According to devlog, that day Toady posted a bunch FotF reply links... and that's it v:
I'm being taken off the top 10 post count by FG&RPers. I must tell myself, it isn't worth it, that we don't need another wave of LW spam...
Too late. You will never reclaim your previous glory. Now all you can bask is the leftover aftertaste./me holds a box of post count enhancing steroids, labeled FG&RP.
/me gives LW a supportive hug.
Now, if only my D & D campaign could be that big...Your game is already got 7 people. That's a ton. Seriously. I've got like 5 players in my game, and it's a ton of work keeping them all busy.
Right, David? :P
Really, there was a crazy members-online jump back in late 2010? What happened back there?
By the way, the early 2010 and 2012 spikes are undoubtedly DF2010 and DF2012, incase you were wondering.
Now, if only my D & D campaign could be that big...Your game is already got 7 people. That's a ton. Seriously. I've got like 5 players in my game, and it's a ton of work keeping them all busy.
Right, David? :PReally, there was a crazy members-online jump back in late 2010? What happened back there?By the way, the EARLY 2010 and 2012 spikes are undoubtedly DF2010 and DF2012, incase you were wondering.
Really, there was a crazy members-online jump back in late 2010? What happened back there?
December 8th, 2010 is still a mystery to me, however.
Its less FG and more RP, actually.It's stupid huge.Realm of the mad godRise of the magic girls has a ridiculous amount of posts for something that isn't a huge rtd or in general discussion, it has somewhere in the rage of 40-50% of the posts the happy thread has according the the bars in the stats center.
Starting in roughly 2014 we changed how members were registered because of the love bots -- about 2/3rds (very approximately) of the signups before then starting especially around 2011 were spammers. Now they only get in every once in a while. I'm still cleaning that up (I've deleted at least ten thousand accounts over the last couple years, and there's a lot left to go). So total member counts are basically useless, though members online isn't really affected by spammers (except maybe that weird day). As far as I can tell, the actual-people membership slowly increases, as does the people online, with the release spikes. Everything else seems to track with that as well, though I haven't done much with it.Christ, how do you even find that many bots?
Christ, how do you even find that many bots?It was pretty bad for a while. :P (And it wasn't helped that a fair share of the bots/bot-people would sign-up, and then wait for quite a while before actually posting anything).
I have a much better question:From what I understand, he spends the first part of his day at home programming dwarf fortress pretty much with minimal interuption, and then later in the day he goes over to his brother Threetoe's place, and that's when he moderates on the forums, without being distracted by programming.
How do you find time to work on DF if you have to delete that many spambot accounts!?
I have a much better question:
How do you find time to work on DF if you have to delete that many spambot accounts!?
On a slightly related note it appears that the post count for the DF Wiki discussion board is broken. The forum seems convinced that the board only has 3 total posts (despite having 152 topics) and thus anyone who has posted 3+ times in it automatically gets this:Spoiler (click to show/hide)
Shame we don't have the older data, from before the forum switch. That was 2008, right? I'm curious how many members there were regularly online back when I joined...Yes, it would be pre-June 15th, 2008.
holycrapwejustbeatthemostmembersonlineatoncerecord
itshouldbeuntouchable
Ptw.
Anyway, I find it really interesting how the "topics per month" is decreasing fairly rapidly, compared to posts, members, etc. which are all increasing steadily.
I feel like that's because the amount of megathreads is increasing, and those cut down on new topics.
Also new users looks like it's been on a steady decline these past few years. That's probably part of it too - new topics seem to correlate with new members better than any of the other graphs.
Can someone yell at me to update this tomorrow? Thanks.Hey, update the thing.
You're right, new members is plummeting with the force of a thousand suns. Any ideas what's going on there?I'm... not entirely sure. Maybe DF has reached it's publicity limit? Maybe we're less inclusive than we used to be? idk
Starting in roughly 2014 we changed how members were registered because of the love bots -- about 2/3rds (very approximately) of the signups before then starting especially around 2011 were spammers. Now they only get in every once in a while. I'm still cleaning that up (I've deleted at least ten thousand accounts over the last couple years, and there's a lot left to go). So total member counts are basically useless, though members online isn't really affected by spammers (except maybe that weird day). As far as I can tell, the actual-people membership slowly increases, as does the people online, with the release spikes. Everything else seems to track with that as well, though I haven't done much with it.combined with the relative length of time since a major update.
-snip-
I'm not worried. It would take considerable effort to influence the numbers enough. You could do it with 10 reasonably active posters doubling their average count for a day, but that's not an easy feat.
