, computation node, etc - needs to go fetch information about that argument, it always knows exactly where to go. You don't need a directory to search in!
We will need a smart way to update argument acceptabilities across servers though, and that's pretty much an open question still!
Get involved if you're interested - there's... well... everything left to do :D
Post by: acetech09 on October 17, 2013, 06:06:13 pm
P.S. I really just have to point out that when I see this thread the first two things that pop into mind is the Agera, a swedish hypercar, and Amora, a personal game project I've been working on. I do get the whole root thing though - I'm mainly suggesting that five-letter words capped by A's with a consonant-vowel-consonant center are tried and tested good names.
Post by: Angle on October 17, 2013, 06:58:01 pm
Edit: Ta Da!
(http://img.ie/images/391ku_thumb.jpg) (http://img.ie/391ku.jpg.html)
Note, in an actual implementation, there would probably be several main indexes, one for each subject, like we have here on bay12. But for the purpose of this graph I only needed the one. As you can see, there is some debating going on here, but there is also a lot of other communication as well. There are of course a great many things that will have to be determined. For example, what are the best practices for using this forum? Should posts be broken up into many smaller ones, or should they be consolidated? What content is valuable, and what is waste? etc, etc.
Post by: Anvilfolk on October 17, 2013, 10:47:05 pm
Watching with interest. Saw this from the beginning, am impressed with how it's come along. I just personally can't see a point to it, but I never really enter in actual online debates so it's out of my usefulness range. If I ever find time in my day I'd be interested in helping out though.
Well, project's open-source, so you can just come and play around with it as well. We can definitely set it up so that there's lots of small tasks, and you can tackle one whenever you have time. I'm thinking the GitHub Issue tracker can be used for posting all sorts of small tasks and to-dos :) Your call! Either way, we'll be here!
The project can serve two purposes. First of all, it helps organise people's thoughts on ongoing discussions. Like Angle posted, the graph really tells people who is talking about what, what each post is responding to, and generally encourages people to be more clear in/ the way they communicate.
Another purpose, which I personally find more interesting, is that it helps you in any decision-making progress. It doesn't need to be a debate, it can be anything where you want to reach a consensus with a group of people. If you use this attack relation to propose things or counter things other people have said, the system can give you how much each proposed thing should be accepted, based on votes and the discussion structure.
I can absolutely imagine this being used in debate teams in high schools, local government, to make company decisions, and just generally all over the interweb, where people care about what they are talking about.
Post by: Anvilfolk on October 20, 2013, 03:46:25 pm
We've gotten together a basic infrastructure for the server, which should be run as a daemon, and are now at the part where we are starting to implement communication between client and server. We have a database up to test things, and all looks good.
It'd be great to get some more interested people to participate. We are trying to keep the GitHub repos filled with issues representing features that still need to be added, so hopefully there's always something you can help do.
We tend to meet and organise in the Google Doc chat, linked in the OP. If there's someone around you can join the chat thing and talk to everyone. If you are interesting, please do drop by and say hello! :)
Post by: Antsan on October 25, 2013, 04:49:54 pm
Post by: Angle on October 25, 2013, 06:02:52 pm
Post by: Anvilfolk on October 25, 2013, 11:09:54 pm
Post by: Xgamer4 on October 31, 2013, 06:47:16 pm
Comment:
As a guy with a B.S. in Mathematics, the conversation about CS naming conventions somewhat amused me. At least CS naming conventions make vague intuitive. Mathematics has open sets and closed sets - but they're not opposites. Sets that are both closed and open are clopen, for example.
Question:
How is this not partially solved by Reddit's thread system?
Post by: Angle on October 31, 2013, 08:20:41 pm
Post by: Anvilfolk on October 31, 2013, 08:32:47 pm
And yeah, hopefully the system is going to be flexible and extensible enough that you can use it for lots of different purposes.
We definitely need to work on proper presentation if we want to get more people involved. Right now we are not really getting the idea out there. I've been talking to a good friend of mine who's all into entrepreneurship and graphics design. I think if he finds the time to participate he'll be a fine addition to the project.
Post by: Anvilfolk on November 01, 2013, 05:00:15 pm
Spoiler (click to show/hide)
Doesn't look like much, being able to remotely log in means most basic systems and hurdles are done. We've got a communications protocol based on BSON that works, we've got client-side and server-side libraries capable of interfacing with each other, and a server capable of talking to the database server. They are also built to be extensible, so that adding new features right now should not be too painful.
Next up is, of course, logging off, then requesting threads by ID, then adding/removing arguments, voting, etc. Once that is done, we've got mostly everything we need to develop an application.
All of these are individual and independent tasks. They're really nice little projects that you could use to get to know the code, so now's a great time to start participating if you haven't already! :)
Post by: Singularity125 on November 01, 2013, 06:43:58 pm
Funnily enough, I think I found the siglink to this in the Roguelike Development Megathread, and this interesting, interconnected sort of data structure also seems like it'd be neat for objects in a game, in some way.
I've also always been interested in knowledge webs. I would set up a mind map for personal use except that I'd never remember to keep the damn thing updated. But in other contexts, like Wikipedia or TvTropes, that interconnected data keeps me busy for hours. Whether that's good for productivity on the other hand... :P
If I weren't so busy I'd help out in some way... I also have no Java experience, although Java code really looks similar to C#, which I do have experience with.
Post by: Anvilfolk on November 01, 2013, 09:56:45 pm
I wouldn't necessarily suggest that you start working on a C# version of AgoraLib (the client-side libraries) right now because things are changing at a very rapid pace in these initial stages, but if you'd like to learn Java and/or the technical parts of the project, drop by the Google Doc. We meet there to chat and bounce ideas, and we'd be happy to explain what's happening to you :)
Or anyone else! :)
Post by: Eagleon on November 02, 2013, 11:12:42 am
getBSON() works like this - it works through each BSONable object, calling its getBSON() method to add to either a BasicBSONObject or BasicBSONList (depending obviously on if it's something like a list of nodes/edges or a single property like information about a post). Then it reads through its own data, adding things like ID numbers, sources, etc - whatever isn't a BSONable object. Finally, it returns the constructed BSONObject. That's it.
Then, to deserialize, you feed any object a BSONObject. In the object's constructor, you retrieve the entries that matter to that object - if it fails, it only fails for that object, and you have the specific BSONObject it used available telling you why. But what's really beautiful (I think) is that it uses BSONObjects to build most of its entries - that is, you call 'new PostInfo(BSON.get(PostInfo.JSONTag))', and PostInfo is built from a BSONObject stored in what the node has received. It's also easy to send partial information back and forth about an object - just leave it out of the BSONObject you're building (using a seperate method than getBSON obviously, maybe getBSONEdges() for instance) and methods in the exact same class that merge them into an existing object. And if you wanted to add something weird down the line like a PostInfo to an edge, all you'd need to do is add it as a field, modify that method, and add it to the BSON-based constructor.
A few extra additions of note - EdgeID/NodeID are gone, replaced by a master ID interface. That can be changed back easily by changing out constants JSON_ID/JSON_SOURCE with unique tags for each BSONable if it causes problems for the database. I suspect that to be the case now, thinking about how this would have to be moved into/from the database - I'm sorry. There's also a nice Property abstract for things like Post information, user data, etc. Might be a useless distinction at this point.
I do think this is a little more extendable on the client/server side than the current system of adding BSON serializers/deserializers to a master list in JAgoraLib. I don't know for sure if the database can handle this kind of nested structure - hence the branch. I didn't want to assume it was useless after doing all of the work to convert it. I can't contribute much more at the moment - got laid off, again, after only five days back on the job. Stressful times! Hopefully this is at least somewhat helpful, and not a diversion from actual work - I should probably have been working on a client but to be fair at the time I was working on it, it didn't look like the server was functional =P
Also Also - Singularity, I encourage you to dive into Java, but only if you're willing to put up with its crap. Specifically, if you're used to using lots of functional programming, it's going to be painful. If you have lots of scripts in other languages that you like to use, it can be done, but it's not designed for that like C#. C# is definitely very similar for OOP, if not identical for some tasks. It was easy for me to work on C# projects in the past, though there are enough differences that you'd want to skim through the list (http://en.wikipedia.org/wiki/Comparison_of_C_Sharp_and_Java) to begin with.
Basically, Java is just an older language built almost entirely for OOP. Other design patterns often have to jump through some ridiculous hurdles just to function, but it's otherwise perfectly usable. C# is more flexible (pointers! I miss pointers... and operator overloading, and real multidimensional arrays, and, and...) because it was designed from the beginning to support interaction with non-OOP, non-C# code. It's quite a bit more modern and featureful. Java will smash you over the head for things that it sees as potentially bug-inducing, even though bugs are still common, while C# (in typical Microsoft fashion =P) has ways to facilitate many of the most horrifyingly buggy things Java has tried to avoid. It surprisingly does a pretty good job of it.
There are more keywords in C# for simple tasks, which can make things easier if you know the language, whereas Java has avoided adding more keywords at all costs, limiting core functionality to preserve backwards compatibility with ancient code - which is why it boggles my mind that Oracle only recently started discouraging keeping around old JREs with security vulnerabilites to exploit. It's probably easier to transition Java->C# rather than the other way around, but I think it's worth working on a project in a different language than you're familiar with just to know that you can.
Post by: Anvilfolk on November 02, 2013, 01:28:38 pm
Your proposal handles something I've been thinking about for the past week or so.
Would it help to have the BSON stuff encapsulated inside objects that should be converted to BSON? I put up a branch with a rough implementation of what I'm talking about - basically every object in the Graph that isn't a basic datatype (although technically we could overload those too if we wanted to go crazy with it) has a BSONable interface with a single getBSON() method, plus a string for its JSON tag.
The general idea I've been following is that we wouldn't want to go crazy with it. The more crazy we go, the harder it is to port to other languages, and we run the risk of having LOTS of different core Agora implementations (for the server, for the lib, for the graph), each with a different supported features. That would make Agora development and communities pretty divided :(
getBSON() works like this - it works through each BSONable object, calling its getBSON() method to add to either a BasicBSONObject or BasicBSONList (depending obviously on if it's something like a list of nodes/edges or a single property like information about a post). Then it reads through its own data, adding things like ID numbers, sources, etc - whatever isn't a BSONable object. Finally, it returns the constructed BSONObject. That's it.
Serialisation, unfortunately, is actually harder than it looks at first :( I think your system is going into infinite serialisation loops when you have a loop in the graph, like A -> B and B -> A (this is going to be super common). Could you check?
If we want to use this method, we'll still need to use something similar to the JAgoraLib serialisation implementation, which is sending the attacks with origin/target NodeIDs rather than Nodes themselves. That should avoid going into an infinite serialisation loop, where A serialises B which serialises A and so forth.
Of course, then you need to deserialise all Nodes, and THEN all Edges, using Node IDs to get to the actual Node.
It's probably easier and more extensible to have an external serialiser. That's how Python does it (pickle and marshall), as well as BSON, and other systems. BSON, for instance, has Lazy and Basic coders/decoders, which are useful in different situations.
Then, to deserialize, you feed any object a BSONObject. In the object's constructor, you retrieve the entries that matter to that object - if it fails, it only fails for that object, and you have the specific BSONObject it used available telling you why. But what's really beautiful (I think) is that it uses BSONObjects to build most of its entries - that is, you call 'new PostInfo(BSON.get(PostInfo.JSONTag))', and PostInfo is built from a BSONObject stored in what the node has received. It's also easy to send partial information back and forth about an object - just leave it out of the BSONObject you're building (using a seperate method than getBSON obviously, maybe getBSONEdges() for instance) and methods in the exact same class that merge them into an existing object. And if you wanted to add something weird down the line like a PostInfo to an edge, all you'd need to do is add it as a field, modify that method, and add it to the BSON-based constructor.
Could you explain the PostInfo stuff a bit more? I didn't really understand how you use it. The idea of having partial information sent to the client is awesome, but I'm worried that it won't have that much use because the server can't send stuff to the client without the client explicitly asking for it first. Since communication is always started by the client, why doesn't he just always get all graph information, rather than partial graph information?
A few extra additions of note - EdgeID/NodeID are gone, replaced by a master ID interface. That can be changed back easily by changing out constants JSON_ID/JSON_SOURCE with unique tags for each BSONable if it causes problems for the database. I suspect that to be the case now, thinking about how this would have to be moved into/from the database - I'm sorry. There's also a nice Property abstract for things like Post information, user data, etc. Might be a useless distinction at this point.
Yeah, this probably screws with the database a bit too much. Why would we want a single ID class though? Node IDs and Attack IDs are pretty different in nature. I mean, we could just have an ID be a String, but it probably just makes it harder to understand the code (e.g. what is this id used for? Is it a Node ID, or a thread ID, or an attack ID)?
I do think this is a little more extendable on the client/server side than the current system of adding BSON serializers/deserializers to a master list in JAgoraLib. I don't know for sure if the database can handle this kind of nested structure - hence the branch. I didn't want to assume it was useless after doing all of the work to convert it. I can't contribute much more at the moment - got laid off, again, after only five days back on the job. Stressful times! Hopefully this is at least somewhat helpful, and not a diversion from actual work - I should probably have been working on a client but to be fair at the time I was working on it, it didn't look like the server was functional =P
Agreed with the first part! We should extract the (de)serialisers from JAgoraLib and make them into their own thing, perhaps as part of AgoraLib, but still not inside JAgoraGraph, for the reasons I stated above.
I've been thinking a bit more about how AgoraLib would actually be used and honestly, I don't think I've reached a good solution.
I believe our best bet to maintain Agora as a nice tight-knit project rather than a million different implementations with different features is relying heavily on the content column of the Node table in the database, which is meant to store flexible BSON content. So, JAgoraLib ONLY handles serialisation of the very basic graph structure, like nodes, attacks, the posters of nodes/attacks. It ignores the BSON content of a node - but sends it over the network (which is trivial). It's up to the user to parse the BSON content of nodes.
The extendibility is obtained by using and parsing the BSON content that comes from the database. This is actually exactly what you have, but for the content, not the graph itself! Let me try to make the two proposals more obvious. I've highlighted the differences.
Proposal #1 (master branch):
- User requests thread
- AgoraLib contacts server, asks for thread
- Server receives request, queries database for all arguments/attacks in that thread
- Server converts database response into AgoraGraph, using server-side library (client never interfaces with DBs)
- Server-side AgoraLib serialises AgoraLib into BSON, and sends it over network
- Client-side AgoraLib deserialises BSON from network into AgoraGraph
- The user's/application's JAgoraNode subclasses parse the the Content (BSON) of nodes into whatever.
Proposal #2 (BSONable branch):
- User requests thread
- AgoraLib contacts server, asks for thread
- Server receives request, queries database for all arguments/attacks in that thread
- Server converts database response into AgoraGraph, using server-side library (client never interfaces with DBs)
- Server-side AgoraGraph serialises itself into BSON
- Server-side AgoraLib sends BSON over network
- Client-side AgoraLib reads BSON from the network
- User/application-side AgoraGraph constructs itself from BSON
So really, the only difference is in who serialises what. I think the best solution is actually a Proposal #3 where we separate the serialiser into its own class. Perhaps serialisation is done into byte[], rather than BSON, allowing us to easily change the network protocol itself?
Then the other part would work somehow like this:
Code: [Select]
public class ApplicationNode extends JAgoraNode {
...
public void loadContent() {
this.x = rand();
this.y = rand();
// this.content is a member of JAgoraNode, populated from the content column of the DB
this.text = this.content.get("textArgument");
if (this.content.get("imageURL") != null)
this.image = Image.load(this.content.get("imageURL"));
}
...
}
So this implementation would be able to handle arguments having some piece of text, as well as one attached image. That all comes from what's inside the content column of the database. We could extend it to have a youtube video, several images, etc etc etc. Anything that you want - but that's application-side. The Agora Project itself could officially support some basic format for Content, but then it's up to the app developers what else they want to use. If some "content" feature starts becoming popular, we could add it to the official "content" description.
Also, do you guys think attacks should also have content?
Post by: Eagleon on November 02, 2013, 02:32:58 pm
Serialisation, unfortunately, is actually harder than it looks at first :( I think your system is going into infinite serialisation loops when you have a loop in the graph, like A -> B and B -> A (this is going to be super common). Could you check?Well, since we're working with bidirectional edges (and a Graph object that stores them seperately), we can do either based on the situation. From edges we can add nodeIDs to a map, and then call a Node constructor that doesn't create edges, but rather just pulls it from the Edge map. Or from nodes, we add edges and call one from Edge that uses the Node map to construct itself. Does that make sense? This was only a very fast, not even tested implementation - I didn't have a database to test it against, couldn't figure out how to modify JAgoraLib to work with it, etc. It probably has a billion uncaught bugs like that.
If we want to use this method, we'll still need to use something similar to the JAgoraLib serialisation implementation, which is sending the attacks with origin/target NodeIDs rather than Nodes themselves. That should avoid going into an infinite serialisation loop, where A serialises B which serialises A and so forth.
Of course, then you need to deserialise all Nodes, and THEN all Edges, using Node IDs to get to the actual Node.
Could you explain the PostInfo stuff a bit more? I didn't really understand how you use it. The idea of having partial information sent to the client is awesome, but I'm worried that it won't have that much use because the server can't send stuff to the client without the client explicitly asking for it first. Since communication is always started by the client, why doesn't he just always get all graph information, rather than partial graph information?I was thinking for client->server, and internal client/server functionality actually - the client sends a node with only the PostInfo inside, and the server can function without the rest using methods inside Node. Rather than put that functionality in the main Comms library, nodes know how to modify themselves using PostInfo objects, because they're the ones that contain/receive them.
Yeah, this probably screws with the database a bit too much. Why would we want a single ID class though? Node IDs and Attack IDs are pretty different in nature. I mean, we could just have an ID be a String, but it probably just makes it harder to understand the code (e.g. what is this id used for? Is it a Node ID, or a thread ID, or an attack ID)?I'm not sure, beyond busy work, haha. Premature optimization, etc. I don't even know if we need seperate ID classes. I do try to complicate things =P
I've been thinking a bit more about how AgoraLib would actually be used and honestly, I don't think I've reached a good solution.I agree for the most part. I don't have the knowledge in advance to evaluate either proposal, but one of the reasons I made the properties abstract was in anticipation of this kind of extra content. There's a default JSON tag other than null for it for that reason - if there's non-standard content to be sent, it would potentially use that class. What I don't want to have happen is stuff being stored in the database the server doesn't know the identity of - as far as I understand that could cause some potentially nasty exploits happening for non-VM languages, with executable code, etc. Of course, you have the same problem with the server examining the content of a property for what it is potentially exposing it to the same thing, but if nothing else, there needs to be some kind of sanity check for the content being uploaded so that people don't just use it for a file server and clog up the whole network with gigabyte-sized nodes. Stop me if I'm not understanding something that prevents that.
