Leafchan just died to GPT menace

So, probably the most popular anonymous board is so predictable in its views, opinions and replies, it has been successfully turned into a bot. Generated replies are barely distinguishable from the real ones. That's a relatively small model trained on just 24 gigs of data collected over 3 years of research on /pol/. The fact these datasets exist means that imageboards are under constant surveillance of researchers and other entities who take Anonymous for a lab rat giving a perfect case to study its behavior. Chances are, soon major boards will suffer from bots using this model. Social media or the intelligence most likely already use their own, much more expensive GPT models. https://github.com/yk/gpt-4chan What do you think? Is the only way left for us to communicate is through using forms of abstract media?

given that 4ch post is already on par with bots in all honesty the bots would prob make more since.

Just goes to show how much of \pol\ is bots. We've known this on \pol\ for years now. It's stupid how people sperg out about \pol\ thinking the bots and shilz are real people, only boomerz and retards actually post on \pol\ while the original userbase has moved on. >pleaze make a better post /sed/z/s/

>>59237 That's actually pretty scary. You think you're with likeminded people but you're actually just surrounded by bots. I guess it's a nice way to insulate "problematic" types. I wonder what the full ramifications are of this.

I think it's a good thing. Talk like a bot, get imitated by one. Now they have an incentive to not spew out inane bullshit. I doubt it'll make a difference, but it's funny to see it backfire on them.

given how frustrating channel four's captchae remain, its my assumption quite a few of the people there arent bots. I really mostly browse paul for news not for commentary or discussion. If you can get me a bot that produces up to date paul level coverage of arbitrary current events then I'll conceed that the site is obsolete. >go away spammer go away jannybot

>>59248 >my assumption quite a few of the people there arent bots that being said, you don't adopt captchae like that unless you have a major bot problem already.

>>59237 I'm not surprised, especially knowing where the data's from. The size is what I found interesting. I always assumed it would take hundreds of gigabytes for something like this to work, but this can be carried around on a cheap stick. I'd assume something similar for Twitter could do with even less given the character limit there.

It looks like the github page is 404'd. Got a mirror? >just >soon It's not new. The archives make it easy to grab all of the text, so it's an easy corpus to access. You don't have to do much processing to it, besides maybe for post numbers. 4chan's userbase has a lot of people who have interest in spamming bot posts. Combine that and you get tons of bots over the years. It used to just be simple spam bots and markov chain bots, but gpt2 bots were used for a while and now gpt3 bots. >>59246 >I think it's a good thing. Talk like a bot, get imitated by one. The problem with this thought process is that it's stuck in some decade old assumption that bots talk in a certain way. Sure, old markov chain bots and the first few generations of AI text bots were very easy to detect and had a clear bottiness to their output, but that is not the case any more. The AIs are only going to get better from this point, too. Modern bots talk like humans. >>59249 In fairness, 4chan added captcha years ago to combat stuff like the hta malware, flooding, and spam bots, not sophisticated text bots.

Cant say I'm really surprised. It's border line impossible to have discussion there anymore. when it's not blindly praising something it's just spamming the same insults over and over again.

So AI can generate pictures and text posts. AI can even solve captchas. What's the point? It's not going to lower the quality of the discussion on \pol\, and it's going to stand out on any forum worth visiting.

>>59241 >I wonder what the full ramifications are of this. One I can think of is bots egging someone on to do something dangerous.

>>59237 link is dead, got another?

>>59257 Yeah, I did think it would be sad if someone was driven to suicide, not even by an actual person, but by a bot.

>>59237 This isn't a problem for the only board still worth visiting over there, /mlp/. There, I have discussions with other people, and this is proven by the art, stories, music, and other things we create. If a machine can create, which they can, they'll just be used to generate pony porn more effectively, which they are. Creative communities avoid these issues.

>>59251 >The AIs are only going to get better from this point If a bot can produce posts that are interesting(to me at least) and on topic, I wouldn't mind. Most 4chan posts don't meet that criteria. I'd prefer an interesting bot to a boring person. Image boards aren't primarily for socializing in my opinion.

>The fact these datasets exist means that imageboards are under constant surveillance of researchers and other entities who take Anonymous for a lab rat giving a perfect case to study its behavior Uh, was there ever any doubt about that? Bots are pretty easy to detect though. They usually do not engage meaningfully with what anyone else says. Not responding to direct questions or answering them in a way that does not make sense is a big one. Yeah a lot of humans get caught in these kinds of bot tests but if they're posting so low quality they're indistinguishable from a bot they might as well be one.