Not sure what you mean. There are no exponents involved, if that's what you're asking.-snip-
Huh. That's pretty neat. Is the data more or less exponential?
These calculations excite me.I'm glad that they excite you! They excite me too!
I demand more predictions.
I suppose one could calculate the average growth or decay rate of posts of an individual, particular board, make a prediction based on previous trends of that board, and sum them all to get a more precise prediction?Indeed you could, if the forum displayed stats like that. Sadly, it simply cumulates the data from all of the boards to give us blanket stats. If it gave individual board stats, then I would jump on that in an instant. I've looked for a way to make that work, but I've come back empty handed. This is what the stats look like, for reference. (http://www.bay12forums.com/smf/index.php?action=stats)
Also, it might make it more accurate to claim (for example) that the announcements board will have 1 new post per month, with 100% accuracy, but other boards, like...DF Adventure Mode Discussion, might not be as reliable for post count records. To get rid of that problem, I assume, day-by-day variance should be analyzed to see which boards are most faithfully being posted in.I already track day-by-day variance to make these predictions, but again, being able to access the individual board stats would be amazing.
And, uh.... STANDARD DEVIATION! Right? Am I statistics now?You're pretty close! I'm not using Standard Deviations in this, because I don't find it really useful. It might work better with that, though. I'll have to try it out. Thanks for the idea!
You might not be able to get stats for all the threads, but the postcounts for the ten most posted-in boards are available.Yes, their total post count. However, I'm looking at daily posts, not total. THEORETICALLY, I could check them every 24 hours and figure the difference to get the daily posts in a board, but that's more maintenance work than I want to handle. Alternatively, it could be automated, but I don't have the skills to make that happen.
May 14th's prediction is 1865. With 4 hours left in the day by the forum's count, and only 1662 posts made today, the prediction is probably going to a little low, but who knows?2065, exactly 200 off. :P 10% off aint bad.
Not sure what you mean. There are no exponents involved, if that's what you're asking.-snip-
Huh. That's pretty neat. Is the data more or less exponential?
You mean just the New Members/Day, sans prediction? I've graphed that. (https://docs.google.com/spreadsheets/d/1o7rk4oGTGNEJ4wD2NJJvjpg-da6s3nB90wPOiv4anNQ/pubchart?oid=943924740&format=interactive) Or are you referring to something else?Not sure what you mean. There are no exponents involved, if that's what you're asking.-snip-
Huh. That's pretty neat. Is the data more or less exponential?
That if one were to just plot the number of registrants/day it would form an exponential curve.
May 16th, we should see ~2054 posts.2186 posts! I'm loving this accuracy.
~2164 posts on the 17th.1954. 9.7% off.
1951 for the 18th.2872 is very off.
So, what do you all want me to do with this predictive program? Continue to post predictions like this every day? Let it run in the background, gathering more data? I don't mind continuing, and I don't mind stopping.
Collect data and do predictions for longer periods at a time.
Try developing better models.
Stuff like that. :v
This tickles my morbid fascination with statistics. I don't fully understand all of those numbers, but I'm inexorably drawn to them.I can explain them to you. That's half of why I'm here.
Which one? They're all in decline. It's normal for it to have a downturn after a major release.
That'd be my guess as well. Remember, the dive started back around late 2013-early 2014ish regarding this:My guess is that since Toady implemented the requirement for accounts to be approved we're finally cutting out all of the many spam accounts that are joining the forum.I'm... not entirely sure. Maybe DF has reached it's publicity limit? Maybe we're less inclusive than we used to be? idkObservation: Daily data from New Members (Monthly too, but to a lesser, vaguer, extent) show a serious decline, now worth making "HMMMM" noises over.You're right, new members is plummeting with the force of a thousand suns. Any ideas what's going on there?
Our lowest ever New Members in one day was October 23rd, 2008, with just one. There have been 2094 days between our last occurrence of only 4 new members in a day, June 28th, 2009 and March 23rd, 2015. The rest of the recent new member counts are back down to 2008/2009 levels. I realize this is mostly due to the lack of major releases, but it's still interesting.Starting in roughly 2014 we changed how members were registered because of the love bots -- about 2/3rds (very approximately) of the signups before then starting especially around 2011 were spammers. Now they only get in every once in a while. I'm still cleaning that up (I've deleted at least ten thousand accounts over the last couple years, and there's a lot left to go). So total member counts are basically useless, though members online isn't really affected by spammers (except maybe that weird day). As far as I can tell, the actual-people membership slowly increases, as does the people online, with the release spikes. Everything else seems to track with that as well, though I haven't done much with it.combined with the relative length of time since a major update.