I believe our best bet to maintain Agora as a nice tight-knit project rather than a million different implementations with different features is relying heavily on the content column of the Node table in the database, which is meant to store flexible BSON content. So, JAgoraLib ONLY handles serialisation of the very basic graph structure, like nodes, attacks, the posters of nodes/attacks. It ignores the BSON content of a node - but sends it over the network (which is trivial). It's up to the user to parse the BSON content of nodes.
The extendibility is obtained by using and parsing the BSON content that comes from the database. This is actually exactly what you have, but for the content, not the graph itself! Let me try to make the two proposals more obvious. I've highlighted the differences.
<snip>
So this implementation would be able to handle arguments having some piece of text, as well as one attached image. That all comes from what's inside the content column of the database. We could extend it to have a youtube video, several images, etc etc etc. Anything that you want - but that's application-side. The Agora Project itself could officially support some basic format for Content, but then it's up to the app developers what else they want to use. If some "content" feature starts becoming popular, we could add it to the official "content" description.
Also, do you guys think attacks should also have content?
And yeah, I think they should. I would absolutely love to adapt this to a neural network model, for Science! (think about the possibilities for treating other portions of content as sensory data, and you're starting to see where I'm going with that) and that would be pretty useful. I'd also like to be able to annotate edges with moderation information, or let users label them, or emphasize votes... as far as I understand it's easy to provide if not identical to what Nodes would need to implement it, and just nice to have.
Post by: Anvilfolk on November 04, 2013, 09:25:45 am
I looked at your proposal for properties, and it scared me that there was no "interface". How would you use it? Would you have an IntProperty, DoubleProperty, StringProperty and then more complex classes like MapProperty, ListProperty and so forth? If that's the idea, you're essentially reimplementing BSON! We might as well just have BSON and allow people to parse that BSON content into their own Graph classes.
That's an extremely good point about possibly having harmful content inside the database. We should definitely make sure we're not allowing it. I think the current situation is as follows: if I'm not mistaken, our JAgoraLib implementation can only transmit BSON objects over the network. Any request you send, or graph you send, or anything, is encoded into BSON. That should include the BSON content of a Node. If you try to send arbitrary bytecode over the network, either as part of a Node's content or otherwise, it should fail because decoding that bytecode should fail as well. We should definitely test this theory out though!
Just checked, and you can send binary data as part of a BSON object. However, that will never be executed unless an application explicitly considers it to be an executable and tries to run it. Otherwise it's just a byte array.
In other words:
- Send malware_byte_code through network, and decoding that into a BSON Object fails. SAFE!
- Send {"executable"=malware_byte_code}, and it's parsed into a BSON object, but the evil binary code is stored as a byte[], so it's harmless.
We should be safe, hopefully :)
Post by: Eagleon on November 04, 2013, 11:31:43 am
The Property class does have an interface - BSONable! Thus PostInfo, UserInfo are BSONable. It exists at all because I'm in the habit of making an abstract class to store methods and static members that are useful across objects that are expected to be similar - almost entirely driven by the fact that in Eclipse, refactoring classes to extend a different one later is much more painful than simply delete-refactoring the stub abstract class. That might be bad, hehe.
We should be safe, hopefully :)Dun dun DUNNN yeah, I was being paranoid. I do wonder how people would use this if they -could- store scripts in each node, and allow them hooks into their API for Agora, but that's something for another day.
Post by: LordSlowpoke on November 09, 2013, 04:07:54 am
I'll be looking out for a workable prototype so I can slap it onto a server and watch hell descend upon it. Both software and userbase-wise.
Please look out for things that need to be thrown into the documentation~
Post by: Anvilfolk on November 09, 2013, 09:23:10 am
Also, there's lots of things to do in programming and general ideas if you're interested! :)
Post by: LordSlowpoke on November 09, 2013, 09:33:21 am
The amount of experience I have when it comes to programming network things is zero, so I don't think I can help you much. Ideas I think I can help with, but they'd be more of the design than implementation variety.
Post by: Anvilfolk on November 12, 2013, 11:09:54 am
Let's aim for one a week, preliminarily set at 8pm EST on Wednesdays, starting tomorrow hopefully. This schedule could change wildly, but we'll try to do at least once a week for a couple of hours.
This'll be an excellent opportunity for you to look at some of the code and how it's organised, ask questions, get involved and generally see what's happening in the Agora world :)
TL;DR version:
Now streaming Agora development once a week!
Post by: Angle on November 12, 2013, 11:32:29 am
Post by: Anvilfolk on November 12, 2013, 07:57:37 pm
Post by: Anvilfolk on November 13, 2013, 10:53:49 am
Post by: Anvilfolk on November 13, 2013, 08:02:30 pm
Post by: Angle on November 13, 2013, 09:49:22 pm
Post by: Anvilfolk on November 13, 2013, 09:59:31 pm
No worries Angle - it was great to have you drop by! I am super tired, so I didn't do such a great job of explaining things.
The idea right now is to make it easier for interested people to get in touch, see what's already in there and so forth. It's also a pretty good situation to guide people through code or answer questions. Let me know throughout these things if I'm going too fast, or if there's something you don't understand/know but would like to look at in more detail!
It's definitely also about me making sure to put some time into the project, as life has been getting more and more hectic :)
If you do end up watching, please do leave all suggestions here. I'd love some feedback!
Post by: Anvilfolk on November 20, 2013, 10:50:17 am
If you've been moderately interested in participating and have some Java coding skills, this is a great opportunity to learn how to implement some of the features we need. We have lots of small things that you should be able to tackle pretty easily, and that are really important at these initial stages.
Just post here if you are interested and we can come up with some voice-chat thing so that I can show you how things work and how you can start programming too :)
Post by: Anvilfolk on November 20, 2013, 07:59:58 pm
Post by: Anvilfolk on November 27, 2013, 05:30:10 pm
We really need to start putting this project out there. Is there anyone that's good with webdesign that wants to pick up making a webpage?
Post by: Angle on November 27, 2013, 05:39:07 pm
Post by: Anvilfolk on November 27, 2013, 05:41:15 pm
I've been having some tough weeks, but I think things are getting better and I might have some more time soon for this soon :) In the meantime we've gotten a few more features implemented, so hopefully we're getting closer to version 0.1 :)
Post by: Anvilfolk on December 03, 2013, 05:22:58 pm
Post by: Anvilfolk on December 03, 2013, 06:14:24 pm
As usual, there's a ton of tiny issues that people who are interested can try to tackle and submit commits for. They should be relatively easy to tackle :)
Post by: Angle on December 03, 2013, 07:24:53 pm
Post by: Angle on January 29, 2014, 05:59:33 pm
Post by: Anvilfolk on January 29, 2014, 08:54:27 pm
Post by: Eagleon on February 07, 2014, 12:52:23 pm
Post by: Anvilfolk on February 07, 2014, 01:17:48 pm
I'll need to find some more time for this, but my evenings have suddenly become really busy with other things :(
Post by: Eagleon on February 11, 2014, 01:18:05 pm
Awesome :) I'm doing terrible with finding some nice java library for pretty graphics, for the client itself, so I dunno. I think there's some way to fix the OpenGL problems somehow, but I've spent a few hours to no avail.I've never worked with OpenGL directly (oh god the abbrevs and utter lack of documentation), but I do know libGDX has lots of very nice wrapper classes for working with graphics (edit: thinking specifically of the PolygonSprite and other sprite classes, which support transparency and clean modification of scaling, rotation, translation, etc as well as packing into PolygonSpriteBatch, which could be faster). It was actually a library I was eying for the java side of things to begin with, since it does OpenGL ES and deploys to android or even iOS so easily if you've set things up right, and is nicely integrated with box2D and Bullet for fancy physics if we want to go that route. I can start pushing everything into GDX's classes to get something on screen at least maybe if I don't mess things up further :D
I'll need to find some more time for this, but my evenings have suddenly become really busy with other things :(
Ubuntu failed me for internet - it doesn't like my wireless one bit, and isn't afraid to drop connection at random for hours at a time to say so! I spent hours Sunday trying to figure out what was happening, but it's been impenetrable. Running ethernet isn't an option unless I want to thoroughly piss off my roommate. Are you still running the test server? I can work on this for about half an hour before work every day, and that should get me back to contributing on a consistent basis, but if anything I'd think linux would like my laptop's hardware even less.
Post by: Anvilfolk on February 12, 2014, 11:26:04 am
Anyway - the reason I was trying to get regular openGL to work is that we would be able to do some nice animations, like circles representing arguments opening up into nice rectangles with rounded edges. I assumed, erroneously, that OpenGL would be nice to vector-like graphics. It's not. We probably just want to go with minimal animations at this point, and just get a prototype out the door :\
Post by: Eagleon on February 12, 2014, 01:39:17 pm
Are you on a laptop? Take a look at disabling USB power saver settings. I've heard those can be big problems, but I'm honestly not sure how to fix them. I think it varies from laptop to laptop, but might be findable in the BIOS somewhere.I've been trying to get my desktop running as a server of my own to understand that part of things better. At this point I'm expecting less trouble setting that up than I've gotten with ubuntu's seeming disdain of wireless internet :P I understand it, ethernet is faster, there's hundreds of wireless adapters out there, no one makes linux drivers because it's an extra thing to support, but arglblgergkguhgff. fbuhfv.
Quote
Anyway - the reason I was trying to get regular openGL to work is that we would be able to do some nice animations, like circles representing arguments opening up into nice rectangles with rounded edges. I assumed, erroneously, that OpenGL would be nice to vector-like graphics. It's not. We probably just want to go with minimal animations at this point, and just get a prototype out the door :\There's probably a way to do it using regular openGL. I should probably learn GL ES or parts of it, because yeah, it's a lot more flexible to be able to do these things yourself without being tied to a library. I definitely wasn't trying to discourage you from that if you have the knowledge already. Just got excited seeing a toolbase that I've used and more or less understand :)
You could do the same sort of thing with PolygonSprite as far as I can see, using getVertices() and a tweening method to translate each one out towards your target shape, making and destroying new sprites each update. Punch a box-hole in the polygon, or decompose it into two tied at their edges if their math can't deal with holes, and you have a border. Or leave it alone and source your texture from the node's text data, and you have the interior. Playing around with TextureRegions, you should be able to make the border maintain rotation and scale to avoid weird distortions if what you're using is a meaningful pattern like wood or something like that.
But yeah, if nothing else, it's completely possible to drop GDX's classes into something that works directly with OpenGL to build closer to metal, or I wouldn't be suggesting this.
Post by: Anvilfolk on February 12, 2014, 01:45:48 pm
The basic code is there, but I hadn't looked at PolygonSprite! Feel more than free to pick up the code and make something work. I'm still super busy at the moment :)
To be honest, as long as we keep a nice separation between the data and the visualisation, it doesn't matter how we implement the initial parts. We can always reprogram the visualisation at some point. At this point I don't think it even matters that we have something that looks decent. Just something that works :D
Post by: Angle on April 14, 2014, 02:32:55 pm
Code: [Select]
ImmediateModeRenderer imr = new ImmediateModeRenderer10();, but when I look at the libgdx api (http://libgdx.badlogicgames.com/nightlies/docs/api/overview-summary.html), I don't see ImmediateModeRenderer10, I see ImmediateModeRenderer20.
Post by: Anvilfolk on April 14, 2014, 02:36:18 pm
OpenGL does not seem to be the best idea anyway. It's far too low-level and we should try to find a nice vector graphics library for Java. I wasn't able to do so, unfortunately. Work's been an absolute nightmare for the past couple of months, and is likely to stay pretty demanding for the next few months/years, but I'd like to get back on track with this if I can find some spare time :\
Post by: Angle on April 14, 2014, 03:13:19 pm
Post by: Anvilfolk on April 14, 2014, 03:14:32 pm
Post by: Angle on April 14, 2014, 03:19:58 pm
Post by: Anvilfolk on April 14, 2014, 03:23:30 pm
It's used in TextClient.java (https://github.com/Agora-Project/JAgoraLib/blob/master/src/org/agora/TextClient.java), which is a simple text client for testing and debugging purposes, i.e. no graphics ;D
Post by: Angle on April 14, 2014, 03:25:49 pm
Post by: Angle on April 22, 2014, 02:15:44 pm
Next, of course, there are the actual policies to be followed. I think that most any topic should be up for debate, but that the rules for this debate must be very strict, but mostly restricted to logical fallacies. For example, it should be legal to discuss things like "We shouldn't vote for What'sHisFace because he's an x and a y", but things like "don't listen to What'sHerFace, she's an x and a y" should not be acceptable, because that's an ad hominem and detracts from the discussion. In some cases, I think people should be held accountable for their arguments - for example, if you want to argue that "all x people are y, and that's a bad thing", then you'd better either make a good argument, or apologize and rescind your position, or face consequences for it.
You can disagree, of course, and I rather hope some of you will - and of course, I look forward to hearing about it. I have to go to work now though, so I'll catch up when I get back.
Post by: Anvilfolk on April 23, 2014, 12:51:12 pm
Alright, so I'm working on putting together a graphical client using java's AWT and swing frameworks.
Sweet! :D I really should learn how to use those! Do they support hardware acceleration? I think the reason I discounted them is 1) I don't know them yet and 2) always thought they might underperform. Keep us updated on how this goes!
I think the system will need fairly heavy moderation in order to function properly. In order to counter balance the possibility of abuse this presents, I think we need a system of moderating the moderators. I think that all moderating activity should be logged and open to criticism. This should prevent moderators from abusing their privileges to promote their ideas at the expense of others.
This might be going overboard with features.
Forums works fine with a simple hierarchy of administrator->moderator->regular user. Why would it be insufficient here? I'm not sure we want to allow "meta-debate" to influence the debate itself, e.g. a debate about politics shouldn't eventually devolve into whether a moderator was right in removing this or that message.
Furthermore, these debating frameworks are self-moderating. The results of debates are a direct consequence of the community that engages in that debate. Communities with lots of trolls will get troll outcomes, and mature communities will get interesting, significant outcomes for their debates because they are able to downvote troll arguments into insignificance. That's something we need to accept, much like the creators of phpBB or SMF did. We're providing a platform. People are going to use it differently in different parts of the internet.
Next, of course, there are the actual policies to be followed. I think that most any topic should be up for debate, but that the rules for this debate must be very strict, but mostly restricted to logical fallacies. For example, it should be legal to discuss things like "We shouldn't vote for What'sHisFace because he's an x and a y", but things like "don't listen to What'sHerFace, she's an x and a y" should not be acceptable, because that's an ad hominem and detracts from the discussion. In some cases, I think people should be held accountable for their arguments - for example, if you want to argue that "all x people are y, and that's a bad thing", then you'd better either make a good argument, or apologize and rescind your position, or face consequences for it.
This is up to each particular host that uses the Agora framework, much like every forum has different policies. Of course, we should absolutely provide a website that is Agora-enabled for people to try it and understand what it's all about. And this website, like any other, is going to require an administration and moderation team as well as moderation policies.
I kind of feel that if a user consistently cites unreliable sources, it's a significant contribution to a debate that he might be doing so again. Why not let it happen organically?
Although I can roll with having a super serious debating subforum where this isn't allowed, and users follow up on citations to find what is wrong with them, etc. Then it makes sense that ad hominem arguments should be against the rules, and removed by moderators.
Post by: Angle on April 23, 2014, 04:45:08 pm
As for policies, I agree that many bad arguments should be allowed, so that the people arguing have something to cut their teeth on. There are a few things that should not be allowed, such as ad hominems and pointless insults - these make the debate excessively hostile and make it far less likely that people will learn new things from it.
There's also the question of what discussions to allow. If a bunch of Neo-Nazis want to debate Naziism, do we allow that? I think we should, though with strict moderation, because it might end up changing some of the Neo-Nazi's minds.
I know that many of these are topics about the use of a particular forum, not about Agora as a whole, but before Agora sees greater adoption, we'll need to demonstrate it's capabilities.
Post by: Fniff on April 23, 2014, 05:59:36 pm
Anyway, nevermind my nerdy indulgence, PTWing with interest.
Post by: Angle on April 23, 2014, 09:56:31 pm
Post by: Anvilfolk on April 24, 2014, 10:15:46 am
Wow, I've rewritten this post about 3 or 4 times now :\ I'll try to keep it short this time!
Moderation policies will be up to whoever is hosting the Agora service. I think as the official Agora website, we probably want to have several subforums for different levels of seriousness, and different levels of moderation, and we should discuss policies for those at some point.
Other than that, different websites will define their own moderation policies, like forums do these days.
I don't think we need moderation for moderators. There is always potential for abuse, be it moderators, administrators, governments... ultimately, you frequent websites with whose community, moderation and administration you're comfortable with.
It's the administration's responsibility to keep moderators in check. If anybody is unhappy with the moderation, they can take their concerns to the administration, or can start a debate on that very issue - after all, that's what Agora is all about :D This is how most forums work these days, and the system works OK.
Having everything super public and open for debate will just really confuse discussions and give them lots of meta-layers that will dilute the issue under consideration. I quite like how Bay12 is moderated, for instance. Someone reports the posts, Toady comes in, removes the flame wars posts, and writes a comment saying what happened. Why wouldn't that work?
Post by: Angle on April 24, 2014, 11:39:45 am
I don't really think there'd be a problem with there being huge debates about a moderators action - either it's obviously wrong, or it's not.
Post by: Anvilfolk on April 24, 2014, 01:01:34 pm
Post by: Angle on April 24, 2014, 02:33:44 pm
Post by: Anvilfolk on April 24, 2014, 02:37:33 pm
Post by: Angle on April 24, 2014, 02:49:55 pm
On another note, I can't seem to get the text client to work. It's giving me this error:
Code: [Select]
[JAgoraLib] Error opening connection to bigornas.bounceme.net:1597 (Connection refused)
[JAgoraLib] Could not connect because socket could not be opened.
Did we change the settings on the server? I can still log in normally through the web page, so I know it's still up. and I updated to the latest version of all the agora projects, so I'm pretty sure it's not that.