>>59265 >Not responding to direct questions or answering them in a way that does not make sense is a big one hate to burst your bubble but have you tried interacting on a big, fast moving, 4chan thread ? people frequently don't reply to questions addressed to them, or do it in nonsensical fashion. Or, huh, maybe these are bots already... The only issue I see with this GPToy is that it has no understanding of quote numbers.

>>59265 >Uh, was there ever any doubt about that? Yeah, I don't see why OP is presenting that as if it hasn't been a given for 15 years at least. The guy who runs textfiles even has an archive of 10 million threads from 2006-2008 https://archive.org/details/4chan_threads_archive_10_billion There are a handful of academic research papers about 4chan culture and many more less formal documentation of it, from sociologists, security people, and AI researchers. >Bots are pretty easy to detect though. They usually do not engage meaningfully with what anyone else says. Not responding to direct questions or answering them in a way that does not make sense is a big one Some of them, sure, but AI is great at this now.

>>59251 looks like this is it https://github.com/tornikeo/gpt-4chan

>>59237 >What do you think? Is the only way left for us to communicate is through using forms of abstract media? I am advocating for the revival of 1337 speak. The bots were announced on an image somewehre that they funded some soykaf to do it and deter imageboard users "by generating garbage". If you didn't notice, this happened around early 2019, 2014-2018 was by far the peak imageboard age, back when things were more centralized. 4chan had been botted/GPT'd for a long time and the bots even used captcha solver providers. It was quite noticeable that the PPH plummeted after the Gcaptcha was removed by hiro (thus preventing the automated bots from posting) and had the lowest traffick on \pol\ and other boards like \b\, meanwhile other boards' traffic remained the same. Also, most GPT projects and AI projects out tehre were trained using imageboard data, this is why they eventually become racist like Tay AI, etc. since there's no other place to get a stream of conversation legally aside from public chatrooms which doesn't exist so they used imageboards. I have noticed posting quality had worsennned around 2019, on imageboards and users have also left out of fear being associated with another "incident" or "leak" while some countries derezzed chans after the christchurch shootings. >>59249 There are many captcha farms, which make people solve captcha per minute in exchange of a few cents. I'm guessing they're either for scraping or botting but 4chan had a botting problem, even worse than twitter (which Elon bought to solve the AI/botting issue too). He knows, maybe one day he will buy 4chan too.

>>59286 >2014-2018 was by far the peak imageboard age, In what sense? >back when things were more centralized Even then, there were smaller chans, which had far better discussion than 4chan.

>>59287 Different Lain, but I'm sure he is referring less to peak quality and probably peak adoption. I know most people joined in, or started using Chans more about that time. Even my friends who lurked for years started spending more time on these sites about that time. And yeah, there have been more chans popping up everywhere, but those are for more niche users. normal people are leaving social media and some are picking up 4chan or reddit.

>>59286 >It was quite noticeable that the PPH plummeted after the Gcaptcha was removed by hiro (thus preventing the automated bots from posting) and had the lowest traffick on \pol\ and other boards like \b\, meanwhile other boards' traffic remained the same. Are there stats for this somewhere? I stopped using 4chan years before the captcha change and can't find a date for when the new one was added. Was hoping to cross-reference it with archives from 4stats to confirm this.

>>59289 From what I can find, the new captcha was implemented in early July of 2021. It does look like there was a brief decrease in posting at that time according to 4pleb's archive of the politics board, but it wasn't a large percentage. It could also be explained by the new captcha confusing users. In searching for the date, I found complaints about it, accounts of people who were confused and didn't understand how to solve it (slidy bar hard). https://archive.4plebs.org/[the politics board]/statistics/activity/ (replace the [the politics board], can't post the rules directly because of spam filters) I don't doubt that people are using bots to post on there, but this data is not indicative of a massive bot presence affected by the captcha change.

In the immediate term, whether the site is really overrun by robots or just human beings so desperate to fit in that they make posts about as interesting as a fleet of robots, the value of the site for any user with an IQ above room temperature in Fahrenheit remains the same. Of course the long term implications are more dire. Somewhat amusing that this thread, about this topic, already has several Lains complaining about the resident jannybot trying to block them, as if they might be bots themselves.

also, even if PPH dropped by a clear amount its wrong to neccisarily removed bots. I know a few occasional posters there who scaled back of totally stopped with the new captchas cause it was just too much of a pain for what its worth.