Basically I'd say that we're probably just approaching the point of being able to observe the effective member join rate. There probably won't be trustworthy data on that particular point until the new restrictions have been in place for enough time to generate sufficient points (so ~6-8 months probably), and when it's been that long I'd be willing to be that what it will look like will be fairly stable at lowish levels. Sort of what it looked like before the 2012 spike and at the moment, except with a bit fewer members; the curve on that graph is somewhat misleading, given that the spike is much sharper and either temporal end much flatter than it indicates.
Those are some sexy graphs.Someone who shares my wierd fetish!
*prods*
*Clears throat*
AHHHHHHHHHHHHHHHHHHHHHH FINALS
AHHHHHHHHHHHHHHHHHHHHHH NO UPDATE UNTIL AFTER
AHHHHHHHHHHHHHHHHHHHHHH
Or Sunday, depending on whether the campus library is open
Yep, that's the normal trend. It really throws a wrench into the gears of my prediction algorithms. I should probably try and exclude them from the data set.I'unno, once we've got enough points to reliably predict new versions it becomes less a problem and more a curious data feature. It wouldn't do to scrub data just because what we do see interferes with what we want to see.
I'm almost waiting for someone to come in and turn this into a learning AI. Then we're boned. Hard.We won't die in a hail of nuclear fire, we'll die from heart attacks from too many cool graphs.
Also: Updoot plox?
It should be easier to just use the release log, instead of the post count to make predictions. I'll try and run up something to figure that out.Yep, that's the normal trend. It really throws a wrench into the gears of my prediction algorithms. I should probably try and exclude them from the data set.I'unno, once we've got enough points to reliably predict new versions it becomes less a problem and more a curious data feature. It wouldn't do to scrub data just because what we do see interferes with what we want to see.
Especially because we can look at stuff beyond new users to get a good idea of how many of those new users stick around.While I would love to figure out that, what should I use as a metric? The closest I can think is Users Online/Day. Maybe comparing the Posts/Day with the UO/D would give me a picture of that.
By the by, your spreadsheets are all down.Really? Huh. Try this. (https://docs.google.com/spreadsheets/d/1o7rk4oGTGNEJ4wD2NJJvjpg-da6s3nB90wPOiv4anNQ/edit?usp=sharing)
I think it's more that SMF might just be unable to count.That's just what they want you to think.
Wonder if subboards, Armok 1 and the various nonsense would account for the discrepancy.[srs]I'm not seeing all of the subboards adding up to 3 million, but if you could check, I'd appreciate it.[/srs]
snipI'll take a crack at it in a bit.
tfw not in the top 1%.Good luck, dude.
Gotta start shitposting.
We are the 1%. Tremble before our shitposts.And/or forum games.
How long ago did ElfCollaborator bypass GreatWyrmGold?
I appear to be halfway through the list there.Basically, but the list is thousands of people long, so it's hard to make sweeping statements about them.
Are most of the people not on the list people who just posted a few things and then never came back?
Aaaaa ha ha ha ha no.
Not gonna happen, I like posts with content in.
Aaaaa ha ha ha ha no.Dude you need to step up your shitposting game
Not gonna happen, I like posts with content in.
You say that as if FFRP wasn't one of the several points of the pinnacle of shitposting. :PWe are the 1%. Tremble before our shitposts.And/or forum games.
Indeed.You say that as if FG&RP wasn't one of the several points of the pinnacle of shitposting. :PWe are the 1%. Tremble before our shitposts.And/or forum games.
I already PTW'd there.
I like posts with content in.
Indeed.
Haaa ha ha ha.
I didn't say how much content.
Posts: 15710 (70.134 per day)Holy shit, why. What do you even do besides Bay12?
Hello
my name is
TheBiggerFish
and I
never
inflate my
postcounts.
/me LOLs dismissively. All posts made are real posts.Hellomy name isTheBiggerFishand Ineverinflate mypostcounts.
The list is not live. Check the Members tab for the actual live list. You're still #30, MZ.Ok. I'm okay with that, I guess. I've been the gatekeeper to the first page for a while now.
however, it's unrealistic: you can only make a post every fifteen seconds/me LOLs dismissively. All posts made are real posts.Hellomy name isTheBiggerFishand Ineverinflate mypostcounts.