Post by: Anvilfolk on April 24, 2014, 02:58:16 pm
You should run a local server! Then the server connects to the database which is running in bigornas, and your client connects locally to the server. Does that make sense?
Post by: Angle on April 24, 2014, 03:15:34 pm
Post by: Anvilfolk on April 24, 2014, 03:16:18 pm
Post by: Angle on April 26, 2014, 06:03:37 pm
Edit: Looks like it is. I've been playing with it, and it seems some of the posts on the test server are messed up, cause when I try to get threads 0 or 1 it gives me errors. 2 works fine, so far as I can tell, and anything above that does nothing, which is not surprising as there are no threads above that point.
Post by: Anvilfolk on April 28, 2014, 01:29:27 pm
Check the database - I am not sure there ARE threads 0 and 1. I've deleted a few of the previous threads because they were no longer in the format that we needed for the DB. But definitely use phpMyAdmin to check the DB out.
Error reporting is TERRIBLE right now. That's one of the things we need to improve, definitely.
Post by: Angle on April 29, 2014, 06:23:47 pm
While I'm at it, though, should I add an object to JAgoraGraph to represent a thread header? Or should I just use argument for that?
Post by: Anvilfolk on April 29, 2014, 09:46:43 pm
I'm not sure whether to the header should be part of JAgoraGraph or some other structure. My original idea was to just have graphs, which you could potentially merge so you'd have multiple threads linked together. Perhaps JAgoraNodes can have the thread information?
Post by: Angle on April 29, 2014, 09:53:33 pm
Post by: Anvilfolk on April 29, 2014, 10:27:26 pm
Post by: Angle on April 30, 2014, 09:53:26 pm
We also might want to consider making posts have lists of the other posts the are linked to, so that we don't have to scan the entire database whenever we need to find a posts relationships.
Post by: Anvilfolk on April 30, 2014, 10:44:44 pm
Allowing posts to exist in multiple threads makes queries less efficient because that makes another table. The way I was thinking this could be at least initially implemented is that each post is originally proposed in a thread, and that's what it's associated to. But the cool thing is that the attacks/arrows/relations don't need to be between arguments in the same thread. So even though an argument belongs to a thread, it can attack or be attacked by any other argument.
The way this is used would be something like this:
- Ask for thread #1
- Get curious about a particular argument from thread #2, which attacks an argument from thread #1
- Ask for thread #2, then merge threads #1 and #2
What currently happens is that arguments from other threads do not get sent to the client (until explicitly requested with their thread ID). They are, however, represented as "placeholders", hence the current isPlaceholder() method in JAgoraNode.
Post by: Angle on April 30, 2014, 11:42:37 pm
Post by: Anvilfolk on May 01, 2014, 09:02:19 am
Post by: Angle on May 01, 2014, 02:07:54 pm
Post by: Anvilfolk on May 01, 2014, 02:11:59 pm
Databases can't really handle recursive queries, which is what you'd need to do if you wanted to get all arguments somehow connected to a given argument. The advantage of each argument having (at least) one thread is that you can get all arguments from that thread at once, then grab all attacks related to those arguments. That's pretty efficient (and it's the way it's currently implemented).
If you grab the initial post, then query again for the posts at 1-step, then again for those at 2-steps, etc etc, you're making a ton more queries. Not to mention that you have to send the database all the posts that you already have, so that you can grab all the posts somehow related to THOSE. And then you get an even larger number of posts, which you have to send again, and so forth.
Similar things would happen if you had to fill in the blanks from the first post and the leaves. Also, there will be cycles, meaning there might be no leaves!
Post by: Angle on May 01, 2014, 02:20:26 pm
Post by: Anvilfolk on May 01, 2014, 02:22:43 pm
For example, all pro/con debates will start out with a cycle, one pro, one con, attacking each other!
Post by: Angle on May 01, 2014, 02:46:22 pm
Post by: Anvilfolk on May 01, 2014, 03:09:00 pm
There's a nice example here (http://centria.di.fct.unl.pt/~jleite/papers/tafa13.pdf), on page 8.
Post by: Angle on May 01, 2014, 04:22:25 pm
Post by: Anvilfolk on May 01, 2014, 07:19:55 pm
Post by: Angle on May 01, 2014, 08:47:49 pm
Post by: Anvilfolk on May 01, 2014, 08:59:38 pm
DBGraphDecoder is the one that does the querying, so I guess you can get the graphs from there. Then you can have a ForumThread class somewhere in JAgoraLib (or should it be part of JAgoraGraph somewhere?). Then you'll need to encode and decode that into BSON.
Post by: Angle on May 01, 2014, 10:12:43 pm
Post by: Angle on May 02, 2014, 05:37:31 pm
Post by: Anvilfolk on May 03, 2014, 02:25:59 pm
Post by: Angle on May 05, 2014, 09:52:43 pm
Post by: Anvilfolk on May 06, 2014, 08:36:55 am
Post by: Angle on May 12, 2014, 08:02:24 pm
Post by: Anvilfolk on May 12, 2014, 08:03:35 pm
Post by: Angle on May 12, 2014, 09:16:17 pm
Post by: Angle on May 18, 2014, 02:00:04 pm
Post by: Anvilfolk on May 18, 2014, 02:05:56 pm
I'm getting a little bit more time recently, and I'll download this and see how it is! Let me know any specific things you'd like my to contribute to!
Post by: Angle on May 18, 2014, 02:12:08 pm
Post by: Anvilfolk on May 18, 2014, 02:20:16 pm
Post by: Angle on May 22, 2014, 11:10:47 pm
Post by: Anvilfolk on May 23, 2014, 10:31:16 am
Just went to try this out but the code isn't compiling at all. What am I missing? It looks like you might've forgotten to add the JAgoraThread file?
Ugh, nevermind, it's Eclipse acting up...
Post by: Angle on May 23, 2014, 10:37:12 am
Post by: Anvilfolk on May 23, 2014, 10:43:59 am
Other than that I see stuff appearing on screen! Sweet :D :D Did you remove the arguments from the first thread? I'm not getting anything from there. They had votes and everything :(
Quick question: is there any way for us to get a ThreadPanel of sorts that can draw itself, rather than ThreadListPanel drawing the JAgoraThreads under PaintComponent?
Also, you should definitely add all of the tasks you feel need to be done in the GitHub issue tracker. That way it's easier for whoever else to pick up on a task that needs doing.
Post by: Angle on May 23, 2014, 10:48:16 am
I could indeed add a ThreadPanel, and will once I've finished with my current work. I'm also going to go back and turn all of the prompts for host, login, etc. into panels in the same fashion as graphpanel and threadlistpanel.
Good point. I'l see about that right now.
Post by: Angle on May 24, 2014, 01:02:33 pm
Edit: Ah hell, I'm exhausted. I'll get it some time this next week.
Post by: Angle on May 27, 2014, 01:39:03 pm
Anyway, do you know a simple way to get back an arguments ID after it's been added?
Post by: Anvilfolk on May 27, 2014, 02:16:36 pm
If we were to add it, it would come from org.agora.server.database.DBAddArgument!
Ugh, I'm sorry I haven't been able to participate more. Grad school is destroying me.
Post by: Angle on May 27, 2014, 02:25:12 pm
Post by: Anvilfolk on May 27, 2014, 02:28:05 pm
Post by: Angle on May 27, 2014, 02:37:05 pm
Post by: Anvilfolk on May 27, 2014, 02:38:36 pm
Post by: Angle on May 27, 2014, 02:47:07 pm
Post by: Anvilfolk on May 27, 2014, 02:49:50 pm
How are you going to make the interface for it? Maybe a button on each argument that says "add new attacker" or something like that?
Post by: Angle on May 27, 2014, 02:55:16 pm
Post by: Angle on May 28, 2014, 12:14:42 am
Post by: Angle on May 30, 2014, 12:49:20 am
Post by: Anvilfolk on May 30, 2014, 09:20:57 am
Post by: Angle on May 30, 2014, 11:59:51 am
Post by: Anvilfolk on May 30, 2014, 02:41:16 pm
Are you doing one or two different queries? Could it be that it needs to be sent as a single query, i.e. query("INSERT ... VALUES (...); SELECT LAST_INSERT_ID();") rather than query("INSERT ... VALUES (...);"); query("SELECT LAST_INSERT_ID();"?
Post by: Angle on May 30, 2014, 03:17:18 pm
And yeah, you're probably right.
Post by: Angle on May 30, 2014, 06:27:38 pm
Post by: Angle on May 30, 2014, 11:48:58 pm
Post by: Anvilfolk on May 30, 2014, 11:49:55 pm
One thing we probably have to worry is whether there are concurrent inserts into the database. We wouldn't want a different ID to be returned. Could that happen in this implementation?
Post by: Angle on May 31, 2014, 12:29:42 am
Possibly. I think dbc has code to prevent that, but I'm not sure.
Post by: Angle on May 31, 2014, 10:22:19 pm
We should do something about getting some other people involved. Even if all they do is add issues to github, that's still pretty useful.
Post by: Angle on June 03, 2014, 02:12:30 pm
Post by: Anvilfolk on June 03, 2014, 02:19:29 pm
If not problems have been reported within a week we can go ahead and remove the JAgoraGraph repository.
Post by: Angle on June 03, 2014, 02:39:59 pm
Sweet!
If not problems have been reported within a week we can go ahead and remove the JAgoraGraph repository.
I tested it and found no problems. We just need a few other people to see of it works for them. On that note, I'm considering how to go about expanding our pool of testers. I was considering sponsoring a debate of some sort on the forum, just to test things out. I think we should probably neaten a few things up first, in particular Post needs to not have text run off the edge. This means either turning it into JComponent of some sort, which should be able to take care of that problem automatically, but may be hard to get to assume the right size and position, or writing the code for this myself. Which do you think would be easier to maintain in the long run?
Post by: Anvilfolk on June 03, 2014, 02:49:45 pm
If a JComponent is just "any" object that can draw itself, then that sounds like it wouldn't be too different than what people do in games, where you need to break text into lines depending on the width of each desired line, then draw each line, etc. Graphics are hard and annoying!
Post by: Angle on June 03, 2014, 02:53:48 pm
Post by: Anvilfolk on June 03, 2014, 02:54:44 pm
Post by: Angle on June 03, 2014, 02:57:41 pm
JComponent is the parent class of most everything in AWT and Swing. So the alternative is to use a different graphics library, or to have it as I do now, where posts are drawn on top of a JPanel with PaintComponent.
Post by: Anvilfolk on June 03, 2014, 04:10:34 pm
I actually tried using OpenGL for a while, but couldn't get transparency to work properly. It's also very low-level, probably more low-level than what we want to work with. Ultimately, I think a vector graphics library would be ideal, but I couldn't find anything that seemed promising for Java.
Post by: Angle on June 03, 2014, 04:53:04 pm
Post by: ShadowHammer on June 03, 2014, 09:55:38 pm
Post by: Angle on June 03, 2014, 10:10:28 pm
Edit: in other news, I fixed the issue with text running of the Post. Still need some work though, I'll go over it with FontMetrics tommorrow.
Post by: Anvilfolk on June 04, 2014, 09:59:22 am
Post by: Antsan on June 04, 2014, 11:53:28 am
One thing we ought to try to do is keep this implementation as multi-platform as possible. That means using libraries that are also multi-platform. Some game-dev libraries will not be, and I'm not sure what the whole OpenGL situation is.I don't know the situation for Java in this regard either, but as OpenGL is the low-level graphics library/interface/standard (I have no idea what the correct term is) it should be an obvious choice if you want multi-platform support.
Post by: Anvilfolk on June 04, 2014, 12:46:06 pm
I'd love it if we could work with something higher level, yet still multi-platform. It's unclear whether there is such a solution for Java at this point.
P.S. fixed the Eclipse files in all of the repositories, and fixed all the warnings. The new login screen is a lot better. At some point we should also add error handling. Perhaps I can try to tackle that.
Post by: Angle on June 04, 2014, 01:07:20 pm
Post by: Angle on June 04, 2014, 09:42:20 pm
I think we should probably consider moderation next. We'll also want to add in voting functionality, probably around the time we add more advanced display options.
Post by: Angle on June 05, 2014, 11:51:45 pm
Post by: Anvilfolk on June 06, 2014, 08:15:00 am
Editing, navigating, some tutorials/help, stuff like that should all be in. That's just my opinion of course! It's also fine if you want to go ahead and bring a few people in for feedback :)
Post by: Angle on June 06, 2014, 11:59:58 am
Post by: Angle on June 06, 2014, 06:33:14 pm
Post by: Anvilfolk on June 09, 2014, 09:41:48 am
Again, I'm just convinced that OpenGL is too low-level...
I've been looking at LibGDX again, but just getting the basic resources/skins system working is such a huge pain.... other than that it appears to have some rather nice UI features that could be interesting.
Post by: Angle on June 09, 2014, 08:59:22 pm
Post by: Anvilfolk on June 09, 2014, 09:08:55 pm
UPDATE Arguments SET content=? WHERE arg_id=?, user_id=?;
We also ought to start sanitizing input at some point!
Post by: Angle on June 09, 2014, 09:23:09 pm
Edit: alright, theres editing working. Do you think we should add a system to keep track of revisions?
Post by: Anvilfolk on June 10, 2014, 08:46:46 am
Argument revisions? No, not really. I think it'd be too much stuff. To be honest, even editing should be a feature that can be turned off at some point! We could actually do it like SMF here: we can store the information of when and who performed the last edit to a post. Shouldn't be horribly hard!
Post by: Angle on June 11, 2014, 03:49:52 pm
Post by: Anvilfolk on June 11, 2014, 03:57:17 pm
Post by: Angle on June 11, 2014, 08:59:04 pm
For now, what do you think we should work on next? I was considering taking a poke at switching to LWJGL, should I do that next or let it wait?
Post by: Anvilfolk on June 11, 2014, 09:18:35 pm
One of the things that's super important is that it looks, feels and works good. In particular, people should be able to "merge" different related discussion graphs, order posts in different ways, etc. This should be programmed in a very general way so that it is easy to add new visualizations. Good principles for game design probably apply here - things like detaching views from actual content, for instance.
The more I dive into LibGDX, the more great features it appears to have that we can use, but on the other hand it seems soooo annoying to actually get to work at first. It does offer really large portability, but at the cost of a relatively confusing and somewhat undocumented API and associated tools. Worst-case it devolves into OpenGL, which is what LWJGL is anyway. I feel very strongly that we should steer clear from very low-level libraries. It's a recipe for disaster and low productivity.
I say we could go for LibGDX instead of LWJGL (check the older revisions with the LWJGL implementation to see the kind of problems and low-level graphics programming we'd be running into). There's also Batik (http://xmlgraphics.apache.org/batik/), which looks interesting and might be closer to what we want.
If we decide on LibGDX, we should probably talk at some point, or I can write some of what I've learnt about it on here!
Post by: Angle on June 12, 2014, 12:56:17 pm
Post by: Anvilfolk on June 12, 2014, 12:57:15 pm
Post by: Angle on June 12, 2014, 01:46:40 pm
Post by: Anvilfolk on June 12, 2014, 01:48:37 pm
Post by: Angle on June 12, 2014, 07:05:15 pm
Post by: Angle on June 27, 2014, 11:54:34 pm
Post by: LordSlowpoke on June 28, 2014, 07:56:19 am
it could lead to a good amount of progress made in a short time period, as long as you know exactly what you want to improve and where
Post by: Anvilfolk on June 28, 2014, 09:43:44 am
I forgot to say something about user types! I honestly don't know at the moment. Diana was working on that, and on permissions and things. I don't think they they currently have a specific meaning, but I could be wrong. I'd guess these would be an integer telling you whether it's a regular user, moderator, administrator, etc.
Post by: Angle on June 28, 2014, 11:08:38 am
My main goal is to get it to the point where I could run a kickstarter or otherwise make it self sustaining.
Edit: Also, Here (http://www.mediafire.com/download/nu4d6adi51ja58t/Agora.zip)'s a couple precompiled jars. You'll need run the server first, then run the client.
Post by: Angle on June 29, 2014, 11:46:32 pm
Edit: Really, I think I should do this regardless, the only real question is whether I should have the advertisements.
Post by: Anvilfolk on June 30, 2014, 09:55:00 am
I still don't think the project is anywhere near a condition where it makes sense to monetize it (independently of my preference not to). Right now there are other alternatives that are free and have a much better interface. The only thing we could possibly offer right now is to have the debate outcomes be computed, which was one of the first implemented things but which needs to be updated.
Post by: Angle on June 30, 2014, 11:10:12 am
Edit: well, I did the research, and it looks like I can't afford to get it professionally developed, unless I spend more of my savings than I'm comfortable with.
Post by: Anvilfolk on June 30, 2014, 02:19:29 pm
Post by: Angle on June 30, 2014, 02:26:28 pm
Edit: Perhaps I should go back to considering running a kickstarter for Agora?
Post by: Anvilfolk on June 30, 2014, 03:50:25 pm
What I meant by technology was basically if we were OK using Java 2d APIs or whether we need something more. For now we should stick with Java, and then sure, eventually we can use another language, but we'll need to reimplement JAgoraLib, for instance.
Post by: Angle on June 30, 2014, 03:56:35 pm
Post by: Anvilfolk on June 30, 2014, 03:59:08 pm
Post by: Angle on June 30, 2014, 04:04:31 pm
Post by: Anvilfolk on June 30, 2014, 04:06:15 pm
Post by: Angle on June 30, 2014, 04:22:01 pm
Edit: Also this (http://trac.erichseifert.de/vectorgraphics2d/), which looks pretty good.
Post by: Angle on July 07, 2014, 05:03:00 pm
Post by: Anvilfolk on July 07, 2014, 05:04:20 pm
Post by: Angle on July 07, 2014, 05:16:16 pm
Post by: Angle on July 07, 2014, 06:45:33 pm
Post by: Anvilfolk on July 07, 2014, 09:21:16 pm
Post by: Anvilfolk on July 08, 2014, 04:04:21 pm
Post by: Angle on July 08, 2014, 04:12:45 pm
Post by: Anvilfolk on July 08, 2014, 04:14:26 pm
Post by: Benata on July 09, 2014, 04:00:34 am
HTML5 is future.
Post by: Anvilfolk on July 09, 2014, 07:52:50 am
Post by: Benata on July 09, 2014, 12:07:19 pm
Funny, I hate Javascript and am OK with Java ;)Weirdo. :D
Post by: Angle on July 10, 2014, 09:51:55 pm
Post by: Anvilfolk on July 10, 2014, 10:40:54 pm
Post by: dorf on July 11, 2014, 04:47:57 am
Each CMS is designed to do something specific really well and that's really easy to use, but if you want to do something custom, you will have a hard time with it. But it can be done!