The biggest tell from bots, is that they are wholly inorganic posters. They're just memes and can only imitate and not create. When bots do post, they spout non-sequiturs and have to 'force' their way in. This can easily be sniffed out if you take a moment or two to read a post. It becomes problematic when there's entire botnets swarming a place. This is the case with the most popular chan and specifically 4pol has been under assault for longer than I can remember. I have noticed their presence here as well, as they try to infiltrate the boards with their low quality threads and sub-tier garbage 'human-talk'. The purpose of this, is to hijack the 'discourse' of the board, so that new people take what they see as the norm, who then imitate behaviour making them almost indistinguishable from bots. >What do you think? Is the only way left for us to communicate is through using forms of abstract media? Perhaps. What I personally consider a good litmus test, is their chatting capabilities. Bots can never chat convincingly. Perhaps this is because of their inability to be quality bullshitters.

>>59297 Your description of obvious bot behavior is indistinguishable from that of thousands of 4tards. I wish I was making a joke here but this is true.

>>59298 >x Challenge not accepted.

>>59298 >>59297 I hate to say "tards" but yeah, its kinda true. You dont find any honest discussion on that site anymore. Just porn, softcore porn, and memes. And half the memes are reposts from other sites. Its becoming a characterized version of itself. Speaking of bots, check out any general thread from /vg/. Ritualposting is strong. A thread there looks identical to the one they came before it. And it was known for years that you couldn't start a Guild Wars 2 thread on /VG/ cause some guy made a bot to fill the thread with comments until it hit its limit. So bots have been a known issue for years.

>>59300 Why not point out what's factually wrong in what I said instead of taking the moment to continue soykafting on 4chan? Oh right,..

>>59301 Are you upset that I agreed with you or something? I replied to two people, so I dont even know which post you are upset about me replying too. Are you a bot?

>>59301 I did though. What are you missing? We were talking about 4chan, you were saying how easy it is to spot "bots" and I pointed out that the average 4channer posts like one already. For that matter, many of them even pride themselves on the repetitive nature of their contributions.. QED, it is not so easy to separate them just by reading.

>>59301 Not that lain but it looks they're saying that the attributes you've associated with bots are also just attributes that are associated with actual 4chan users. Non-sequiturs, memes, imitated behavior, and unconvincing chat style are all classic 4chan culture that predates sophisticated bot farms. It makes sense that they would give specific details about the low quality posting habits of actual people on 4chan to clarify. I think perhaps you read their criticism as if they were implying you sounded like a bot? That's how I read it at first. One of the goals of bot farms is to lower the level of discourse like you said, but for 4chan that's not really a concern. It's already very low. Just look at old archives. Another goal of bot farms is to make people doubt they're talking to real people even when they are. I think this thread is proof that's successful. Another goal of bot farms is to use low quality, detectable bots so that people can get a false confidence in their ability to recognize bots. Then high quality bots that are not distinguishable are used and are more convincing. >can only imitate, not create bots can create. It's particularly not difficult to create images that meet 4chan quality. They don't really need to, though, since only a small amount of 4channers produce OC. >I have noticed their presence here as well, as they try to infiltrate the boards with their low quality threads and sub-tier garbage 'human-talk'. I have also seen posts here that I believe are bot posts, but I also suspect most of the low quality garbage that gets posted is just from 4chan users who haven't yet learned that not all imageboards have the same brewing soykaf/trolling culture.

>>59304 >make people doubt they're talking to real people even when they are >use low quality, detectable bots so that people can get a false confidence in their ability to recognize bots This is how undercover police and spy agencies strategize in the flesh and blood world when attempting to infiltrate a target group or simply blend in with a more general subculture long term. Convince people that anyone else is the agent except them. Even police "ghost cars" are camouflaged in a similar way. You've probably known someone who swears he can spot any unmarked car. The obvious cars are SO VERY obvious. The ones the ones they don't want you to spot? You probably won't.

>>59268 >Some of them, sure, but AI is great at this now. Not from the various GPT demos I've seen

>>59304 >One of the goals of bot farms is to lower the level of discourse like you said, but for 4chan that's not really a concern. It's already very low. Just look at old archives. Yeah, I know what you mean. But I actually had few times on 4chan where I went into discussion and I didn't try to act like a dick. I tried to be reasonable and try to talk to someone like a human and we began having a calm and friendly conversation. You could notice it wasn't bot since the conversation stood out like a sore thumb and we actually engaged each other. Although in the mean time the rest of the thread was spammed by low effort hateful replies. I am not sure if these are bots or not. I think a lot of it actually is just deranged people. Which too is quite sad. Other funny thing though, and it's something I can't really give proof of because to be there you have to keep it semi private, is that there are actually whole discord groups dedicated to spamming low effort bait. It basically is just making a thread with a provoking title and some picture of a porn crop or something from what I've seen. Although that doesn't really justify paranoia. It's best to ignore people like that. Not that I don't even like that kind of humor since I have a soft spot for dumb brewing soykaf, but sometimes it's just too much and I feel it lost it's charm.