Nice trick with the timestamps though.
however, it's unrealistic: you can only make a post every fifteen seconds/me LOLs dismissively. All posts made are real posts.Hellomy name isTheBiggerFishand Ineverinflate mypostcounts.
Nice trick with the timestamps though.
It occurs to me that the 15-second limit sets the maximum number of posts per day per person to 5760.
Indeed.
All posts made are real posts.see this is why people complain
Indeed.All posts made are real posts.see this is why people complain
Now I'm waiting for their to be an unofficial thread where everyone has a bot auto-posting every 15 seconds exactly :P
Yes, probably.Now I'm waiting for their to be an unofficial thread where everyone has a bot auto-posting every 15 seconds exactly :P
Sourced from this tool? (http://www.bay12forums.com/smf/index.php?topic=155912.0)
After almost two years, I've come to a difficult conclusion. Providing this service is kind of tedious, frankly. Recently, I've just lost interest. I don't think I can maintain this any longer. Lots of people enjoy this, I know, but dropping it is what's best for me.
Feel free to run this in my stead, just shoot me a PM. Otherwise, let it languish, it's what's best. Obviously, it's not easy putting this down; it's occupied the last two years of my life. Lately, I've been reading Descartes, and I thought that this quote really resonated with me. "Sometimes, you have to read the first letter of every sentence."
Spoiler (click to show/hide)
Personally, I'm only surprised you didn't stop sooner.It's still April 1st :)
It is curious, actually. It doesn't seem to match up with the release of 0.42.06 at all. Anyone have any ideas?If it's still going on, you might be able to get a clue by looking through the member list—you could see if they're upper-boarders, or lower-boarders, or spambots, or whatever.
Is it just that some activity got pulled forward by less than a day? Any indication that posting patterns around midnight were odd?IDK. All I know is that forum time is UTC-6.
Updated. (https://docs.google.com/spreadsheets/d/1o7rk4oGTGNEJ4wD2NJJvjpg-da6s3nB90wPOiv4anNQ/edit?usp=sharing)Right now your algorithm is averaging the past N daily differences to predict today's daily difference. I suggest adding a couple low-hanging-fruit improvements to see if they help. Statisticians call these control variables and data scientists call it feature augmentation.
I've also started to poke around with the prediction algorithms again. Doing a field test today: the algorithm suggests that there should be between 1400 and 1746 posts today. The low number of posts on the 29th is weighting the data weirdly, but it's actually not affecting the prediction too much.
I've run the algorithm over the data for about a year and a half, and its margin of error is about +/- 225 posts, and an average of +/- 11% off. That's not really great yet. I'm still working on refining it; I'd really like to see the margin get down to +/- 100 posts.
Well, i might, for laughs, try and use a neural network to predict the contents of posts
Dear god, I don't even want to hear about that thing in action.Depends on the source. Mafia posts would be much different from rp posts or emotion thread posts. FotF posts would probably be weird but interesing.
Sounds even better, and indont have to r program my network!Well, i might, for laughs, try and use a neural network to predict the contents of posts
Just feed everything in a relevant thread/board into the shitpost simulator and you should get some fairly accurate results, with some proofreading.
Posts from the politics threads, however, will be indistinguishable.Dear god, I don't even want to hear about that thing in action.Depends on the source. Mafia posts would be much different from rp posts or emotion thread posts. FotF posts would probably be weird but interesing.
I bow down to you, master of colors and wizardsPosts from the politics threads, however, will be indistinguishable.Dear god, I don't even want to hear about that thing in action.Depends on the source. Mafia posts would be much different from rp posts or emotion thread posts. FotF posts would probably be weird but interesing.
You have no power here, Citrus Mage! This is my realm, where I am like God!nur nur nurSpoiler (click to show/hide)
Well, i might, for laughs, try and use a neural network to predict the contents of posts
You can't just 'done that' to us and not give us an example.Well, i might, for laughs, try and use a neural network to predict the contents of posts
Been there, done that, regrettably don't have the hardware to make a really stronk net. But I do have the software to!
You can't just 'done that' to us and not give us an example.Well, i might, for laughs, try and use a neural network to predict the contents of posts
Been there, done that, regrettably don't have the hardware to make a really stronk net. But I do have the software to!
700 posts or so to go until I make top ten posters. Whoo-ee.If you hurry, you'll have made it there within one year of joining.
700 posts or so to go until I make top ten posters. Whoo-ee.If you hurry, you'll have made it there within one year of joining.