I don't know how Agora works, but it seems to me that it could be implemented as a single-page app (GMail, Facebook...). Which means the choice of CMS is irrelevant (or even unnecessary in some cases) as the integration of Agora would be trivial.
Post by: Anvilfolk on July 11, 2014, 06:23:16 pm
With Wordpress (or any other CMS) you trade simplicity with flexibility.
Each CMS is designed to do something specific really well and that's really easy to use, but if you want to do something custom, you will have a hard time with it. But it can be done!
I don't know how Agora works, but it seems to me that it could be implemented as a single-page app (GMail, Facebook...). Which means the choice of CMS is irrelevant (or even unnecessary in some cases) as the integration of Agora would be trivial.
Thanks for the tip!
So I decided to dive into some SVG implementation and went with SVG Salamander. The implementation is under the weirdly named branch JFreeSVGClient of JAgoraClient, but it features arguments moving around, and they look nice. I feel like this is probably what we want to go with for the Java implementation.
Post by: Angle on July 11, 2014, 06:55:08 pm
Post by: Anvilfolk on July 11, 2014, 07:46:03 pm
Post by: Angle on July 11, 2014, 09:12:44 pm
Edit: I deleted those, and sure enough, it works now. I'll note, you didn't make you project exit when the window is closed. You need to add "this.setDefaultCloseOperation(EXIT_ON_CLOSE);" to the frame constructors.
Post by: Anvilfolk on July 11, 2014, 09:58:31 pm
What do you think of the looks of the SVG version? It's got anti-aliasing, very flexible (as flexible as SVG!), we can easily move arguments around, there's functionality for getting the argument at a given position (when we click on them), support for viewports/scrolling, I think there's some native support for animations too... I'm pretty excited about it. It's very dynamic, looks good, and there's basically no dependencies either.
I feel like it looks like the web version that we wanted without being javascript, which is a plus for me personally. Not saying the web interface isn't useful - it's probably even more portable - but I don't know a lot about javascript and we'd have to reimplement a whole bunch of stuff.
Post by: Angle on July 11, 2014, 10:08:36 pm
I went ahead and committed my changes, no sense duplicating that effort.
Post by: Angle on July 12, 2014, 11:47:18 pm
Post by: Angle on July 17, 2014, 04:03:20 pm
Post by: Anvilfolk on July 17, 2014, 04:04:21 pm
Post by: Angle on July 17, 2014, 06:32:38 pm
Also, remember when I was talking about using the thing for formal logic? Check this (http://rkirsling.github.io/modallogic/)out.
Post by: Anvilfolk on July 17, 2014, 09:57:26 pm
This looks kickass!
Post by: Angle on July 20, 2014, 04:19:27 pm
Post by: Anvilfolk on July 20, 2014, 04:21:18 pm
Post by: Anvilfolk on August 03, 2014, 06:14:42 pm
Post by: Angle on August 03, 2014, 06:29:31 pm
Edit: I made it close when the window is closed, and pushed to GIT. also, your debug messages look eerily similar to mine.
Post by: Anvilfolk on August 03, 2014, 07:13:20 pm
Post by: Angle on August 03, 2014, 07:50:48 pm
Post by: dorf on August 04, 2014, 03:59:16 am
However, the project has been inactive for 2+ years. It _is_ complete, but it does have smaller bugs and limitations.
If you'll ever hit the limitations, I recommend switching to artemis-odb (https://github.com/junkdog/artemis-odb), a forked version (and still active) of the original Artemis.
The intention of this post is purely informational. I'm not affiliated with either of them, nor do I think you should immediately switch to the forked version.
Post by: Anvilfolk on August 04, 2014, 09:09:30 am
Post by: Angle on August 24, 2014, 01:48:28 pm
Post by: Angle on September 03, 2014, 11:54:12 am
Post by: Anvilfolk on September 03, 2014, 12:24:07 pm
Post by: Angle on September 03, 2014, 12:35:42 pm
Ad now I've tested the thing, and yup, it works. That was laughably easy. Now I just need to make them understand each other.
Post by: Angle on September 05, 2014, 04:35:15 pm
Post by: Anvilfolk on September 08, 2014, 01:27:40 pm
My life has been a total mess recently, so there's little chance for me to work significantly on this for this semester, I think.
Post by: Angle on September 08, 2014, 02:56:13 pm
Post by: Anvilfolk on September 08, 2014, 02:59:04 pm
Or we can just check whether a message comes with authentication. If it does, we check it. If it doesn't we check whether server allows for this query to be "publicly available"?
Post by: Angle on September 08, 2014, 03:19:06 pm
Post by: Angle on September 11, 2014, 02:17:52 pm
Edit- wait, that's not it. I unsigned the jars, and they'll still run but they still won't find the BSON library, despite it being in the lib folder with JAgoraLib, which I could swear was finding it earlier.
Post by: Anvilfolk on September 11, 2014, 02:21:34 pm
Post by: Angle on September 11, 2014, 02:35:15 pm
Post by: Anvilfolk on September 11, 2014, 02:36:37 pm
Post by: Angle on September 11, 2014, 02:45:12 pm
Post by: Anvilfolk on September 11, 2014, 02:48:04 pm
Do people actually have to run Java on their computer then?
Post by: Angle on September 11, 2014, 02:56:21 pm
For the moment, I'm working on seeing if I can compile the BSON library myself and sign it. I found the source, and even loaded it into netbeans but it's not loaded as a java project. SO I need to figure out how that works now.
Post by: Anvilfolk on September 11, 2014, 03:01:05 pm
Post by: Angle on October 05, 2014, 07:04:21 pm
Post by: Anvilfolk on October 06, 2014, 10:25:17 am
Querying for one argument at a time is incredibly inefficient, which is why right now we are querying thread by thread. But nothing is stopping an attack from connecting two arguments from different threads. The threads are simply an aid to navigating Agora like you do in a forum!
We do still need to know how to handle "merging" graphs. What's happening right now is that attacks that connect the thread you asked for to another thread will have a "bogus" argument on one end, with just its ID and threadID. Using that threadID though, you can grab the thread. We just then need to merge the old thread we asked for first with the new thread.
Post by: Angle on October 06, 2014, 10:38:21 am
The main problem there is that threads can easily become huge - well over a hundred thousand posts huge. And in this system, they'll likely get even bigger. So grabbing an entire thread at once is also rather inefficient. Perhaps if we grab a post and all the posts in response to it?
I'll look into merging graphs. It doesn't sound too hard...
Post by: Angle on October 06, 2014, 07:21:40 pm
Spoiler (click to show/hide)
Post by: Anvilfolk on October 07, 2014, 09:27:08 am
Right now, we have basically a huge interconnected graph. It's just annotated with threads so that it can be navigated like a forum. I think it's a good idea for us to have more query types, but it's also important that we not go overboard with it so that we have to implement a bajillion functions in every AgoraLib implementation :)
Post by: Angle on October 07, 2014, 11:14:42 am
Post by: Anvilfolk on October 07, 2014, 01:17:19 pm
Post by: Angle on October 07, 2014, 04:24:15 pm
Post by: Angle on October 10, 2014, 05:38:56 pm
Post by: Anvilfolk on October 10, 2014, 06:32:15 pm
Post by: Angle on October 10, 2014, 06:49:39 pm
Post by: Angle on October 13, 2014, 09:28:31 pm
Post by: Anvilfolk on October 13, 2014, 09:29:52 pm
As for me, I've been working since 9am and it's 10.30pm :D This is also setting back my plans for Agora :P
*shakes fist at work*
Post by: Angle on October 20, 2014, 01:30:07 pm
Post by: Anvilfolk on October 20, 2014, 01:33:50 pm
Post by: Angle on October 20, 2014, 01:35:00 pm
Post by: Angle on November 03, 2014, 11:22:17 pm
Post by: Angle on November 06, 2014, 05:43:58 pm
permission java.net.SocketPermission "127.0.0.1:1597", "connect,resolve"; to your java.policy file, and will need to run JAgoraServer first. But after all that it should work. Once I have the client to a good point, I'll work on setting up a proper site, and should be able eliminate most all of the hassle.
Post by: Angle on November 10, 2014, 10:59:40 pm
Post by: Anvilfolk on November 10, 2014, 11:02:56 pm
Can't test anything right now, totally swamped and tired again. Sigh. Things will be better next year!
Post by: Angle on November 10, 2014, 11:14:03 pm
Post by: Anvilfolk on November 10, 2014, 11:14:43 pm
Post by: Angle on November 10, 2014, 11:16:37 pm
Post by: Anvilfolk on November 10, 2014, 11:17:43 pm
Post by: Angle on November 10, 2014, 11:30:34 pm
Post by: Fniff on November 11, 2014, 01:29:56 pm
Post by: Angle on November 11, 2014, 02:23:32 pm
Post by: Angle on November 12, 2014, 06:42:14 pm
Post by: Anvilfolk on November 12, 2014, 06:43:02 pm
Post by: Angle on November 12, 2014, 06:43:57 pm
Post by: Angle on November 22, 2014, 08:44:57 pm
Post by: Anvilfolk on November 22, 2014, 09:02:33 pm
Post by: Angle on November 22, 2014, 10:27:26 pm
Post by: Angle on November 26, 2014, 06:07:12 pm
Post by: Angle on December 06, 2014, 04:17:50 pm
Spoiler (click to show/hide)
Post by: Angle on December 10, 2014, 03:30:33 pm
Edit: And they're up again. was that your doing, or was the internet just being difficult?
Post by: Anvilfolk on December 11, 2014, 08:57:16 am
By the way, I've been talking to people who use MongoDB, and it seems that it might be a much better fit for Agora, since it stores graph-like structures. This might make it easier to perform complex graph queries.
Post by: dorf on December 11, 2014, 09:21:30 am
There is a myriad of NoSQL databases available now, though.
I've had an excellent experience using Blueprints (https://github.com/tinkerpop/blueprints) for my master's degree. It abstracts and decouples from the underlying NoSQL storages so that you can easily switch databases before you settle.
Of course they also expose the underlying storage instance in case you want to access to low-level or DB-specific features.
Note that you will lose normalization and the convenience of SQL (if that is important to you) when switching to NoSQL.
Post by: Angle on December 11, 2014, 07:20:03 pm
Post by: Anvilfolk on December 11, 2014, 07:31:04 pm
Post by: Angle on December 19, 2014, 03:31:53 pm
Post by: Angle on December 26, 2014, 06:03:36 pm
Post by: dorf on December 30, 2014, 07:48:30 am
What's even nicer is you going public so I can finally test this thing out! :)
Also keeping an eye out for JSAgoraClient.
Post by: Angle on December 30, 2014, 02:45:31 pm
I will also point out, you can test out the JSAgoraClient already, it's just a bit of a pain. I can guide you through it, if you like. Just be warned, there are like twenty steps.
Post by: dorf on December 31, 2014, 10:00:34 am
A part of me wants to contribute in a more meaningful way but I'm not sure I'm motivated nor skilled enough to see it through.
Post by: Angle on December 31, 2014, 02:30:00 pm
Post by: eerr on December 31, 2014, 06:44:38 pm
Post by: Angle on December 31, 2014, 08:34:39 pm
Post by: Angle on January 02, 2015, 07:00:11 pm
Post by: Angle on January 04, 2015, 01:08:57 pm
Post by: Arx on January 04, 2015, 02:03:33 pm
Post by: Angle on January 04, 2015, 02:47:49 pm
Spoiler (click to show/hide)
Then that means it's working. Next, you'll need to PM me your desired username and password so I can register you. I'd advise making up a completely new password, preferably random keyboard mashing that you save to a text file, because I don't think my database is very secure.
Post by: Angle on January 04, 2015, 06:22:37 pm
Post by: Angle on January 05, 2015, 01:33:52 am
Post by: Antsan on January 05, 2015, 12:49:25 pm
If you are fine with leaving that thing running for a month for maybe… I guess 8 to 24 hours of testing by me, put it up again.
Post by: Angle on January 05, 2015, 01:17:11 pm
Post by: Antsan on January 05, 2015, 02:44:56 pm
Also, the entires in there seem to be sorted into grant blocks and I am not sure which of those blocks the additional permission should go into.
Also also, I don't know what a "java web start" is (i found some information, but that was for Windows, I use Debian – is there even java web start for Linux?).
Post by: Angle on January 05, 2015, 03:00:01 pm
There is a java web start for Linux. For me, it's found under the internet section of the start menu. There are like five different java control panels though, so you may need to check several. I use Xubuntu, which is derived from Debian, so it should be more or less the same or you.
Post by: Antsan on January 05, 2015, 03:41:18 pm
There seems to be a user-level policy file at
Code: [Select]
/home/<user>/.config/icedtea-web/security/java.policybut I don't know whether it is actually in effect or not.Currently the site displays neither a graph nor a button saying "Refresh Graph". The rest is there.
[Edit]
After finding this thread (http://ubuntuforums.org/showthread.php?t=2256329&p=13185297#post13185297) I went to the "Extended Applet Security" tab and added a new row where the action is "Always trust this" and http://angle.webfactional.com/ as Document-base and Code-base.
That doesn't seem to work.
[/Edit]
Post by: Angle on January 05, 2015, 03:49:56 pm
Spoiler (click to show/hide)
Edit: The icedtea web control panel is not actually what you want, despite sounding like it's the right thing. As I said, there are like five different control panels. For me, the right thing is called Oracle Java 8 Web Start. Anyway, I need to go to work now. I'm leaving the server up, I'll be back in like 8 hours or so. Good luck!
Post by: Antsan on January 05, 2015, 04:00:24 pm
[Edit]
After finding this thread (http://ubuntuforums.org/showthread.php?t=2256329&p=13185297#post13185297) I went to the "Extended Applet Security" tab and added a new row where the action is "Always trust this" and http://angle.webfactional.com/ is Document-base and Code-base.
That doesn't seem to work.
[/Edit]
My IcedTea looks decidedly different from yours.
There is a security tab, but it is suspiciously devoid of anything but checkboxes.
There is no graph drawn. Switching between tabs works.
Post by: Angle on January 05, 2015, 07:19:32 pm
For me, the thing is called Oracle Java 8 Web Start. There are a another two or three java control panels that also won't do what you want, either.
Post by: Angle on January 06, 2015, 03:30:51 am
Post by: dorf on January 06, 2015, 08:43:05 am
If you dislike JS, you could use Unity (C#) however that requires a (surprisingly painless) add-on installation in the browser.
Post by: Angle on January 06, 2015, 01:55:00 pm
As an additional note, the server is still up, for anyone who still wants to test it.
Post by: dorf on January 06, 2015, 03:49:59 pm
Post by: Angle on January 07, 2015, 04:38:35 am
Post by: Antsan on January 07, 2015, 05:03:27 am
Sorry about that. Thank you for trying to walk me through setting it up.
Post by: Angle on January 07, 2015, 02:07:48 pm
Post by: Angle on January 11, 2015, 03:14:50 pm
Post by: Angle on January 13, 2015, 12:27:00 am
Post by: Antsan on January 13, 2015, 05:56:50 am
Post by: dorf on January 13, 2015, 06:30:10 am
But what should Angle do when a single thread (graph) is HUGE? Serve a HUGE image, or make the posts unreadably small?
Also, what happens to the interactivity of the application? Sending plain data to browser and letting the browser visualize and modify can offer much better user control. And uses less bandwidth.
If this is considered a prototype with intention of rewriting later, using the simplest of methods will probably be best.
Post by: Angle on January 13, 2015, 07:04:02 am
Post by: Anvilfolk on January 13, 2015, 11:43:17 am
I don't think the same is true if you're just running a Java application locally. There you can easily open a socket connection to somewhere else.
And yeah, dorf, the idea is definitely that you'll simply send the data, and the client will interpret it and show it to the user. Sending images would be very unoptimal for large graphs/debates!
Post by: Antsan on January 13, 2015, 01:19:03 pm
Post by: Angle on January 18, 2015, 12:03:48 pm
Post by: Angle on January 28, 2015, 10:38:09 pm
Post by: Angle on February 01, 2015, 01:17:36 pm
Post by: Anvilfolk on February 01, 2015, 01:20:43 pm
Post by: Angle on February 09, 2015, 10:46:29 pm
Post by: Anvilfolk on February 09, 2015, 10:57:46 pm
Post by: Angle on February 09, 2015, 11:37:50 pm
Post by: Anvilfolk on February 09, 2015, 11:42:54 pm
Post by: Angle on February 09, 2015, 11:55:11 pm
Post by: Angle on February 13, 2015, 05:57:01 pm
Post by: Angle on February 15, 2015, 04:17:12 pm
Post by: Angle on February 16, 2015, 12:19:57 am
Post by: Angle on February 17, 2015, 01:42:14 pm
Post by: Anvilfolk on February 17, 2015, 01:45:19 pm
I had to allow plugins and also accept the certificate. This happens a lot for Java, I think it's normal. I don't think there's a point in spending money to make that go away right now :)
Post by: Angle on February 17, 2015, 01:49:51 pm
Hmm, you didn't need to go and add "angle.webfactional.com" to your site exception list in order to get that option at all?
Post by: Anvilfolk on February 17, 2015, 01:51:03 pm
Post by: Angle on February 17, 2015, 01:53:16 pm
Post by: Anvilfolk on February 17, 2015, 01:54:33 pm
Can you try to use Chrome/Chromium and see if you get the option? It might be browser-dependent. If that doesn't change it, it might be OS dependent!
Post by: Angle on February 17, 2015, 02:01:55 pm
Well, regardless, what next? Should we consider slapping an Open source license on it? Are there any features that need to be implemented post haste? I'm thinking we should keep registration fairly limited for now, because our only moderating capabilities come from directly editing the database. Still, it'd be nice to start getting some testing done. We also probably need to refine the interface a bit - the posts are both really small and rather poorly positioned right now.
Post by: Anvilfolk on February 17, 2015, 02:15:11 pm
I think with all the changes, the part where the outcomes are computed isn't really working anymore. Database changes and all that. We'd need to have a way to do that periodically. I don't think it's super feasible to do it every time there's a change.
I'd really like to look into changing the database from an SQL one to a graph-based one, since that is so much more adapted to the kind of data we are storing. That will require quite a bit of behind the scenes changes, but hopefully it will be more efficient in the long-run. I don't really know if this is the case, but we should look into it.
