Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Retropunch]

As far as skills go, I much, much prefer the method of skills increasing with use rather than any other method of point allocation. I think it's necessary to have clear skill progress and not to obfuscate the advancement process, but point allocation seems to take away from any sense of immersion. When I was working on CyberRogue (which I still am slowly!) I tried quite a number of systems and this seemed to work best, especially with a mix of combat and non-combat skills.

I think the trick with this is to make skills overlap quite significantly, so you don't end up with the 'training shields by letting a rat bite you 1000 times' thing. For instance, training any weapon could increase your skill mostly in that weapon, but also improve skills in all others of a similar type, as well as athletics and strength. Similarly, training lockpicking could also train (to a much lesser extent) other skills which require precision such as surgery and trap making. With enough overlap, character progression should become quite natural, but still allow the player to specialise.

[Scoops Novel]

Will gambling be in?

[puke]

The more I think about this project (library), the more I realize it is beyond my maths. 

I was wanting to generate each book from a seed, but also to ensure that the generated books never overlap.  I'm not sure where to start with that one, ensuring uniqueness.  Most algorithms that generate keys and hashes rely on an extremely low probability of overlap, but the purpose of this is the absurd thought experiment, one would want to actually ensure that each book was unique and not just pseudo-unique.

oh well [/derail]

[Ultima Ratio Regum]

Dang.

... but still no magic. Sorry!

And-i-hoped-magic-would-exist-but-no. HOW SAD (But-i-still-hope-this-changes-soon-or-at-the-far-future)
Still-your-game-is-great.

Never happening, I'm afraid. There's a million other roguelikes with magic! Asking me to add magic, at this point, is like going up to Netflix half-way House of Cards, say, and saying "please add magic!". It just doesn't fit into the game's design *at all* and it just wouldn't make any sense to integrate it, and would clash strongly with both the gameplay and the narratological/thematic goals. Also, I must say, I don't feel adding magic inherently makes a game better, any more than adding survival mechanics inherently makes a game better (indeed, in many instances, it makes games worse)!

Haha, you could totally algorithmically generate the full text of every one of the 410 page books.  It might be tricksey if you wanted to be able to turn to a random page in a book without generating the whole thing... but even now I'm thinking of ways you could solve that.  This would be a cool little project...

Have you been reading my design document?! But I do very shortly intend to get the game generating books, both fiction and non-fiction, and as with everything else, tie that to world history; so maybe you have a particular cult that uses a particular symbol (an animal, say), then someone might have written a book which is a parable about that cult by using that animal symbol, then maybe a library somewhere else has acquired that book, but misinterpreted it as a historical treatise on a mythical creative which dwells in that far-off land... and thus the truth becomes obscured, but the player is still given a clue to point them in the right direction!

As far as skills go, I much, much prefer the method of skills increasing with use rather than any other method of point allocation. I think it's necessary to have clear skill progress and not to obfuscate the advancement process, but point allocation seems to take away from any sense of immersion. When I was working on CyberRogue (which I still am slowly!) I tried quite a number of systems and this seemed to work best, especially with a mix of combat and non-combat skills.

I think the trick with this is to make skills overlap quite significantly, so you don't end up with the 'training shields by letting a rat bite you 1000 times' thing. For instance, training any weapon could increase your skill mostly in that weapon, but also improve skills in all others of a similar type, as well as athletics and strength. Similarly, training lockpicking could also train (to a much lesser extent) other skills which require precision such as surgery and trap making. With enough overlap, character progression should become quite natural, but still allow the player to specialise.

This all makes a lot of sense to me: to be honest, though, there's a good chance I'll end up with an extremely minimalist skill system, or possibly even no skill system at all (!) having everything for combat etc dependent on base stats and the weapons/armour you're actually wielding, and have most other important skills reside in the *player* (i.e. deduction) rather than in the player *character*.

Will gambling be in?

Probably!

[Piotrhabera]

Well. There may be mods that may ACTUALLY add magic. but i am not into mods now. (almost all games can be modded.)
Anyways. Will be there multi-tile animals? (dragons in DF were as big as an tower and as long as a rather long chunk of stone walls and yet they are in one tile) And will be there a possibility to found your own cities & make a new kindgom? Select your religion, city types and city pattern...

[Ultima Ratio Regum]

Well. There may be mods that may ACTUALLY add magic. but i am not into mods now. (almost all games can be modded.)
Anyways. Will be there multi-tile animals? (dragons in DF were as big as an tower and as long as a rather long chunk of stone walls and yet they are in one tile) And will be there a possibility to found your own cities & make a new kindgom? Select your religion, city types and city pattern...