That would be it, yes.From the looks of it, an average of 83 posts per day.700 posts or so to go until I make top ten posters. Whoo-ee.If you hurry, you'll have made it there within one year of joining.
how the fuck...
Not anywhere near 2000...It does help when you reply to every single thing that anyone ever says to you :)
Well yeah, but that's what one does, is it not?Sometimes.
I suppose there are times when one wouldn't want to reply.Well yeah, but that's what one does, is it not?Sometimes.
Not anywhere near 2000...Depends on whether you count GWG (http://www.bay12forums.com/smf/index.php?action=stats) or not (http://www.bay12forums.com/smf/index.php?action=mlist;sort=posts;start=0;desc).
Total Posts: 3694444This must have numerological significance...
I just noticed that the Bay 12 board (the one that's mostly monthly announcements) reports 49 posts in 100 topics. This bothers me :(Uhhhh....Yeah, that's very strange.
- Go to the Statistics Center (http://www.bay12forums.com/smf/index.php?action=stats) after the end of the month in forum time. A new month should have started by now.
- Scroll down to Forum History.
- Copy the Summary for the month you want to input. This is the "[month] 2016" line, followed by the totals for the month.
- Open the Spreadsheet (https://docs.google.com/spreadsheets/d/1o7rk4oGTGNEJ4wD2NJJvjpg-da6s3nB90wPOiv4anNQ/edit#gid=1222546767).
- Open the Monthly Tab, and scroll down to the bottom.
- Paste the data you just copied from the Statistics Center at the end of the data.
- Open the Daily Tab, and scroll down to the bottom: ~2900
- Expand the monthly summary in the Statistics Center.
- Copy all of the data for the individual days.
- Paste that data at the end of the data in the Daily Tab.
???Profit!- The charts should have automatically updated! Thank you for your help!
Updated. Sorry about the lateness!That's cool!
The update definitely doesn't have anything to do with the fact that I'm using it in a job application. Definitely not. Nooooooope. Nothing. Nada.
ohgodihopethisworks
How steady are the lower boards, and do long-term trends in each board match?You're asking questions about data that I can't access. The data is stored as a sum of all boards, I'm afraid. There's no breakdown by board, as much as I wish there was.
I believe it might have something to do with the rise in Discord use for idle chatter in FG&RP.Who's using discord for fgrp chatter?
I believe it might have something to do with the rise in Discord use for idle chatter in FG&RP.Who's using discord for fgrp chatter?
Also, is it *a* discord for fgrp, or several little ones for individual rps?Lots of the prolific posters, and several little ones.
I believe it might have something to do with the rise in Discord use for idle chatter in FG&RP.That's an interesting hypothesis. I'll have to look into it.
I'm happy to find this tread, I have been doing the same stats about the data for something like 2 years :)That's awesome! Can I look at your data, to compare our methodologies?
It has more sense when you add a 365-day trend (less "accidents" like new versions or special events).
Perhaps you should add the donation data (published, each month, by Toady), that's (I think) really interesting, too.
There is a quite strong correlation between donations and activity (in general) until patreon came. After that, activity fell, and donations progressed.
Hahaha I wonder if you can correlate ban streaks with world eventsAlmost certainly. I just don't really want to go there, ya know? That gets me a little too close to the action, which I like to stay far away from.
That's awesome! Can I look at your data, to compare our methodologies?
I'll look into including that too! Thank you!
Hahaha I wonder if you can correlate ban streaks with world eventsWell, now's an excellent time to test this hypothesis. I didn't know BFEL very well, but I'm going to guess he was banned thanks to something to do with the inauguration?
I woudn't be surprised if a significant percent of recent bans are ameripol-related.
Xeron November 9, 2016 Ban Trolling
Codician November 9, 2016 Ban Trolling
Vendayn November 11, 2016 Ban Trolling
Posts = 2506.71 + 2.18*Days since last update + -283.18*February + -576.60*March + -615.83*April + -676.63*May + -762.90*June + -928.52*July + -1077.16*August + -1415.67*September + -1341.61*October + -1552.25*November + -1115.57*December + 120.98*Monday + -101.43*Saturday + -11.37*Sunday + 65.69*Thursday + 121.15*Tuesday + 96.09*Wednesday
R2 = .5074
All variable except Sunday and Thursday are statistically significant at the 95% level.
Yes, there was a spambot in GD on that date.
I guess that answers my unasked about whether or not deleted posts count towards the total. It seems that they do.Has the Mystery of December Eighth finally been solved?!