Right now I'm totally pinned down by a conference deadline on Friday... but I've been working on a side project that is giving me a whole lot of experience with SVG, so hopefully I can get a nice, good looking app at some point!
Post by: Angle on February 17, 2015, 02:17:15 pm
Post by: Anvilfolk on February 17, 2015, 02:23:28 pm
I'm hoping this side project of mine will eventually allow me to get the Agora client working a lot faster. I'm learning a ton about lots of things like architecture, event-based systems, etc, all of which are applicable for the Agora client! This way we won't have to refactor it all the time as we learn new things!
Post by: Angle on February 17, 2015, 02:26:05 pm
I'm hoping this side project of mine will eventually allow me to get the Agora client working a lot faster. I'm learning a ton about lots of things like architecture, event-based systems, etc, all of which are applicable for the Agora client! This way we won't have to refactor it all the time as we learn new things!
Go for it! The client is pretty hacked together right now, and could use a major touch up.
Also, it occurs to me that it might be worthwhile to make a proper home page for Agora, with links to this page and all that. I'll go ahead and slap together some thing basic. Any suggestions for things I should put on it?
Post by: Anvilfolk on February 17, 2015, 02:28:11 pm
I think right now the only thing that's worth doing is sort of the mission statement. An online debating system with formal computed outcomes that make theoretical sense. Could be used for pure and simple debate, or for decision making. Provides more granularity than the competition. Is extremely flexible.
That kind of stuff!
Post by: dorf on February 18, 2015, 05:00:05 am
Is extremely flexible.(http://xaxor.com/wp-content/uploads/2012/10/Etremely-flexible-people17.jpg)
Go, team Agora!
Post by: Anvilfolk on February 18, 2015, 09:23:23 am
Post by: Angle on February 21, 2015, 12:41:49 am
Post by: Angle on February 23, 2015, 03:17:35 pm
Post by: Antsan on February 24, 2015, 02:41:31 am
I don't know about the user interface. As soon as other people start using the library they'll probably want to design their own UI either way, so it might be sensible to move that to the back.
Post by: Angle on February 24, 2015, 03:30:09 pm
Post by: Antsan on February 25, 2015, 04:40:27 am
Post by: Angle on February 25, 2015, 01:32:55 pm
Post by: Arx on February 25, 2015, 02:02:25 pm
Post by: Antsan on February 25, 2015, 02:17:54 pm
Well no, I'm talking about modifying the UI for usability. Right now the posts are super small, and if there's a lot of them they tend to cover each other up. You can see what I'm talking about right here (http://angle.webfactional.com/JAgoraHTTPServer/).That certainly seems like it might be a bit obstructive.
I see a menu on the left and a large blank area.
Post by: Angle on February 25, 2015, 02:23:02 pm
Post by: Antsan on February 25, 2015, 02:48:29 pm
Post by: dorf on February 26, 2015, 04:21:04 am
Agora currently locks up my Firefox until I kill it with task manager, which is probably a problem.It works here on Win 7 SP1 64-bit with Java 8U31 when using:
- FF 4, 10, 34.0.5, 35.0.1 and 36.
- Latest Chrome
- Opera 15.0.1147.130
Probably doesn't work on:
- IE 9 (it shows the alert window, but then the canvas is blank)
- Safari 5.1.4 (because I can't get it to detect Java for some reason)
- Opera 12.10 (it's having problems with Java Applets for some reason)
Post by: Angle on February 26, 2015, 06:50:11 am
Agora currently locks up my Firefox until I kill it with task manager, which is probably a problem.
Hmm. What version are you using?
Spoiler (click to show/hide)
...Huh. Do you have to add it to your site exception list, for java, or does it just give you the option to run it anyway?
Post by: dorf on February 26, 2015, 06:56:07 am
Post by: Arx on February 26, 2015, 09:19:47 am
Windows 7, 2GB RAM 1.5GHz Intel dual-core, if that helps.
Post by: Angle on February 26, 2015, 05:22:08 pm
Agora currently locks up my Firefox until I kill it with task manager, which is probably a problem.
Hmm. Have you tried clearing your cache? You might be running an old version, some of them have that problem. Ad of course, did you do the whole Site Exception List thing?
Post by: Angle on March 05, 2015, 05:33:40 pm
Post by: dorf on March 06, 2015, 02:54:22 am
So how many people have the applet working and would be interested in helping me test it if I enable registration?Tester reporting in.
Post by: Antsan on March 06, 2015, 06:08:21 pm
Would it maybe be useful to make a poll for this?
Post by: Angle on March 06, 2015, 06:22:56 pm
Post by: Angle on March 07, 2015, 04:56:41 pm
Post by: dorf on March 09, 2015, 04:00:01 am
I've made a few weird posts where I've described my intent, like connecting a post to every other node.
I've noticed that the graph doesn't update correctly after adding a multi-parent post. It doesn't show all the connected parents. I have to double-click the blue button to show that it seems.
Sometimes when I expand posts they appear at the top-left corner of the applet. They seem to snap into place when double-clicking (multiple times) the blue button.
I won't comment on general UI/UX at this stage of the project.
I've figured out the graph UI pretty quickly. The green button took a bit longer because of the bug(?) described above.
Post by: Angle on March 09, 2015, 07:30:28 am
Post by: Angle on March 16, 2015, 02:01:49 am
Post by: Cryxis, Prince of Doom on March 18, 2015, 05:05:16 pm
Post by: Angle on March 25, 2015, 06:50:47 pm
Post by: dorf on March 27, 2015, 09:09:04 am
Post by: Angle on March 27, 2015, 12:13:03 pm
Post by: Angle on April 03, 2015, 08:39:42 pm
Edit: More specifically, the GPL (http://choosealicense.com/licenses/gpl-2.0/) looks pretty good.
Post by: Angle on April 06, 2015, 01:25:16 pm
Post by: Angle on April 14, 2015, 02:46:12 pm
Post by: Urist McScoopbeard on April 14, 2015, 03:05:17 pm
Post by: Angle on April 16, 2015, 01:57:41 pm
Post by: Antsan on April 16, 2015, 03:13:14 pm
Post by: Angle on April 16, 2015, 03:18:44 pm
Post by: Skyrunner on April 18, 2015, 01:33:38 pm
Post by: Angle on April 18, 2015, 01:35:10 pm
Post by: Skyrunner on April 18, 2015, 01:41:12 pm
Post by: Radio Controlled on April 19, 2015, 05:38:38 am
Post by: Angle on April 19, 2015, 02:55:00 pm
Post by: Angle on April 19, 2015, 10:38:21 pm
http://mongodb.github.io/node-mongodb-native/api-bson-generated/bson.html#bson-deserialize
Post by: Angle on April 21, 2015, 01:20:44 pm
Post by: Eagleon on April 21, 2015, 03:11:50 pm
Post by: Angle on April 21, 2015, 03:13:29 pm
Post by: SirQuiamus on April 21, 2015, 03:26:45 pm
The server side code will still be in java, but the client should soon be entirely in Javascript.That's good news. I'm also a bit leery of JavaWS and Java in general, but I'd be willing to try the Javascript version.
Post by: Angle on April 21, 2015, 04:38:15 pm
For example, this latest attempt to get Agora working: It says use a buffer? What kind of buffer? hell if I know. The thing I'm using is a re-purposed node.js module though, so I figure it must be the node.js buffer. So I jump through a billion hoops to make that class work in my thing, installing the node package manager and several of it's libraries. I finally iron out al the bugs that causes and... It still doesn't work. Same error as before.
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAGHGHGAHGHAGHAGHAGHAGGHGA...
*sigh*. I'll wait a bit, try soe things out, and if none of them work, I'll try rewriting it to use the io.js buffer and see if THAT works.
Post by: Eagleon on April 21, 2015, 06:04:26 pm
It's funny, cause I like java, but it just doesn't look like it's gonna work out for us. So instead I'm stuck using bloody javascript, which I am developing a significant dislike for.I got that io.js link from this page https://github.com/feross/buffer (https://github.com/feross/buffer) and judging from
For example, this latest attempt to get Agora working: It says use a buffer? What kind of buffer? hell if I know. The thing I'm using is a re-purposed node.js module though, so I figure it must be the node.js buffer. So I jump through a billion hoops to make that class work in my thing, installing the node package manager and several of it's libraries. I finally iron out al the bugs that causes and... It still doesn't work. Same error as before.
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAGHGHGAHGHAGHAGHAGHAGGHGA...
*sigh*. I'll wait a bit, try soe things out, and if none of them work, I'll try rewriting it to use the io.js buffer and see if THAT works.
Quote
The goal is to provide an API that is 100% identical to node's Buffer API. Read the official docs for the full list of properties, instance methods, and class methods that are supported.the api appears to lift the method names from that class, to the extent that that is considered the official documentation, other than the two or three differences listed on the git page. Sorry if that wasn't clear.
Post by: Angle on April 21, 2015, 06:12:42 pm
Also: I just found out that the error I was getting was because I was directing the queries to the wrong address, and was thus just getting another copy of the html page in response.
AAAAAAAGAGGHAGAG...
Post by: Angle on April 21, 2015, 06:29:23 pm
Post by: Angle on April 22, 2015, 12:37:02 am
Post by: Angle on April 24, 2015, 06:04:41 pm
Spoiler (click to show/hide)
and here's the list supported by the buffer class that is used as input for BSON:
Spoiler (click to show/hide)
and no, utf16, the default value for xmlhttp, is different enough from utf16le that they don't work together.
Post by: Eagleon on April 25, 2015, 09:49:59 am
Post by: Angle on April 25, 2015, 04:22:24 pm
Post by: Angle on April 30, 2015, 09:30:23 pm
Post by: Angle on May 04, 2015, 04:25:22 pm
Post by: dorf on May 05, 2015, 01:33:59 am
Post by: BoredVirulence on May 05, 2015, 03:20:04 pm
I see that you started your design hoping to use java for the server side. I don't like javascript, but you pretty much have to use it on the client side, and if your using it there why not on the server as well? It could solve a lot of your issues with communication, a server running node.js and a web framework (I like express.js personally) communicate easily in JSON to the client side.
If I were building this I would probably do it using Node.js, Express.js (although thats really out of familiarity) and probably Durandal on the clientside. AngularJS has nice features, but can be really cumbersome, and I find Durandal to provide the same features in a cleaner way.
I'm somewhat interested, so I'll go through the google doc's page, and maybe stumble through the github a little. Although I'm more likely to fork the project using the frameworks I described than actually contribute. Especially Java, I hate Java (not the language, the environment forced on you, and also the security vulnerabilities).
Post by: Angle on May 05, 2015, 03:42:51 pm
Post by: BoredVirulence on May 06, 2015, 08:29:06 am
And of course, if I do fork it, you're welcome to take any of the resulting implementation. We'll see what happens.
Post by: Angle on May 06, 2015, 12:01:04 pm
Post by: Angle on May 09, 2015, 01:11:59 pm
Post by: BoredVirulence on May 09, 2015, 05:28:21 pm
Just a list of functions or partial URL's your client side needs. I could pump out a skeleton server with the proper routes, and a few examples on how to fill in the rest, in a few days.
I don't know what your using for the client side, I intended on forking it as a single project and would have re-written the client side too (I'm using it as an opportunity to learn Aurelia, it looks cool). But writing a quick server using express and node is super easy. Either way, most javascript frameworks speak JSON inherently. So my server would be passing and receiving JSON. And really, I see no reason why you would need BSON. It might save some bandwidth, but the complexity surely isn't worth it, especially not when JSON works out of the box.
Post by: Angle on May 09, 2015, 05:31:44 pm
I have access to the bash shell, so I should be able to install most anything.
Post by: BoredVirulence on May 09, 2015, 06:12:50 pm
Install node. Its especially easy in a UNIX like environment.
There will probably be a few utilities you need to install globally with npm (the package manager for node). Specifically a task runner will be helpful, I've worked with grunt and gulp. It will probably be grunt, because I'll probably steal most of the configuration from my last node + express project.
After that it should be as simple as cloning a git repo.
Do you want me to try and package a database with it? MongoDB pairs well, I'm sure there are others. NoSQL databases are nice for quick start up, and tend not to be bad on performance, but if performance is a concern I think Cassandra has some pretty good specs. I can probably plug a relational database in as well. I'll focus on the server first though.
Also, is this more of a web app, where the majority of the GUI is controlled by the client side in javascript? Or does the server spit out a number of different slightly dynamic pages? I imagine the "web app" approach would work better for this kind of site, and it keeps the server side very clean and simple.
Post by: Angle on May 09, 2015, 06:21:12 pm
And yeah, it's a web app. The GUI is entiely on the cient end, with the server only responsible for working with data.
Post by: BoredVirulence on May 11, 2015, 08:42:29 am
I can hand it over now if you would like. I plan on working on my fork, which would use the same server, but my next steps will diverge from what you've done already.
You should be able to take the server, add a MySQL driver to it, and add whatever routes you need for your client. Since your using your client side as a separate project I assume you aren't worried about having the server "serve" the client side app to the browser. If you are there are ways to do it, but it would take some time to set up.
The only other thing to add is some simple user authentication, but to add it you really need to hook up a database, and there should be an agreement on how it works with the client side. I have some code examples from an old project I could throw your way that demonstrate some authentication. Its been a while since I worked with that, and it worked out of the box so there was little that needed changing. A good example if you can follow it comes from the MEAN.js stack.
Edit: You can find the example server here:
https://github.com/BradyPowers/AgoraServer
I'll continue working on the fork. I'll be adding Cassandra as a NoSQL database, and I'll see about packaging a client side using Durandal. I'll probably put the actual fork in a separate repository, so if you need any help with the server I uploaded (its very minimal) we have a common repository. You have push access. Or you can just fork it and run with it.
Post by: Angle on May 11, 2015, 12:33:24 pm
I've got an incredibly simple, bare-bones server running express + node. It does nothing other than provide an example restful API. Its literally only 140 lines of code, and does virtually nothing.
I can hand it over now if you would like. I plan on working on my fork, which would use the same server, but my next steps will diverge from what you've done already.
I already have a server I can set the project up on, I don't need to use yours. I just want the code.
What are your next steps? I would assume you'd start adding the coding to handle requests, and all that.
You should be able to take the server, add a MySQL driver to it, and add whatever routes you need for your client. Since your using your client side as a separate project I assume you aren't worried about having the server "serve" the client side app to the browser. If you are there are ways to do it, but it would take some time to set up.
I already have both tomcat & htdocs on my server. I could just use one of those, right?
The only other thing to add is some simple user authentication, but to add it you really need to hook up a database, and there should be an agreement on how it works with the client side. I have some code examples from an old project I could throw your way that demonstrate some authentication. Its been a while since I worked with that, and it worked out of the box so there was little that needed changing. A good example if you can follow it comes from the MEAN.js stack.
That is something I'll be wanting eventually.
Edit: You can find the example server here:
https://github.com/BradyPowers/AgoraServer
I'll continue working on the fork. I'll be adding Cassandra as a NoSQL database, and I'll see about packaging a client side using Durandal. I'll probably put the actual fork in a separate repository, so if you need any help with the server I uploaded (its very minimal) we have a common repository. You have push access. Or you can just fork it and run with it.
Awesome, thank you!
Post by: BoredVirulence on May 11, 2015, 02:23:53 pm
I'm technically working on my own project. Its a fork, that is it does the same thing, but its an alternate implementation. So I will do things to a separate repository that won't be completely compatible. I'm doing it because I want a good web project to play with a few technologies I've been wanting to play with. Any work I complete related to this project can be used by you however you see fit. For instance, if my fork were to pass yours in functionality with less complexity, you have my express permission to incorporate it into the your project.
Since you are having such a problem with getting Java to communicate with your client side Javascript, I provided a simple Javascript server (back-end is synonymous here). When I say server, I mean the code that makes a machine behave like a server. The point of the Javascript server was to show you how you can accomplish those same tasks in Javascript. It is incompatible with your Java work though, commiting to it would require abandoning Java on the server. Technically you can get the two servers to communicate with eachother, but it would likely involve solving your current issues, and introduce a lot of unnecessary complexity. You may be able to use some Java technologies with the Javascript server, but thats likely to reintroduce issues getting Java and Javascript to work.
The server I provided can be used by you as you see fit, the repository too. If you aren't sure which direction to go with, feel free to develop them both side-by-side for a while. My fork will take place in a different repository.
Things you can do...
Install a driver for your database. Looks like there is one named mysql (type: npm install --save mysql). Here is a page on it. (https://www.npmjs.com/package/mysql)
This will allow you to connect to your MySQL database.
Since this server will be incompatible with your Java server, you will need to investigate to how to reimplement functionality you already have. Consider a little bit of effort here an investigation into which server you want to keep. I would also try and implement some of the routes your client side needs, and see if you can get it communicating.
Things I will be doing with my fork...
I will be adding a database (Cassandra in my case) to my fork. Next I will see if I can get Browserify (http://learnjs.io/blog/2013/12/22/express-and-browserify/) to serve (bundle, build, ship, whatever) a client side Durandal app to the browser (when a browser makes a request of the server, how else would the client side run?).
After that I will have a simple full stack application. From there I would implement most of the required functionality on the back-end. Test it using curl where applicable. Then implement the client side to provide an interface that utilizes the back-end restful API. At this point we have a partially functioning Agora implementation. Further iteration would refine it.
Post by: Angle on May 18, 2015, 08:27:18 pm
Post by: Eagleon on May 18, 2015, 09:49:17 pm
I have a guy suggesting that we rewrite the thing in Meteor.js with Polymer and Mongo-db. Does that sound like a god idea, that will make things easier in the long run?Is he offering to rewrite it? ;) Otherwise I would stick with an implementation and follow it through. That's a platitude, I don't really have the first clue about web development or JS - you're probably in the best position at this point to decide if Meteor or Polymer are good fits, or something you can work with.
Post by: Angle on May 18, 2015, 09:59:02 pm
Post by: Eagleon on May 18, 2015, 10:13:27 pm
He's offering to rewrite it for $650. I think it's probably worth it, as it includes the server and client, and should bring us pretty much back up to where we were with the java server.I can't contribute financially, things are nuts with work and they might not keep me on if I can't get the work restriction I have lifted. I can look at his portfolio if he has one if you want a second pair of eyes - tomorrow, I've had four hours of sleep in the past 16 haha. But yeah, keep options open?