I hate to continually disappoint... but I have no intention of enabling mods in the slightest, and if anything, I'm writing the code in such a way that making mods should be extremely challenging. There WILL, however, be multi-tile animals! Things like horses/camels/elephants as modes of transport should take up multiple tiles; as for founding cities... er... not as SUCH, but you'll certainly be able to toy with the fates of nations. Do you know the "Plots" system in Crusader Kings II? Well, something along those lines is being worked on whilst I also work on NPCs in general...

[Retropunch]

This all makes a lot of sense to me: to be honest, though, there's a good chance I'll end up with an extremely minimalist skill system, or possibly even no skill system at all (!) having everything for combat etc dependent on base stats and the weapons/armour you're actually wielding, and have most other important skills reside in the *player* (i.e. deduction) rather than in the player *character*.

I really like the idea of the players own skills being important, and that's one of the reasons I'm so excited for URR. However, I would caution against not having a skill system or, what I feel is worse, having it just dependant on base stats that leave the player possibly stuck with a character config they don't like for a whole play-through.

I feel quite strongly about this, as I believe that the advancement of the player character/avatar is integral to the gaming experience, especially in an open-world/RPG game where there is no linear story arc that develops your character. Without skills or some internal advancement, it's difficult for a character to really 'grow', which is important for the player to become more attached to and invested in the character and the game world. Or more simply, every character becomes completely the same, leaving no real replay value in that respect.

Furthermore, without skill advancement, the challenge becomes static - especially without magic, as it won't be like the player can get 'better' equipment. Sure, they might be able to find/buy more stuff to be of more use in a number of situations, but as soon as you find the 'best' armour/weapon in the game world, you automatically become the best fighter you can be.

I'm not saying you should stick to a traditional RPG style of skills, but I feel the character needs some 'internal' development to make the player feel that they're advancing - as you'd improve at skills in Real life(tm)! I could imagine this could be subtle, with no direct manipulation by the player. Perhaps skills could only be in a number of broader areas, such as 'seafaring, tracking, melee combat, ranged combat, speechcraft, athletics' etc. which would increase and decrease with time spent on activities associated with them (I'm sure you can think of more interesting systems!)



 

[Piotrhabera]

No-magic rougelikes are more interesting. but UnReal World has rituals. URR might too have, One can learn ancient tribal rituals if one either sees someone doing it or finds a book eplaining it or he/she is simply taught it. I would be very happy that URR includes rituals. But if no rituals you shall add in URR: Do not worry. I will not be dissapointed even a slightlest bit.

[Servant Corps]

I was wanting to generate each book from a seed, but also to ensure that the generated books never overlap.  I'm not sure where to start with that one, ensuring uniqueness.  Most algorithms that generate keys and hashes rely on an extremely low probability of overlap, but the purpose of this is the absurd thought experiment, one would want to actually ensure that each book was unique and not just pseudo-unique.

Step 1: Generate a book. Print it out to the public.
Step 2: Save that book in a file (call it "list of generated books").
Step 3: Generate a second book.
Step 4: Query the list of generated books to see if your second book has exactly the same text as the first book. If it does, then simply re-generate your book and then query your list again.
Step 5: If your newly generated book does not match any books on your list, then it's unique. Congrats! Save that book in your "list of generated books".
Step 6: Repeat steps 3 to 5 until your computer runs out of memory.

Also, somebody did end up creating a Library of Babel book generator, so you could use that code as inspiration.

But I do very shortly intend to get the game generating books

Dr. Mark, are you expecting them to generate good books? I looked up NaNoGenMo (National Novel Generator Month*), and actually, they ended up succeeding in producing loads of novel generators on out there, all of them "open source" and available on GitHub.

And, all of them equally terrible in their own special ways. The ones that make the most sense are the ones that have "templates", but they are the also the ones that require the most work to do. Other ways to generate a novel would be to scrape content from websites, write short stories and stick them so close to each other so that readers are tricked into thinking these short stories are somehow related, replace words in existing novels, use recursive patterns, and create a simulation of different characters interacting with each other and printing out the results (the CK2 approach). And there are probably more ways to write a novel out there but I don't want to get distracted looking at all these generators when I'm still trying to get my own "novel generator" functional.

Put it frankly, while it may still be possible to generate a full-fledged novel that people would want to read, it'll take a lot of effort, and you still have an actual game to ship as well! I'd recommend demoting your goal down to just generating short stories, or...quite possibly, creating a generator that produces a short story, and treat that short story as a synopsis for the novel in question, so that people can imagine the novel without you needing to come up with said novel. Think of how Tarn Adams handled poetry in Dwarf Fortress.