>>59304 >One of the goals of bot farms is to lower the level of discourse like you said, but for 4chan that's not really a concern. It's already very low. Just look at old archives. >Another goal of bot farms is to make people doubt they're talking to real people even when they are. I think this thread is proof that's successful. >Another goal of bot farms is to use low quality, detectable bots so that people can get a false confidence in their ability to recognize bots. Then high quality bots that are not distinguishable are used and are more convincing. games within games

>just at least 6 years to late, OP

So what does this mean for the future of alt-chans? If captcha can be worked around, then do we have to make a new system? Hey, what happened to the idea of Linked Intranets? I was told YEARS ago that it was the future, and right about now I want it

>>59297 It is a very hard issue especially for anonymous image boards since the lack of accounts makes it harder to track who is real and who is a bot. For now the best solution would likely be more difficult captchas. The truth is that most anonymous imageboards on the clearnet will likely be over run, the idea of linked intranets sounds good, but with the increasing prevalence of these bots and bot programs, malicious actors would likely still sneak in. The big issue with most solutions to the problems are that they kill traffic or kill anonymity. Anyone deeper in the field have any novel solutions?

>>59305 >Even police "ghost cars" are camouflaged in a similar way. You've probably known someone who swears he can spot any unmarked car. The obvious cars are SO VERY obvious. The ones the ones they don't want you to spot? You probably won't. This is a perfect analogy and also a huge pet peeve for me. It's fuarrrking maddening hearing "I can tell from the headlights, trust me." >>59307 >Other funny thing though, and it's something I can't really give proof of because to be there you have to keep it semi private, is that there are actually whole discord groups dedicated to spamming low effort bait. I believe you, I remember irc servers long ago dedicated to it. I've seen funny cycles where people spam bait, other people think the bait is actual people who are upset and then spam bait against it, both groups think they're trolling others but really they're just trolls all around and nobody really cares. People seem to attribute more malice to it now. it seems like some people are getting caught up in it and believe actual agendas are being pushed. And of course, it's funny that some people think that a thread on 4chan is agenda, so trolls, or just people having fun, will also pretend to believe that and make fake proof of it, too, or make chats where they larp as agenda pushers to feed that too. It goes full circle.

Text bots are really bad at figuring out rigid, mechanical rules. Therefore one option is to force every post to follow certain rules of composition which are never announced. Humans (at least, moderately intelligent ones) can figure out quickly that every post has to end with the word 'fnord' but bots will tend to try fitting it into their language model and put it in sentences. Of course, any serious glowtard would catch on pretty quick and adapt their program, but you could change the rules periodically and make an example announcement post to clue people in what the rules might be. Therefore the text bots would require constant supervision. If you make the rules complicated enough it would actually be fun in itself to figure out what they are well enough to soykafpost fnord.

>>59318 >>59326 Why not have anonymous accounts? So like a forum, without usernames. And instead of using emails to make an account, you'd have to generate one by solving a computationally heavy problem. 8chan.moe did something similar once to stop a spam campaign. The maximum number of accounts could also be limited.

>>59327 because then anyone with enough money can create more accounts than anyone else and flood the board.

>>59302 My apologies, I misread what you said.. yesterday was a bad day for me since my uncle passed away and I had to say my goodbyes. I was drunk and if it's any consternation: I was angry at everyone. To reply to your post: fuu-chan is a shadow of its former glory. It's hard for me to find anything worth of value in there, safe for the Durham threads and some tips on /ck/ and /out/. It's such a waste and I am afraid this is what has happened on 4ch, is in store for all anonymized image boards in the coming years. And no, i'm not a bot .. just a very terrified someone who hates what the Internet has become. It's coming to the point that I am so paranoid that I do no want to post anything online. I hate the idea that every word I type is processed by computers and analyzed how it fits in my general (online) behaviour pattern, with the purpose to understand and rule over me. It's no fun for me to be here anymore when I constantly have to weigh every word and I am seriously considering 'unplugging' from the Net (not that it really matters since there's camera's and smart devices everywhere which can keep an eye on me..) I am very afraid, lain.. just so very afraid of what's to come. >>59303 Pride is my capital sin and I did not like to be called out for my hypocrisy,

>>59328 And then they'd be derezzed. If someone has enough money, there's literally no full proof way of keeping them out.