Post by: Angle on May 18, 2015, 10:25:45 pm
Post by: dorf on May 19, 2015, 10:30:13 am
I hear they like to break their API from version to version and that old tutorials will not work in newer version.
meteor.js fits Agora well.
I've never heard of Polymer, so I don't know whether it belongs with Agora or not. Can you ask the dev in what way Polymer fits into the picture? Also make sure the dev knows whether you'll be rendering the forum in HTML5 canvas/canvas3d or not.
MongoDB is well supported in a JS environment and can be used as a graph DB to a degree. Neo4j is a fully-featured graph DB which comes with a lot of graph-related functions & algorithms. It's written in Java though. Ask dev why he thinks a document-based DB is more suited for a graph forum vs. a graph DB.
MongoDB might be OK for Agora, but that depends on the use-cases and the sizes of the graph.
TL;DR: I most probably wouldn't use such a tech configuration for Agora.
That said, I'm not your lead dev, and I don't have all the design facts before me.
Post by: Angle on May 19, 2015, 11:29:02 am
The reason he suggests Mongo-db is because that's what works best with Meteor. I'll talk to him about alternatives.
Post by: BoredVirulence on May 20, 2015, 01:06:42 pm
Meteor would do what my fork would do. But where I would use 2 frameworks (one for the client side, another for the server) Meteor is just one.
So yeah, it seems like a fine idea. It would bring pretty much the same benefits as my fork. Except it would be more consistent.
There is nothing wrong with MongoDB. NoSQL databases generally perform only slightly worse than properly normalized SQL database. MongoDB isn't the best for performance though, redis or Cassandra would be a faster choice. Both would probably work well with Meteor. Honestly, its not too difficult to swap in another NoSQL database if required, so I don't think that its an important design decision here, they all operate similarly.
NoSQL databases are nice because they don't require you to normalize them. You can nest data structures if necessary, and you can leave a document ID and join it later. They better represent the data than a series of tables that have to be manually assembled. I do think a NoSQL database better fits your graph structure than a SQL database.
Not familiar with Polymer. I'm not sure if its necessary. I guess Meteor might not provide enough DOM manipulation for Agora. Thats why I was using Durandal, to manipulate the DOM for to make it a single page app. I'm not familiar enough to say Polymer is good or bad.
TL;DR, Meteor should be fine. I can't speak for Polymer (Nor do I see why its necessary, but I'm guessing Meteor doesn't manipulate the DOM enough). MongoDB will be fine, especially initially. NoSQL databases are quite nice, I think the model fits your data better than a SQL database, and if you need more performance you should be able to swap a faster NoSQL database in without too much effort.
If the $650 isn't breaking the bank, go for it. It doesn't look like a bad decision. Just pay him in installments so you can see work is being done.
I'll still work on my fork, slowly (I have a lot of projects I like to work on).
Post by: Angle on July 09, 2015, 03:36:57 pm
Post by: Antsan on July 10, 2015, 03:29:38 am
When entering '…' (the Unicode character, [Alt Gr]+[.] on my keyboard) the mask for entering a new topic reloaded. I cannot reproduce this, so that might have been a fluke.
Post by: Jarvis on July 11, 2015, 12:36:44 am
Post by: Antsan on July 11, 2015, 05:14:48 am
Post by: Angle on July 11, 2015, 11:03:16 am
Post by: Jarvis on July 11, 2015, 12:50:03 pm
Post by: Angle on July 11, 2015, 01:55:30 pm
Post by: Jarvis on July 11, 2015, 03:39:57 pm
Post by: Antsan on July 11, 2015, 03:48:20 pm
Post by: Angle on July 11, 2015, 03:57:03 pm
Post by: Jarvis on July 11, 2015, 04:06:56 pm
Post by: Angle on July 11, 2015, 04:28:18 pm
Also, is anyone familiar with meteor? If so, would you be willing o take the time to look at the code for this (https://github.com/Agora-Project/agora-meteor) and point out any problems you see there? It's cool if you're too busy, I get it.
Post by: Angle on October 02, 2015, 06:15:17 pm
Post by: Antsan on October 05, 2015, 06:16:09 am
Post by: Angle on October 05, 2015, 02:54:07 pm
Post by: dorf on October 06, 2015, 06:41:12 am
Post by: Antsan on October 06, 2015, 06:48:55 am
Post by: Angle on October 06, 2015, 12:00:52 pm
Post by: Angle on August 15, 2016, 05:06:25 pm
Post by: Antsan on August 16, 2016, 04:23:35 am
I cannot read my overly long post, though. I still can delete other people's posts.
It probably would be sensible to replace the link to the forum in the first post.
Post by: Angle on August 18, 2016, 11:09:10 am
It probably would be sensible to replace the link to the forum in the first post.Done. I'm pretty sure that the example forum doesn't work right on chrome. I'm already working on a nwere version, though, so I'd rather focus on pushing that out. Hopefully it won't have the same problem.
In other news, does anyone know anything about Meteor, Atmosphere, NPM, D3, or Cytoscape? I'm currently trying to put the next version together and I need to figure out which of these technologies to use. Right now I'm trying to use meteor, atmosphere, and cytoscape, but one of the extensions to cytscape I want to use relies on NPM and isn't very friendly with meteor. I'm thinking I might just go back to using d3 instead of cytoscape, but that had it's own problems...
Post by: dorf on August 19, 2016, 07:39:41 am
but one of the extensions to cytscape I want to use relies on NPM and isn't very friendly with meteor.If the extension is simple enough: have you tried pasting the extension's code into your software somewhere, somehow?
Also, there seems to be an official way to install NPM packages into meteor (https://guide.meteor.com/using-npm-packages.html#installing-npm).
Post by: Angle on August 19, 2016, 03:41:01 pm
If the extension is simple enough: have you tried pasting the extension's code into your software somewhere, somehow?
Also, there seems to be an official way to install NPM packages into meteor (https://guide.meteor.com/using-npm-packages.html#installing-npm).
I considered doing that. The trouble with that is that it's pretty kludgey and will probably cause problems in the long term, and would also require a lot of work up front. And yeah, there are lots of ways to get NPM packages to work with meteor, but I'm not just using meteor, I'm using atmosphere, and atmosphere is not nearly so compatible. Right now, I have three main branches under developement. One uses meteor and D3, one uses meteor and cytoscpae, one uses meteor, atmosphere, and cytoscape, and is the furthest along, and one uses meteor, atomsphere, and cytoscape. I'm trying to decide which of these to continue development with, and how exactly to structure my code, and it's a difficult decision.
Post by: dorf on August 21, 2016, 03:55:19 pm
this looks interesting
Post by: Angle on August 22, 2016, 10:46:16 pm
In other news: http://en.arguman.org/Yeah, I've seen a few others like that. None of them seem to have really caught on though.
this looks interesting
In other news, I managed to overcome my dilemma and am making progress again! In a little bit I'll see about updating the demo, or just putting up another one entirely.
Post by: Angle on August 26, 2016, 03:03:26 pm
Post by: dorf on August 26, 2016, 04:02:16 pm
Post by: Angle on August 26, 2016, 04:05:57 pm
Oh, and what do you think? Is there anything in particular you think need to be done next? Maybe auto-sizing posts? Or making it so you can click the blue button and get a pop-up menu of posts, and load them individually?
Post by: crazyabe on August 26, 2016, 04:09:37 pm
Post by: Angle on August 26, 2016, 04:12:28 pm
PTW.
Did you check out the test site? It's mostly functional, if somewhat obtuse.
Post by: crazyabe on August 26, 2016, 04:21:33 pm
Post by: Angle on August 26, 2016, 04:22:31 pm
Post by: dorf on August 27, 2016, 05:47:28 am
Excellent, I can see your posts! Mind if I delete them to keep the forum tidy?Sorry for not elaborating, I thought you just wanted to know whether it works.
Oh, and what do you think? Is there anything in particular you think need to be done next? Maybe auto-sizing posts? Or making it so you can click the blue button and get a pop-up menu of posts, and load them individually?
Yeah, of course you can delete my ugly posts. I would have done it myself, but I didn't find the button that does that.
I think that at this early phase we can talk more about defects than functionality. If you think UI/UX improvements are not necessary at this phase, I'd suggest implementing more features and fixing bugs.
Personally, I'd like to see a functional prototype.
Currently it's difficult to navigate the posts. The graph is too wobbly. To the point that I couldn't click any buttons because it hasn't stabilized yet. Navigation needs to be really, really smooth before any kind of public release.
The mouse-wheel zoom is a nice feature and helps with readability!
Thread view might be nice. You press the blue button on a post somewhere in the graph and you get the whole conversation in a thread-like interface. That's what I imagined when you mentioned pop-ups anyway. This will get tricky when posts will have multiple parents though. S/he will have to annoyingly select the whole path.
Post by: Angle on August 27, 2016, 01:43:19 pm
Sorry for not elaborating, I thought you just wanted to know whether it works.
Yeah, of course you can delete my ugly posts. I would have done it myself, but I didn't find the button that does that.
I think that at this early phase we can talk more about defects than functionality. If you think UI/UX improvements are not necessary at this phase, I'd suggest implementing more features and fixing bugs.
Personally, I'd like to see a functional prototype.
Currently it's difficult to navigate the posts. The graph is too wobbly. To the point that I couldn't click any buttons because it hasn't stabilized yet. Navigation needs to be really, really smooth before any kind of public release.
The mouse-wheel zoom is a nice feature and helps with readability!
Thread view might be nice. You press the blue button on a post somewhere in the graph and you get the whole conversation in a thread-like interface. That's what I imagined when you mentioned pop-ups anyway. This will get tricky when posts will have multiple parents though. S/he will have to annoyingly select the whole path.
Ah, good to know! I'll work on getting the graph less wobbly, and start implementing the thing for loading posts individually. I am considering how best to implement a kind of "conversation follower". I think first we'll need to add types to replies, so that your reply can be an attack, or a support, or reference another post, etc, etc, and depending on that we can track the flow of a conversation. Combined up and down voting and SAA, we can track which posts are more or less important, and have the conversation tracking ignore all below a certain threshold. That's a little ways off, though.
Post by: dorf on August 27, 2016, 03:48:52 pm
I'd go for no wobbling myself. Make it as static as possible.
I had a student show me his work: some interactive JS graph-like visualization thingy. Whenever I clicked/changed something the whole graph spun off the screen completely and then spun back into view wobbling around until it finally stabilized. I really hated that. I think other users will too.
Once you'll have the reply types figured out the conversation follower feature should fall into place nicely. The "meat" of the conversation will probably have a recognizable pattern (OP post -> attack -> attack -> attack, with various supporting replies mixed in). Such conversations can probably be easily visualized. Or at least easier than graph visualization!
Post by: Angle on August 31, 2016, 01:53:32 pm
Heh, I had a few more ideas I could have mentioned. I had to deliberately stop myself to write them out since, like you've said, some dreamy features (https://github.com/Agora-Project/AgoraDevDocs/wiki/Outcome-Computation) are a little ways off :)
I'd go for no wobbling myself. Make it as static as possible.
I had a student show me his work: some interactive JS graph-like visualization thingy. Whenever I clicked/changed something the whole graph spun off the screen completely and then spun back into view wobbling around until it finally stabilized. I really hated that. I think other users will too.
Once you'll have the reply types figured out the conversation follower feature should fall into place nicely. The "meat" of the conversation will probably have a recognizable pattern (OP post -> attack -> attack -> attack, with various supporting replies mixed in). Such conversations can probably be easily visualized. Or at least easier than graph visualization!
Well, I managed to cut down on the wobbling significantly, though not completely, and make it so that you can right click on a post to see what all the other posts connected to it but not loaded in the graph are, and click them to load them. I'm also considering a significant number of issues towards a UI rework.
*allowing people to pin posts
*making posts prettier
*Making it so that you can click anywhere on a post to select it.
*Making it so that you can make new links by clicking and dragging.
*making it so that when you want to add a new node to the graph, it just adds a post like when you load an existing one, except that it's text fields are editable. You can then select any number of posts and connections that haven't been added yet, and submit them by clicking the submit button.
After that, we have a couple more functionality oriented changes
*Making more types of connections between attacks, so that you can have attack, supports,
*SAA! Yes, voting and solving for relative value of different posts!
Any idea what all should be prioritized, or anything you'd like to add?
Post by: Angle on September 03, 2016, 06:10:43 pm
Post by: Angle on September 10, 2016, 01:24:21 pm
Post by: Eagleon on September 10, 2016, 03:29:18 pm
More feedback! Sorry that I still can't contribute more time to this, I've got my first real commercial music project going =3 pretty exciting
- Everything should, visually, have its own layer not necessarily using the same cue, just different. Most of this is about that.
- Dragging posts is still awkward because there's no visual cue that a post is being dragged. A small icon is all you'd need if there's no way to get a transparency of the post itself. Also nice would be to draw the connections as you drag them, because where it ends up isn't always too predictable, so you have to search for a second or two after it's dropped.
- The post title should be above or next to the red dot, maybe right-justified into it, or worst-case limited in length. As it is, some of the text is hidden if the title is too long. The dot also doesn't need to be nearly so large if you keep it at the same scale while zooming. Might look a little strange at first, but it'd be worth it for being able to focus on the posts themselves more.
- While zoomed out, we probably don't want to read the post text itself so much as the titles. Scaling them differently might work?
- It should be easier to visually transition from title to post, while still making the title most prominent - maybe a line/divider in bracket shape below the title?
- The background should, IMO, be darker than the posts to make them pop out more, or else the border of the posts should be a bit stronger. If you relied on the border, you could potentially get rid of the right side and make it look like it's being peeled up from the page.
- Having a seperate page for composing a post feels a little jarring. Many times I'll reference what's been written in previous posts in a conventional BBS, and this would make that impossible without opening another window. But maybe a pop-up is what you need, since you still want to be able to move stuff around to read through the network. I dunno.
Post by: Angle on September 10, 2016, 05:02:12 pm
I didn't mind the floatiness once the posts froze when you interacted with them, I think this form will break more once you get into lots of cycles with attacks, but that's less important than getting this polished so that we can have fun with it while it's developed :) It's usable in its current state either way IMO.
A lot of people were complaining about posts moving, hence the change. It also lets me ditch the obnoxious code for freezing posts.
More feedback! Sorry that I still can't contribute more time to this, I've got my first real commercial music project going =3 pretty exciting
Hey congrats! I'll look it up.
Everything should, visually, have its own layer not necessarily using the same cue, just different. Most of this is about that.
What all do you mean? Links are already displayed behind posts. Do you mean like in terms of contrast and what stands out and draws the eye?
Dragging posts is still awkward because there's no visual cue that a post is being dragged. A small icon is all you'd need if there's no way to get a transparency of the post itself. Also nice would be to draw the connections as you drag them, because where it ends up isn't always too predictable, so you have to search for a second or two after it's dropped.
I actually just fiddled with that. It's still a little finicky, but it's a fair bit better now than it was.
- The post title should be above or next to the red dot, maybe right-justified into it, or worst-case limited in length. As it is, some of the text is hidden if the title is too long. The dot also doesn't need to be nearly so large if you keep it at the same scale while zooming. Might look a little strange at first, but it'd be worth it for being able to focus on the posts themselves more.
- While zoomed out, we probably don't want to read the post text itself so much as the titles. Scaling them differently might work?
- It should be easier to visually transition from title to post, while still making the title most prominent - maybe a line/divider in bracket shape below the title?
- The background should, IMO, be darker than the posts to make them pop out more, or else the border of the posts should be a bit stronger. If you relied on the border, you could potentially get rid of the right side and make it look like it's being peeled up from the page.
Yeah, rearranging the post displays will probably be my next task. I need to do a bunch of cleaning up on them. Give the posts proper windows, make sure nothing overlaps, rework the colors, etc, etc.
Having a seperate page for composing a post feels a little jarring. Many times I'll reference what's been written in previous posts in a conventional BBS, and this would make that impossible without opening another window. But maybe a pop-up is what you need, since you still want to be able to move stuff around to read through the network. I dunno.
I actually already have a plan for that - I'm gonna change the button so it just adds a blacnk post to the graph, and you can click on it and type in it and make links between it and the other nodes, and even add multiple of them and then submit them all at once.
Post by: Eagleon on September 10, 2016, 05:15:21 pm
The music will be in a CYOA style novel, who knows when the game will be released, I'm just happy I'm starting to get to that point haha. My insane paranoia is that it'll all be "but the music sucks! D+ try again" when it gets to greenlight stage x3 Just have to win a grammy, and then I can stop hating on myself.
Post by: dorf on September 11, 2016, 04:27:47 am
I switched the forum to use a static layout. It should eliminate the problems people have been having clicking on posts and such. What do you think, people? Is it an improvement?Yeah, I like it!
I've had a bug where I've opened connecting posts of the 'Forum Games' and they've appeared parentless at the bottom-right side of the graph. Unfortunately I cannot reproduce the bug.
Tooltips are helpful, although jarring. You might wanna try using the browser built-in tooltip by manipulating the title attribute of #posts-graph.
My insane paranoia is that it'll all be "but the music sucks! D+ try again"It's hard putting yourself out there to be potentially criticized but what's the alternative? Being creative and not showing your work to anyone?
Post by: Angle on September 12, 2016, 04:38:03 pm
Oh, and what all would be necessary for you to start using it now? I kinda want to start testing it out ASAP so I can get a good feeling for how it works and how it could be made better.
Post by: dorf on September 13, 2016, 06:15:08 am
- category: the top posts (Forum, Creative Forums...)
- thread: user-created post (OP) that is directly under a category
- The main categories should probably always display in the same order. If you open the application a bunch of times and click the "load connected posts" button, the order of categories will be different almost every time. Since you allow users to manipulate the graph ad-hoc, maybe have a "remember last state" option?
- Make threads and categories visually different. Make user-created threads and mod-created categories visually different.
- Loading icons/animations will keep users from leaving the site.
- I feel the starting view should be more zoomed-in maybe. The font is quite small by default. The buttons are too tiny for mobile-use and uncomfortable for mouse-use.
- The current layout makes me want to create twitter-like posts instead of large posts such as this one.
- Disabled buttons when they are not applicable. When a post doesn't have children the button should be grey'd-out.
- Maybe add some numbers. E.g. posts could have total number of unread/hidden children posts.
- Streamline! Make the most frequently used actions as fluid and simple as possible for the user. Use more recognizable icons for creating posts and loading connecting posts.
- Feature-wise I'm unsure. I guess that's dictated by Agora's use-cases. So, if you're going for an social argument platform, implement features that will go toward that goal.