*2013 NaNoGenMo entries
*2014 NaNoGenMo entries

[puke]

Step 1: Generate a book. Print it out to the public.
Step 2: Save that book in a file (call it "list of generated books").
Step 3: Generate a second book.
Step 4: Query the list of generated books to see if your second book has exactly the same text as the first book. If it does, then simply re-generate your book and then query your list again.
Step 5: If your newly generated book does not match any books on your list, then it's unique. Congrats! Save that book in your "list of generated books".
Step 6: Repeat steps 3 to 5 until your computer runs out of memory.

Also, somebody did end up creating a Library of Babel book generator, so you could use that code as inspiration.

Whoah, no, god no that would never work.  I mean, it would work in that your computer would run out of memory, but you could never actually create the complete set of books.  You'd run out of atoms in the galaxy.  The set would have greater volume than the observable universe.

You have to generate them on demand, and not sequentially.  You have to discard things volatily after generating them, and just be content that you can regenerate them with the same seeds if you ever need to observe them again.

That in its self isnt super hard, but making sure you cant produce a duplicate.. There might be a trick to it, I suspect someone well versed in fractal maths might have an idea. 

On the other hand, I have an equal suspicion that writing an actual mathematical proof for a method of generating non-unique strings is the sort of thing that would keep academics busy for years.

Either way, far beyond my abilities. 

Also, somebody did end up creating a Library of Babel book generator, so you could use that code as inspiration.

This was exactly the kind of thing I was initially thinking about, but it is just generating random strings and random pages.  There is no attempt at uniqueness.  I also don't see any method of tracking the seed to regenerate the same book if needed.  The whole 410 pages of garbage would need to be saved.

The odds of a repetition before someone gets board and stops generating books is pretty low, but if it was run to its theoretical limits... I'm not even sure how to figure out the odds of a repetition occurring, I suspect it is a  combinitoric problem. 

every possible way of writing a book would be a member of a set, "n" would be the length of the set, n! should be equal to the number of ways that they /dont/ overlap, and n^n are number of ways they /do/ overlap?  so (n!):(n^n) are the odds of generating a duplicate book?

I think I screwed that up, I'm not a statistics guy.

Edit for context:  there would be 2.5E13120000 books in this library.  The upper estimate for atoms in the Observable Universe is 1E83. That is a difference of 13119917 Orders of Magnitude.

[puke]

every possible way of writing a book would be a member of a set, "n" would be the length of the set, n! should be equal to the number of ways that they /dont/ overlap, and n^n are number of ways they /do/ overlap?  so (n!):(n^n) are the odds of generating a duplicate book?

This is retarded.  not only are my numbers wrong, but the premise is broken. 

If "n" are the possible ways of writing a book, n is necessarily equal to the number of non overlapping books.  n! and n^n are meaningless.

Obviously, not a statistics guy.

More edits for context:

If you make the problem more "manageable" by reducing it to 5 possible characters, 5 characters per line, 5 lines per page, and 5 pages per book.. You still end up with 2.4E87 possible books which is 4 orders of magnitude larger than the number of atoms in the observable universe.

I mean, what kind of seed would you have to use for that?  Would there need to be one unique seed for each unique book? 

If so, you'd need a seed of length "z" for x=y^z where "x" is the number of unique items to produce and "y" are the number of legal characters for your seed. 

So, for example, if you allow any of 255 possible bytes to be part of the seed and you are working with my reduced set size of 2.4E87, you'd need a 37 byte long seed (https://www.wolframalpha.com/input/?i=2.4e87%3D255%5Ex)

For the actual library, you seed would need to be 5.5E6 characters (bytes) long.  You seed would need to be FIVE MEGABYTES in size. 

I thought you could reduce this problem by breaking it down into more manageable chunks.  having a unique seed for each room, and then for each bookshelf, and then finally each unique book on the shelf.  But how would you guarantee that you dont generate the same book on multiple shelves or in multiple rooms?  You'd have to use the room and shelf seeds as part of your book generating seed, which I think throws out any benefit there might have been from breaking the problem up into smaller chunks.

[Retropunch]

TL:DR - It's impossible to do properly. You could a fair imitation of it in the number that the player would actually see, but it'd just be millions of books of rubbish, which wouldn't really be any fun for anyone.