Post by: Angle on September 13, 2016, 07:10:17 pm
What do the rest of you think? Sound good, or would you suggest something else?
Post by: Angle on September 16, 2016, 05:21:54 pm
Post by: Angle on October 02, 2016, 03:53:01 pm
I ask cause me and a friend are going over the code and he wants to make some major changes. I agree that the project needs some major changes, but I'm not sure that these are the right ones.
Post by: Angle on October 14, 2016, 06:00:09 pm
Spoiler (click to show/hide)
Post by: dorf on October 16, 2016, 12:49:20 pm
Nice work.
Post by: Angle on October 16, 2016, 02:12:44 pm
Post by: Angle on November 10, 2016, 05:46:39 am
You will likely need to refresh your browser after navigating to the forum page in order to get the forum to work. We're gonna try and catch that mod soon, but for the moment it persists.
Anyway, what do you think? Look good? Right now you can load posts by clicking the 'Show Connecting posts' button in the lower left of a post, you can close it by clicking the red X, you can drag it around by clicking on the titlebar (But not the title), you can drag the entire forum around by clicking on the back ground and dragging, and you can add new replies by click the reply buttons. Clickint\g reply to multiple posts will connect your reply to those multiple posts. Replies are fairly self-explanatory - they have a title and a body, and you can write whatever you like in them and then click submit to add them to the graph.
You can view it here: http://forum-demo-88817.onmodulus.net/forum
Post by: Antsan on November 10, 2016, 09:01:09 am
Post by: MoonyTheHuman on November 10, 2016, 09:10:30 am
Post by: Angle on November 10, 2016, 12:36:50 pm
The number in parens behind the "Show connecting posts" is confusing. I thought it would be the number of subnodes, but as it is it seems to be the height of the tree the button belongs to. I'm not sure what purpose that measure has – I think number of subnodes is more relevant for the function of that button, since it predicts what strain the user will put on his system by clicking on that node, abstractly speaking.
That shows the number of subnodes and of supernodes. I'm eventually gonna make it show that number, minus the ones that already present on the graph.
Intresting. i would like to help ^_^ what languages are used? (i know js and html, but what else?)
CSS and a thing called Meteor (https://www.meteor.com/), which isn't a language per se, but requires learning a lot like one. You can view the source for the project here (https://github.com/Agora-Project/agora-meteor) (forum code) and here (https://github.com/Agora-Project/agora-meteor-demo) (Test website).
If you need any help figuring things out don't hesitate to ask!
Post by: dorf on November 11, 2016, 05:24:09 am
Their right border moved like 1px in then out, etc.
Post by: Angle on November 12, 2016, 04:44:50 pm
Once I've opened the whole tree, two posts in the tree were shaking their right border while I was moving the mouse.
Their right border moved like 1px in then out, etc.
I'm not seeing that bug, but I'll keep an eye out for it. It's probably a schrodinger bug, it'll show up eventually.
On another note, does anyone have any advice for what order I should start implementing other things? I have a lot of features left to implement and I'm not sure how to prioritize them. These include allowing posts to be expanded so you can read a large post easily, zooming, proper user profiles, SAA, replying to posts and web pages outside the forum, moderation, BBCode, reworking the graph layout, etc.
What all do you guys think should be prioritized, and in what order?
Post by: dorf on November 13, 2016, 02:22:28 pm
I'd wager that SAA is a core feature to Agora.
Add BBCode and proper user profiles last.
Post by: Angle on November 13, 2016, 03:29:29 pm
Add spam detection and user registration (with OpenID support) somewhere in there.We already have user registration, and adding OpenID support should be pretty easy - there are packages for logging in with facebook, google, etc. I can probably just drag and drop them right in.
I'd wager that SAA is a core feature to Agora.
Add BBCode and proper user profiles last.
Huh, really? I figured I should prioritize having enough features that people can use it normally, so I can build up a user base before going for the high end stuff like SAA. :/
Post by: dorf on November 17, 2016, 08:32:42 am
Since I thought Twitter and mobile games were merely fads, you can assume that my opinion doesn't count for much :)
Post by: Angle on November 25, 2016, 05:15:32 pm
It's just my opinion/interests.
Since I thought Twitter and mobile games were merely fads, you can assume that my opinion doesn't count for much :)
No, that's fair. I was just surprised.
In other news, I'm considering changing the view to implement a sort of 'broad overview' which would allow people to view the forum as a bunch of dots connected by smaller lines, like this:
For the interface, what about having a panel which shows the posts as a series of nodes linked to others, and they can be selected to bring up a windown showing the content of that post? something (vaguely) like this:
(http://img.ie/wmum1.gif)
ANy thoughts? Does it seem worthwhile? Are there other changes you'd rather see?
Post by: Angle on December 09, 2016, 05:20:47 pm
Especially if you have advice on what to add next - I'm thinking of tackling BBCode, but I'm not certain. I have a friend who's working on adding a way to display the entire forum, or at least a large chunk of it as a graph, but that's still being implemented. You can see a preview at the bottom of the forum page, though.
Post by: Angle on December 31, 2016, 12:10:44 am
Post by: Angle on January 04, 2017, 06:06:06 pm
Post by: Angle on January 28, 2017, 07:06:30 pm
Post by: Urist McScoopbeard on January 28, 2017, 11:33:09 pm
Post by: Angle on January 28, 2017, 11:42:22 pm
Just dropping by to say this is a great project.
Thank you! Any suggestions for things that would make you want to use the forum? Right now I'm mostly trying to get things to the point where people will use it, so that I can see how it works in practice.
Post by: Angle on January 30, 2017, 11:56:00 pm
Post by: Angle on January 31, 2017, 11:07:31 pm
Post by: Retropunch on February 05, 2017, 03:44:19 pm
Basically, I think it just needs some big changes to how you show the basic stuff. My suggestions would be as follows:
- Highlight the current node you're looking at. Make it a green dot or similar. (this is an absolute MUST)
- Show the alternate nodes you could have jumped to from the last node (I'd suggest that they be fainter/slightly transparent). It's needlessly time consuming to have to jump back to see alternate answers to the same post.
- Label the nodes with their subject. You need to know roughly what you're looking at straight away - do the first three words or something for subjects. This might spread stuff out a bit and make it looks slightly messier, but the whole point of this is that I can see more naturally how conversations are structured - I lose that if I have to hover over everything.
Whilst it's a big change, I'd also suggest having text on the left and the node-tree on the right. Currently you've got small posts, but if you end up with lengthy ones (like this!) you'll lose the one thing that makes Agora stand out (the fact you can see where your entry fits in the tree of stuff) by pushing it all the way down to the bottom underneath the text. Having the text in a scrollable box to the left hand side will help a lot.
Honestly, I love love LOVE the idea, but it's got to be more (or at least equally) efficient than a normal forum for it to be useful. Sorry this has been such a critical first post - I think it's a great and I definitely want to see it come to fruition. I honestly believe it could become the new forum standard if it was made easy enough to use - I'd certainly deploy it across my sites if it did!
Post by: Angle on February 06, 2017, 04:21:33 am
So, I've just looked at this and I love the idea behind it. However, I've got to be brutally honest and say that I think through your quest for the whole 'node based discussion' you've gone too far into making it more difficult to use than normal forums. I've done a lot of systems documentation in the past, and this suffers from the classic problem of a lack of basic, quickly readable 'surface' information. You're forced immediately to do a deep dive into stuff, and rather than it being better than a normal forum, it becomes way more time consuming to use and just plain confusing.
Basically, I think it just needs some big changes to how you show the basic stuff. My suggestions would be as follows:
- Highlight the current node you're looking at. Make it a green dot or similar. (this is an absolute MUST)
- Show the alternate nodes you could have jumped to from the last node (I'd suggest that they be fainter/slightly transparent). It's needlessly time consuming to have to jump back to see alternate answers to the same post.
- Label the nodes with their subject. You need to know roughly what you're looking at straight away - do the first three words or something for subjects. This might spread stuff out a bit and make it looks slightly messier, but the whole point of this is that I can see more naturally how conversations are structured - I lose that if I have to hover over everything.
Whilst it's a big change, I'd also suggest having text on the left and the node-tree on the right. Currently you've got small posts, but if you end up with lengthy ones (like this!) you'll lose the one thing that makes Agora stand out (the fact you can see where your entry fits in the tree of stuff) by pushing it all the way down to the bottom underneath the text. Having the text in a scrollable box to the left hand side will help a lot.
Honestly, I love love LOVE the idea, but it's got to be more (or at least equally) efficient than a normal forum for it to be useful. Sorry this has been such a critical first post - I think it's a great and I definitely want to see it come to fruition. I honestly believe it could become the new forum standard if it was made easy enough to use - I'd certainly deploy it across my sites if it did!
No no, thank you! I would love to relieve more in depth straight forward criticism like that. It's pretty ate for me right now, so I'll go over your post in detail in the morning.
Post by: Angle on February 06, 2017, 06:32:52 pm
So, I've just looked at this and I love the idea behind it. However, I've got to be brutally honest and say that I think through your quest for the whole 'node based discussion' you've gone too far into making it more difficult to use than normal forums. I've done a lot of systems documentation in the past, and this suffers from the classic problem of a lack of basic, quickly readable 'surface' information. You're forced immediately to do a deep dive into stuff, and rather than it being better than a normal forum, it becomes way more time consuming to use and just plain confusing.
Basically, I think it just needs some big changes to how you show the basic stuff. My suggestions would be as follows:
- Highlight the current node you're looking at. Make it a green dot or similar. (this is an absolute MUST)
- Show the alternate nodes you could have jumped to from the last node (I'd suggest that they be fainter/slightly transparent). It's needlessly time consuming to have to jump back to see alternate answers to the same post.
- Label the nodes with their subject. You need to know roughly what you're looking at straight away - do the first three words or something for subjects. This might spread stuff out a bit and make it looks slightly messier, but the whole point of this is that I can see more naturally how conversations are structured - I lose that if I have to hover over everything.
Whilst it's a big change, I'd also suggest having text on the left and the node-tree on the right. Currently you've got small posts, but if you end up with lengthy ones (like this!) you'll lose the one thing that makes Agora stand out (the fact you can see where your entry fits in the tree of stuff) by pushing it all the way down to the bottom underneath the text. Having the text in a scrollable box to the left hand side will help a lot.
Honestly, I love love LOVE the idea, but it's got to be more (or at least equally) efficient than a normal forum for it to be useful. Sorry this has been such a critical first post - I think it's a great and I definitely want to see it come to fruition. I honestly believe it could become the new forum standard if it was made easy enough to use - I'd certainly deploy it across my sites if it did!
These changes are for the expanded view posts, yes? The ones showing the full text of a post, with the graph down at the bottom showing nodes above and below? What about the main forum view? I just changed things so it should show up as the default front page now.
As for your suggestions, I've been thinking of doing the first one. I'll go ahead and bump it up and do it today. The second is interesting, and I'll definitely think about it. The third one I'm skeptical about, but I've heard it before, so I'll definitely consider it. As for moving the text and the node tree to be side by side, I could definitely do that. I'm worried how well it'll work out on smaller displays, though.
Post by: Mesa on February 06, 2017, 06:56:02 pm
Why not go with something like Markdown instead of BBCode?
I know BBCode is the 'traditional' markup used for forums, but it feels really needlessly verbose for typing manually - this might be just me but I generally dislike using rich text editors, or at least having to fall back onto buttons a lot when I 'm doing a lot of formatting.
Not to mention that BBCode is absolutely heinous on any device that does not have an access to a proper keyboard - writing B12 posts on my phone is the worst thing sometimes.
I will admit that I don't know how difficult it is to implement either Markdown or BBCode and whether or not you are willing to throw one out in favor of the other, but I really do think that despite being, well, a fair bit more powerful than Markdown, most people do not need that much power and they are especially not fond of the added syntactic verbosity.
Especially with how the rest of the interface seems to be designed in the 'new-age' flat/material/responsive-design sorta way, pairing that with BBCode just strikes me as strange.
Just my 2 cents though. I've been extremely extremely casually watching this thread for a while and I like how much it's progressed so far, but I really feel that BBCode is...not outdated, but not really fit for The Modern Age™.
Post by: Retropunch on February 06, 2017, 07:31:58 pm
These changes are for the expanded view posts, yes? The ones showing the full text of a post, with the graph down at the bottom showing nodes above and below? What about the main forum view? I just changed things so it should show up as the default front page now.
As for your suggestions, I've been thinking of doing the first one. I'll go ahead and bump it up and do it today. The second is interesting, and I'll definitely think about it. The third one I'm skeptical about, but I've heard it before, so I'll definitely consider it. As for moving the text and the node tree to be side by side, I could definitely do that. I'm worried how well it'll work out on smaller displays, though.
Glad you can take the criticism on board, and I'm thrilled you're taking on board some of my suggestions! I honestly think it's a fantastic idea.
I am referring to the expanded posts in my suggestions. In terms of the main view, I honestly don't know if it works well; mostly due to a lack of labelling.
I really do understand that you're trying to sort of make it more 'natural', but without seeing labels I just can't tell where to even start. It's like coming onto Bay12 for the first time, and instead of seeing the different sub-board headers (Other Games, Lets Plays etc) you just saw blank spaces and had to hover over them to see what they were - would that be more user efficient or less?
If you're really convinced that some people may prefer it blank, then put it as a toggle option, but I really can't imagine how people could try to navigate through a forum without labels. Again, sorry my criticism is so strong on that, but it's something I feel is a pretty strong blocker to mainstream enjoyment.
On top of the labelling issue (and something that would help it not look so cluttered), I'd really suggest that you fold up the main view when you enter the page, and allow users to click through it to open it up. As in, it'd just show the first and second row - If you clicked on one of the second row nodes, it'd open everything attached to that node (and so on and so on) - a double click could then open the post itself. A button/whatever to show everything would be good though. Most of this is because on the vast majority of forums, there's tons of boards people aren't interested in. More than that though, if this shows every post on a 30,000+ post site, it's just going to be a complete mess.
In terms of the side by side view - I'd suggest keeping it vertical for mobile devices, but would definitely think it's a big improvement as side by side for non-mobile devices.
Post by: Angle on February 06, 2017, 09:11:21 pm
This is about as much as I can offer in terms of feedback right now but...
Why not go with something like Markdown instead of BBCode?
I know BBCode is the 'traditional' markup used for forums, but it feels really needlessly verbose for typing manually - this might be just me but I generally dislike using rich text editors, or at least having to fall back onto buttons a lot when I 'm doing a lot of formatting.
Not to mention that BBCode is absolutely heinous on any device that does not have an access to a proper keyboard - writing B12 posts on my phone is the worst thing sometimes.
I will admit that I don't know how difficult it is to implement either Markdown or BBCode and whether or not you are willing to throw one out in favor of the other, but I really do think that despite being, well, a fair bit more powerful than Markdown, most people do not need that much power and they are especially not fond of the added syntactic verbosity.
Especially with how the rest of the interface seems to be designed in the 'new-age' flat/material/responsive-design sorta way, pairing that with BBCode just strikes me as strange.
Just my 2 cents though. I've been extremely extremely casually watching this thread for a while and I like how much it's progressed so far, but I really feel that BBCode is...not outdated, but not really fit for The Modern Age™.
I didn't even know markup existed. :/
I'll definitely consider it - I'm hoping to have proper editors, so that people don't need to deal directly with the BBCode so much, but failing that I'll definitely consider using markup instead.
Glad you can take the criticism on board, and I'm thrilled you're taking on board some of my suggestions! I honestly think it's a fantastic idea.
I am referring to the expanded posts in my suggestions. In terms of the main view, I honestly don't know if it works well; mostly due to a lack of labelling.
I really do understand that you're trying to sort of make it more 'natural', but without seeing labels I just can't tell where to even start. It's like coming onto Bay12 for the first time, and instead of seeing the different sub-board headers (Other Games, Lets Plays etc) you just saw blank spaces and had to hover over them to see what they were - would that be more user efficient or less?
If you're really convinced that some people may prefer it blank, then put it as a toggle option, but I really can't imagine how people could try to navigate through a forum without labels. Again, sorry my criticism is so strong on that, but it's something I feel is a pretty strong blocker to mainstream enjoyment.
On top of the labelling issue (and something that would help it not look so cluttered), I'd really suggest that you fold up the main view when you enter the page, and allow users to click through it to open it up. As in, it'd just show the first and second row - If you clicked on one of the second row nodes, it'd open everything attached to that node (and so on and so on) - a double click could then open the post itself. A button/whatever to show everything would be good though. Most of this is because on the vast majority of forums, there's tons of boards people aren't interested in. More than that though, if this shows every post on a 30,000+ post site, it's just going to be a complete mess.
In terms of the side by side view - I'd suggest keeping it vertical for mobile devices, but would definitely think it's a big improvement as side by side for non-mobile devices.
I think you're talking about the overview, not the main forum view. The main forum view is the one with all the individual posts displayed as cards, where you click through to open them up. You can find it here (http://forum-demo-88817.onmodulus.net/).
As for your suggestions on the overview, The labeling thing is something I've heard before, and as it seems to be a frequent suggestion I'll probably start seeing about implementing it. Not sure on the specifics, though. As for opening nodes individually on the overview, that's supposed to be what the main forum view is for. I do have plans to change what all is displayed, though, and allow people to close and remove nodes from the view. Still figuring out the details.
Oh! And just now, I pushed some updates that should make the graph much more readable with lots of nodes. Still working on a few more small changes, most notably allowing it to shuffle nodes around in a layer for improved viewability.
Edit: And now node shuffling is also implemented, and it looks much better. Gonna go and look at implementing these suggestions for the other views next, and maybe fix some old bugs.
Post by: Retropunch on February 07, 2017, 11:33:55 am
I think you're talking about the overview, not the main forum view. The main forum view is the one with all the individual posts displayed as cards, where you click through to open them up. You can find it here (http://forum-demo-88817.onmodulus.net/).
As for your suggestions on the overview, The labeling thing is something I've heard before, and as it seems to be a frequent suggestion I'll probably start seeing about implementing it. Not sure on the specifics, though. As for opening nodes individually on the overview, that's supposed to be what the main forum view is for. I do have plans to change what all is displayed, though, and allow people to close and remove nodes from the view. Still figuring out the details.
Ahh unfortunately the 'main forum view' doesn't really work for me (and therefore I didn't realise that was the main view) - if I click connecting posts, I see the 'load all' and 'show list' flash and then disappear - even if I'm fast enough to click on them they don't do anything. I've tried in a few different browsers.