Generating 'sensible books' is not a problem though. Think about the game history we already have in URR (and DF!!) if you expanded on that enough it would make a sensible - if slightly stiffly written - history book. Most books of that time were written in that style though, so it's not necessarily a problem. The whole problem with the NaNoGenMo thing is that it's trying to emulate current day fictional novels, rather than books that are already grounded in a world.

This is *much* easier. I've got a chatbot for CyberRogue which seems a lot more realistic (in my opinion) than internet chatbots, as you're talking about stuff in universe. It's situated in a 'physical' space, which makes it much more lifelike. Granted, it's not going to pass the Turing test, but as long as you stay in-universe, it can function well. It grabs bits of the news feed, it has preferences from the consumables available and knows a number of random people in the world that you also know. This definitely transfers to books, so I'm sure the Dr. will be able to turn out some pretty great stuff!
 

 

[Ultima Ratio Regum]

I really like the idea of the players own skills being important, and that's one of the reasons I'm so excited for URR. However, I would caution against not having a skill system or, what I feel is worse, having it just dependant on base stats that leave the player possibly stuck with a character config they don't like for a whole play-through.

Hmmmm, interesting thoughts. Re: your third paragraph, I don't agree issues are quite that "inevitable" - "Sure, they might be able to find/buy more stuff to be of more use in a number of situations, but as soon as you find the 'best' armour/weapon in the game world, you automatically become the best fighter you can be. " - but why can I not extend that to be the entirety of development? Or, at least, say 50% of character development; 50% base stats, 50% items, so something very minimalist (one could reasonably argue that the Souls games do something along those lines, with few skills, no skill trees, etc). I think I could make that work: have enough variation/resolution in the best/worst items, make anything above the basics extremely rare and extremely hard to acquire, and I still think that could actually work really well! I'd actively like to experiment putting the sense of player development into primarily items, and knowledge, rather than stats (and I could go a bit extreme and make *everything* item/learning based... which I might) - I know it's crazy! But at the top of my design document is a sentence to the effect of "Don't do what everyone else has gone", and putting all player progression into items and knowledge would certainly be different. And I think it would be good, and fresh, and WOULD yield progression-feeling!

No-magic rougelikes are more interesting. but UnReal World has rituals. URR might too have, One can learn ancient tribal rituals if one either sees someone doing it or finds a book eplaining it or he/she is simply taught it. I would be very happy that URR includes rituals. But if no rituals you shall add in URR: Do not worry. I will not be dissapointed even a slightlest bit.

Agreed re: non-magic being more interesting! There will be "rituals", but not ones which truly have an effect, if you get what I mean; rituals that propagate and continue religious belief, but not which actually HAVE physical effect in the world.

Dr. Mark, are you expecting them to generate good books? I looked up NaNoGenMo (National Novel Generator Month*), and actually, they ended up succeeding in producing loads of novel generators on out there, all of them "open source" and available on GitHub.

And, all of them equally terrible in their own special ways. The ones that make the most sense are the ones that have "templates", but they are the also the ones that require the most work to do. Other ways to generate a novel would be to scrape content from websites, write short stories and stick them so close to each other so that readers are tricked into thinking these short stories are somehow related, replace words in existing novels, use recursive patterns, and create a simulation of different characters interacting with each other and printing out the results (the CK2 approach). And there are probably more ways to write a novel out there but I don't want to get distracted looking at all these generators when I'm still trying to get my own "novel generator" functional.

Put it frankly, while it may still be possible to generate a full-fledged novel that people would want to read, it'll take a lot of effort, and you still have an actual game to ship as well! I'd recommend demoting your goal down to just generating short stories, or...quite possibly, creating a generator that produces a short story, and treat that short story as a synopsis for the novel in question, so that people can imagine the novel without you needing to come up with said novel. Think of how Tarn Adams handled poetry in Dwarf Fortress.

Interesting discussion! Re: the specific above quote, my answer is: absolutely, but books will be extremely short. That's the trade-off - I want a book, or a piece of poetry, to be only a couple of scenes/paragraphs, but to be MEANINGFUL and relevant for the player's investigation; not a full book of dross. A book that just describes the relevant parts of a battle and gives you a hint as to where it took place, or a piece of poetry that describes a symbol meaningful to a particular culture, or whatever, rather than entire novels (since who wants to read the kind of bilge that a novel generator is going to spew out in this day and age?!). A blurb, as you suggest, is an interesting idea; but I think I'll go with the Skyrim model (as little as I like emulating anything Skyrimian) where "books" are extremely short, but have them generated in serious detail.