Post by: Angle on February 07, 2017, 12:53:48 pm
Post by: Angle on February 12, 2017, 07:21:49 pm
Post by: Retropunch on February 12, 2017, 08:15:52 pm
Well, it took longer than expected, but I fixed up a couple of the bugs with chrome. It should be much more cooperative now. Anyone give it a try and see if it works better for other peoples chromes, and not just mine?
It works now, thanks!! Great work all over.
I've got a bit of feedback though(again, sorry to sound critical, it's awesome work!):
- Firstly, clicking on connected posts/load all resets my view to somewhere in the middle of all the posts - it makes me lose track of where I am unfortunately.
- Secondly, but most importantly, this really, really needs a zoom in/out function!
- I'd recommend that when you close one post, it closes all the posts beneath it in the chain. You could have two close buttons, one that closes everything, and one that keeps the posts under it.
- There's a lot of empty space on each card - real estate is at an absolute premium in this kinda design, so each card should be as tight as possible design wise. I wouldn't both with
- I'd change the 'connecting posts (load all and show list)' and 'more' buttons to just being three buttons you can click on from the card - even if you want to add more options (and hence need a fold out more button) then at least allow the user to open up connecting posts and show a list in one click.
All this being said, I think there's a bit of a bigger problem here; you've pretty much got two completely different forum systems running at the same time. The card based view is good, and the node based view is good, but they seem like two separate thing and I don't think they mesh together well.
I'd honestly suggest sticking with cards or nodes and developing that fully. I much prefer the node based design (with some more headings) as the card based design gets really, really quickly unmanageable - imagine loading up even the amount of posts we have on this thread as cards - let alone a thread with 400+ pages.
You can always go back and implement the other system later as an alternative view (or completely different forum software), but currently it's sorta pulling things in two directions at the same time.
Post by: Angle on February 12, 2017, 08:52:40 pm
It works now, thanks!! Great work all over.
I've got a bit of feedback though(again, sorry to sound critical, it's awesome work!):
- Firstly, clicking on connected posts/load all resets my view to somewhere in the middle of all the posts - it makes me lose track of where I am unfortunately.
- Secondly, but most importantly, this really, really needs a zoom in/out function!
- I'd recommend that when you close one post, it closes all the posts beneath it in the chain. You could have two close buttons, one that closes everything, and one that keeps the posts under it.
- There's a lot of empty space on each card - real estate is at an absolute premium in this kinda design, so each card should be as tight as possible design wise. I wouldn't both with
- I'd change the 'connecting posts (load all and show list)' and 'more' buttons to just being three buttons you can click on from the card - even if you want to add more options (and hence need a fold out more button) then at least allow the user to open up connecting posts and show a list in one click.
All this being said, I think there's a bit of a bigger problem here; you've pretty much got two completely different forum systems running at the same time. The card based view is good, and the node based view is good, but they seem like two separate thing and I don't think they mesh together well.
I'd honestly suggest sticking with cards or nodes and developing that fully. I much prefer the node based design (with some more headings) as the card based design gets really, really quickly unmanageable - imagine loading up even the amount of posts we have on this thread as cards - let alone a thread with 400+ pages.
You can always go back and implement the other system later as an alternative view (or completely different forum software), but currently it's sorta pulling things in two directions at the same time.
Again, thank you! Feedback is what I really need right now, criticism or otherwise. Even better if it's nice and polite criticism.
- Yeah, I'm gonna go ahead and break off that functionality and make it so it doesn't constantly do that.
- I'm definitely considering this. I actually had an idea (http://forum-demo-88817.onmodulus.net/forum/post/Jb8gQdXC2PXg49qKn) to merge the functionality of the overview and forum view graphs with zooming, though I'm not sure it's a good idea. I could also just implement normal zooming - we had that at one point but removed it cause it didn't work super well. Shouldn't be too hard to re-implement and get working properly. Should it trigger off of mousewheel or have a zoom bar, do you think?
- I don't think it should always do this, but adding a button for it would be easy enough.
- Yeah we're planning to rework the cards at some point. Not sure on the specifics yet though. We may want to add more information? Like word count, date last edited, etc. And of course change up the buttons while we're at it.
Yeah, we've been thinking about that. I was considering adding a gmail style form for writing new replies, and having that work with the node system? I'm not too sure yet though. Regardless, we are going to have some kind of system for skipping though links more quickly no matter what we do - loading each layer a bit at a time is silly. Probably wait until after we have voting and maybe SAA, though, so we have a metric by which to choose what posts to display.
EDIT: I just added a button to close all nodes descending from a node, recursively. It doesn't close the node itself, but that only takes one extra click anyway. Though I did make it so that the buttons only available if all of the nodes children are open, in retrospect I should change that to only require one open. I'll do that later though. Oh, and it'll take a few minutes for the server to catch up.
Post by: Retropunch on February 13, 2017, 05:14:21 am
Again, thank you! Feedback is what I really need right now, criticism or otherwise. Even better if it's nice and polite criticism.
- Yeah, I'm gonna go ahead and break off that functionality and make it so it doesn't constantly do that.
- I'm definitely considering this. I actually had an idea (http://forum-demo-88817.onmodulus.net/forum/post/Jb8gQdXC2PXg49qKn) to merge the functionality of the overview and forum view graphs with zooming, though I'm not sure it's a good idea. I could also just implement normal zooming - we had that at one point but removed it cause it didn't work super well. Shouldn't be too hard to re-implement and get working properly. Should it trigger off of mousewheel or have a zoom bar, do you think?
- I don't think it should always do this, but adding a button for it would be easy enough.
- Yeah we're planning to rework the cards at some point. Not sure on the specifics yet though. We may want to add more information? Like word count, date last edited, etc. And of course change up the buttons while we're at it.
Yeah, we've been thinking about that. I was considering adding a gmail style form for writing new replies, and having that work with the node system? I'm not too sure yet though. Regardless, we are going to have some kind of system for skipping though links more quickly no matter what we do - loading each layer a bit at a time is silly. Probably wait until after we have voting and maybe SAA, though, so we have a metric by which to choose what posts to display.
EDIT: I just added a button to close all nodes descending from a node, recursively. It doesn't close the node itself, but that only takes one extra click anyway. Though I did make it so that the buttons only available if all of the nodes children are open, in retrospect I should change that to only require one open. I'll do that later though. Oh, and it'll take a few minutes for the server to catch up.
I'm happy to help - even if I do feel a bit awful coming on and criticising every time I post! I'm truly a big fan though, I've always felt that forums need a bit of a shakeup.
Re Zooming: I'd suggest both mouse wheel and zoom bar - some computers don't have mouse wheels, and laptop trackpads can be awkward to work with.
Re Extra card data: I'd definitely keep it as minimal as possible - any extra information at that stage is just going to be info overload - if they want to get that info then they can look at the post itself.
In terms of a redesign - I'd really try to have everything on one page if possible. Having the node/card diagram on the right with the full text of the clicked post on the left for instance, could work well. Mobile screens would obviously have to have a separate layout, but I believe that forum usage is still largely skewed towards desktops/laptops.
Post by: Angle on February 27, 2017, 09:56:24 pm
I'm happy to help - even if I do feel a bit awful coming on and criticising every time I post! I'm truly a big fan though, I've always felt that forums need a bit of a shakeup.
Don't worry about it! Polite useful criticism is a real treasure. Think of it this way - anything you don't point out politely now, some asshole can be a dick about later. XD
Re Zooming: I'd suggest both mouse wheel and zoom bar - some computers don't have mouse wheels, and laptop trackpads can be awkward to work with.
Mmm, good point. We'll add both of those, then.
Re Extra card data: I'd definitely keep it as minimal as possible - any extra information at that stage is just going to be info overload - if they want to get that info then they can look at the post itself.
Yeah, this sounds best.
In terms of a redesign - I'd really try to have everything on one page if possible. Having the node/card diagram on the right with the full text of the clicked post on the left for instance, could work well. Mobile screens would obviously have to have a separate layout, but I believe that forum usage is still largely skewed towards desktops/laptops.
Hmm, we'll deifnitely consider it. Keeping things on one page would be really nice. I'm not entirely sure we want to make the view quite that crowded, though. We'll try some stuff and see.
For the moment we're working mostly on refactoring, minor presentation changes, and bug fixes. After that we'll see about some more major changes.
Post by: Angle on March 06, 2017, 06:41:00 pm
Post by: Angle on April 10, 2017, 07:44:32 pm
Spoiler (click to show/hide)
Post by: Angle on May 03, 2017, 08:06:01 pm
https://agora-2866.nodechef.com/
Post by: Parsely on May 03, 2017, 11:01:29 pm
Post by: Angle on May 03, 2017, 11:50:34 pm
Post by: Angle on May 10, 2017, 05:16:50 pm
Post by: Angle on June 03, 2017, 03:42:16 pm
Post by: Angle on June 03, 2017, 06:23:54 pm
https://agora-2866.nodechef.com/
Post by: Angle on June 14, 2017, 01:10:47 pm
https://agora-2866.nodechef.com/
Post by: Antsan on June 14, 2017, 01:24:10 pm
I'm not sure how exactly that could work, though. If the currently visible nodes are spread evenly across screen width, how do you handle the user scrolling sideways?
Also when zoomed in very far the levels stay aligned, which means that a discussion with short posts takes up the same vertical space as a discussion with very long posts. This does seem a little more important to me than the issue above.
Maybe arrange the topics in columns at a certain zoom level and instead of a continuous scroll, let users do whole-column-scrolling. The question is what columns to consider when scrolling, because not all columns are "filled" equally far.
When zooming out so far that only dots are visible the dots that were expanded before seem to be loading something. That seems like a bug to me – you're displaying the same dot as everywhere else, what are you loading?
Overall this does look a lot better than when I last looked at it, though.
Post by: Angle on June 14, 2017, 06:51:23 pm
I'm looking at this on a 1920x1080 monitor and for that long thread it seems like at the zoom levels where the, uhm, headers are still missing, there's a lot of unused space to the sides. Maybe stretch the visible levels out so that they stretch over the whole screen horizontally? I think that would make it look less cramped.
I'm not sure how exactly that could work, though. If the currently visible nodes are spread evenly across screen width, how do you handle the user scrolling sideways?
Yeah, I'm not sure how to deal with that either. Especially given that a thred might suddenly baloon out, and theres no way to predict how many nodes you're gonna have next to each other other than counting them, so you could end up with really wide flat nodes.
Also when zoomed in very far the levels stay aligned, which means that a discussion with short posts takes up the same vertical space as a discussion with very long posts. This does seem a little more important to me than the issue above.
I figured the answer to that would just be to let people zoom in more or less as they pleased? So in a discussion with larger nodes, people will naturally zoom in farther to see them? How does that work out for you in practice?
Maybe arrange the topics in columns at a certain zoom level and instead of a continuous scroll, let users do whole-column-scrolling. The question is what columns to consider when scrolling, because not all columns are "filled" equally far.
I'm not sure I know what you mean. Like, have solid columns show up that users can scroll normally? How does that work with splitting conversations?
When zooming out so far that only dots are visible the dots that were expanded before seem to be loading something. That seems like a bug to me – you're displaying the same dot as everywhere else, what are you loading?
Did they disappear after a second? The loading spinners often hang around for a second or so for me, but they disappear pretty quick after that. :/
Overall this does look a lot better than when I last looked at it, though.
Thank you! :D
Post by: Antsan on June 15, 2017, 09:41:49 am
Quote
I figured the answer to that would just be to let people zoom in more or less as they pleased? So in a discussion with larger nodes, people will naturally zoom in farther to see them? How does that work out for you in practice?The problem is not the discussions with larger nodes but the ones with smaller nodes. Those just are spaced out very much vertically. Actually, they are spaced with constant height and I guess that's my problem. When zooming in so far that I can comfortably read the text in the smaller nodes, I can see three at once on my monitor with 1080 pixels of height. I'd like to see more. Aligning vertically related nodes top-to-bottom would seem more sensible to me.
Quote
I'm not sure I know what you mean. Like, have solid columns show up that users can scroll normally? How does that work with splitting conversations?Conversations don't rejoin in the current view, as far as I can see. Arrange nodes only in columns instead of a grid. Columns can contain a splitting conversation by containing n columns where the conversation splits into n parallel conversations.
"Scrolling by column" would then mean that there are buttons at the left on and the right. Clicking on one of those would move the view by one column into the appropriate direction. So you'd basically scroll between parallel conversations.
Vertical scrolling would work as it does currently.
Quote
Did they disappear after a second? The loading spinners often hang around for a second or so for me, but they disappear pretty quick after that. :/Yes, they disappear. I'm just confused that they show up at all when zooming out to a level where the graphic to put into their place is a constant.
Post by: Angle on June 15, 2017, 01:08:04 pm
The problem is not the discussions with larger nodes but the ones with smaller nodes. Those just are spaced out very much vertically. Actually, they are spaced with constant height and I guess that's my problem. When zooming in so far that I can comfortably read the text in the smaller nodes, I can see three at once on my monitor with 1080 pixels of height. I'd like to see more. Aligning vertically related nodes top-to-bottom would seem more sensible to me.
Hmm, yeah we can see about dynamically spacing nodes. At a guess though, this is gonna be a real pain to implement. How important do you think it is? :/
Conversations don't rejoin in the current view, as far as I can see.
Well, not yet. They probably will though, once we reimplement that feature.
Arrange nodes only in columns instead of a grid. Columns can contain a splitting conversation by containing n columns where the conversation splits into n parallel conversations.
"Scrolling by column" would then mean that there are buttons at the left on and the right. Clicking on one of those would move the view by one column into the appropriate direction. So you'd basically scroll between parallel conversations.
Vertical scrolling would work as it does currently.
So you mean columns within columns? That sounds awfully complicated. How does it work when you have constant splitting? Do posts that would currently be above each other but not directly connected appear in the same column? Or do you mean more like reddit? :/
Yes, they disappear. I'm just confused that they show up at all when zooming out to a level where the graphic to put into their place is a constant.
That's just the code being slow to catch up. Not sure how much I can do about that but I'll take a look at it at some point.
Post by: Antsan on June 15, 2017, 02:34:18 pm
Quote
Hmm, yeah we can see about dynamically spacing nodes. At a guess though, this is gonna be a real pain to implement. How important do you think it is? :/It does slow down reading a bit since this means more scrolling and scrolling means loading nodes. Testing it out it's pretty minor, though.
Quote
Well, not yet. They probably will [rejoin] though, once we reimplement that feature.Hm, okay, that does change things a bit.
Quote
So you mean columns within columns? That sounds awfully complicated. How does it work when you have constant splitting? Do posts that would currently be above each other but not directly connected appear in the same column? Or do you mean more like reddit? :/Yeah, that was the idea, definitely not like reddit. Those two questions to show some serious problems with my imagined approach, though, and so do rejoining conversations.
Post by: Angle on June 15, 2017, 04:42:21 pm
It does slow down reading a bit since this means more scrolling and scrolling means loading nodes. Testing it out it's pretty minor, though.
Hm, okay, that does change things a bit.
Yeah, that was the idea, definitely not like reddit. Those two questions to show some serious problems with my imagined approach, though, and so do rejoining conversations.
Well, as it happens, I'm currently thinking that I'll be working on changing the layout eventually. Not sure how exactly, but I'll figure something out. probably be more or less like it is now, but with a different algorithm for arranging nodes. I think I'm gonna implement federation (Different servers talking to each other so you can view posts on any of them from any other.) via ostatus first though? And probably proper user profiles and alerts and all that. Though I might get someone else to work with layouts...
Post by: Angle on June 24, 2017, 08:53:22 pm
Post by: Angle on July 15, 2017, 12:01:14 am
Post by: Angle on July 29, 2017, 07:38:27 pm
Post by: Angle on March 26, 2018, 11:25:08 pm
https://agora-2866.nodechef.com/forum
Post by: dorf on March 27, 2018, 04:56:56 am
The preview feature is functional and cool-looking.
The zoom feature and moving around the graph is sluggish for me on almost-latest version of Chrome on Windows.
Post by: blueturtle1134 on March 27, 2018, 06:49:05 pm
- Different types of connections, denoting different relations between nodes. For example, reply, subordination, approval, contradiction, etc... Also ability to add connections post-node-creation, so someone can go "oh, these are related!" *bamf*
- Different types of nodes denoting different functions. Some should be headers, for example, that other stuff goes under. Or +1 posts can get a different color.
A non-linear forum is an interesting concept. It'd fix the problem of unrelated conversations in the same thread - if you don't have anything to add to the person above you you're just replying to OP you can just reply to OP.
Post by: Angle on March 27, 2018, 09:24:16 pm
Noticed this and just want to throw out some suggestions:
- Different types of connections, denoting different relations between nodes. For example, reply, subordination, approval, contradiction, etc... Also ability to add connections post-node-creation, so someone can go "oh, these are related!" *bamf*
- Different types of nodes denoting different functions. Some should be headers, for example, that other stuff goes under. Or +1 posts can get a different color.
A non-linear forum is an interesting concept. It'd fix the problem of unrelated conversations in the same thread - if you don't have anything to add to the person above you you're just replying to OP you can just reply to OP.
Aye, those are all features I'm planning to work on. I'm also thinking of a connection between posts that basically says "these posts say the same thing", which would cause replies to one to show up as replies to both? I'm not sure though.
Nice to see this is still being worked on.
The preview feature is functional and cool-looking.
The zoom feature and moving around the graph is sluggish for me on almost-latest version of Chrome on Windows.
As for the lag, I'm also still trying to optimize a bit more, too, so hopefully it'll work better. I'm curious as to what sort of computer it was running on? Like, an old laptop, or a new gaming computer...?
Post by: dorf on April 05, 2018, 06:21:00 am
I've tried it on Chrome and FF. Both are similarly sluggish, FF might be a bit less sluggish.
The size of the window also plays a factor.
Post by: Angle on April 05, 2018, 02:25:02 pm
Post by: dorf on April 09, 2018, 01:45:41 am
The site loads in ~5 secs, the times for each request range from 100 ms to 400 ms.
Not sure how to measure websocket's lag, but it seems quite fast.
Also, on Chrome the dots are only partly connected: https://imgur.com/a/5JQ3A
On FF they are fully connected.
Post by: Angle on April 09, 2018, 05:44:28 pm
Post by: Angle on June 09, 2018, 01:34:40 pm