This definitely transfers to books, so I'm sure the Dr. will be able to turn out some pretty great stuff!

Yes! I think I can connect books to the other sources of information for symbols, ideas, histories, phrases, religions, the whole nine yards; I'm working on (and will be working on the background for the next couple of releases) the basic overall system for "information -> clues", meaning that, in one sentence:

The game needs to be able to take any piece of *data* - symbol for religion, name of nation, engravings on tomb, name of NPC, pattern on a vase - and transform it into any kind of *clue* - a book, a conversation, a painting, etc...

And obviously that will yield books!

[Retropunch]

I really like the idea of the players own skills being important, and that's one of the reasons I'm so excited for URR. However, I would caution against not having a skill system or, what I feel is worse, having it just dependant on base stats that leave the player possibly stuck with a character config they don't like for a whole play-through.

Hmmmm, interesting thoughts. Re: your third paragraph, I don't agree issues are quite that "inevitable" - "Sure, they might be able to find/buy more stuff to be of more use in a number of situations, but as soon as you find the 'best' armour/weapon in the game world, you automatically become the best fighter you can be. " - but why can I not extend that to be the entirety of development? Or, at least, say 50% of character development; 50% base stats, 50% items, so something very minimalist (one could reasonably argue that the Souls games do something along those lines, with few skills, no skill trees, etc). I think I could make that work: have enough variation/resolution in the best/worst items, make anything above the basics extremely rare and extremely hard to acquire, and I still think that could actually work really well! I'd actively like to experiment putting the sense of player development into primarily items, and knowledge, rather than stats (and I could go a bit extreme and make *everything* item/learning based... which I might) - I know it's crazy! But at the top of my design document is a sentence to the effect of "Don't do what everyone else has gone", and putting all player progression into items and knowledge would certainly be different. And I think it would be good, and fresh, and WOULD yield progression-feeling!

My problem is that keeping things item based and non-magical runs the risk of items becoming unrealistic within the game world. If the game keeps to actual period weapons, the enhancement of the 'next level up' of weapons/armours would have to be a huge leap - a steel sword becomes MUCH better than an iron sword, even though they are essentially the same thing.
Most importantly, in any renaissance/medieval setting there would have been tons of blacksmiths and armourers which would have been able to create weapons and armour, as unless you made items of ridiculous cost it'd be pretty easy to buy all the best stuff pretty much instantly (especially as you start off as a noble person).

I DO like the idea of progress through items (like the Souls games), but I just can't see it working as 50% in a purely 'realistic' setting without having to make items nearly magical in enhancement. Furthermore, the skill of the user would increase over time - not showing that seems to also be unrealistic. If you trained for 6 months with a sword, you'd get much, much better with it!

I have to say though, my biggest problem is base stats - being tied to a few stats that you chose at the start of the game without having a clear knowledge of mechanics/what is useful can be frustrating. Most games that don't allow you to progress those stats through levelling up (D&D CRPGs for instance) give you magical items/potions to enhance stats instead. Keeping the player frozen in time seems (as I pointed out in the last paragraph) slightly unrealistic, and punishes the player for making a poor choice at the beginning. Whilst this would be fine if it was a quick RL, I imagine many players will continue through one game for quite a while in comparison to other RLs, which means their ability to keep resetting their character as they learn more about what base stats are useful may be limited.

Lastly, and I certainly don't mean to come across as belligerent or that I don't believe in what you're doing -  but I think that purposefully going out of your way to 'not do what everyone else has' can be dangerous. The reason that these mechanics have endured is not because other systems haven't been tried before (because they have!), but because they work! I'm not saying to follow their lead exactly, and CERTAINLY try new things, I just feel you should be cautious about dismissing these established systems out of hand. Most players like to progress their character, and they dislike being bound too much to luck/RNG.

Sorry for the length, I just feel it's important to give another viewpoint, as I really believe that good character progress (in whatever form it takes) makes or breaks an RPG.

 

[Zireael]

This is *much* easier. I've got a chatbot for CyberRogue which seems a lot more realistic (in my opinion) than internet chatbots, as you're talking about stuff in universe. It's situated in a 'physical' space, which makes it much more lifelike. Granted, it's not going to pass the Turing test, but as long as you stay in-universe, it can function well. It grabs bits of the news feed, it has preferences from the consumables available and knows a number of random people in the world that you also know. This definitely transfers to books, so I'm sure the Dr. will be able to turn out some pretty great stuff!

I like that chatbot already - sounds like something I want to have in my game (a logical, if maybe not 100% realistic, universe that reacts to what the player does).