I bet it can't keep up with current news.

>>59327 Since literally no-one has asked, I'll post a potential solution to the bannable anonymous accounts problem in the hopes someone smarter than me can actually prove/implement it. Associate each user post with a unique asymmetric key pair and divide posts into "good", "bad" and "invalid" sets. In order to submit a new post, a user must provide a ring signature over the set of "good" posts - anonymously proving that at least one of those posts was made by them. This scheme by itself isn't particularly compelling, since a user can simply rely on a single "good" post. However, by restricting the maximum ring size and/or invalidating posts according to a certain criterion or rolling window, correlation attacks are made possible. If the correlation is something that the user can estimate, the user is forced to decide between continued "bad" posting and their anonymity; the only way to decrease correlation is to create "good" posts to enlarge their anonymity set. A user with no valid posts or only posts with an undesirably high correlation from re-use is effectively banished without having been de-anonymized. Users who don't care about their anonymity and continue spamming anyways will end up with extremely high correlation which can then be used to automatically invalidate/remove their posts. This scheme can be extended by requiring an additional signature over the user accounts (asymmetric keypairs, again) in order to prevent people sharing their posts' private keys. This is because the private key associated with the account can be used to de-anonymize all of the user's previous signatures (in some schemes), so sharing would require a higher level of trust. Potentially, there could be some way to create a hierarchy of linkable derivative keys so that members can properly invite new users, but be liable for the spam their gifted keys generate. If this is possible, it could be used to curtail rich users/glowies since they would need to continuously obtain top-level keys from the admins.

>>59329 I miss having a place I could go to have fun. Guess I'll have to find or make a new one. Lain is a good place for now

>>59329 >My apologies, I misread what you said.. yesterday was a bad day for me since my uncle passed away and I had to say my goodbyes. I was drunk and if it's any consternation: I was angry at everyone. Sorry about your uncle lain, that is like kicking dead whales down the beach. >>59328 For a lot of use cases, that's fine. Alt chans for example usually aren't being spammed by wealthy actors, but by bored people who think it's funny. It also cuts out ad bots or stock/crypto grift bots if the amount to spend on posting isn't worth the return from it. It's not perfect and it lessens anonymity because every post is tied to the token, but it's not without some merit for some use cases. >>59340 This is a cool idea, but I have one criticism. Since the thread topic is bots, let's consider that propaganda bots don't just spam propaganda. Bots often post normal content to blend in to prevent detection or to attract people to them. For example, a lot of social media bots maintain normal pages for some interest and only occasionally insert propaganda posts. The obvious attack on this system is to just post enough good posts to safely make bad posts, which is doable with little effort.

>>59340 I hate it, but proof-of-work certainly works for this, and can prevent further spam. My solution is a token system. Paranoid users may get a new token for each message; others may not want to do this. Proof-of-work can be used to limit tokens. Tokens may develop reputations the owner may use to judge them. During bouts of spam, tokens may be withheld.

>>59237 >What do you think? Is the only way left for us to communicate is through using forms of abstract media? Create a site where only a kind of people post. You don't need to destroy anonymity for that, who you are and where you live is not important, but perhaps your knowledge or interest still need to be checked. Yes, it is like kicking dead whales down the beach, but unless you test every new user, it will become impossible to stop bots.

>>59340 My problem with this is who defines someone else's post as a "good", "bad" or "invalid"? If other users do that, then it's just a reddit with automatic shadow ban.

I think obscurity is being under rated as a defense mechanism, and I think ease of entry on entry level sites like number four is also undervalued. Personally, I don't think I'd be here on lain today if the captchas on the cloverleafchan were as bad as they are today. I visited the site on a lark, and because it was so easy to post I figured might as well. Fancy things with key handling are well and good, but they make it less interesting to new users just sort of checking the space out. Having a big and accessible chan like Mr Mo0ts used to be is good for places like lain because it brings people in, weakly vets them, and passes them on to more obscure places. To this end a gatewaychan needs to be A) accessible, easy to join, and have some info that leads to obscure places, B) vaugely interesting and fun, keeping people around long enough and with some real discussion, and C) miserable and unpleasant so people we don't want leave the system, and people we do want end up leaving gatewaychan. Lain of course isnt super obscure, and we get our share of spam and CP and other trash, but I do think this sort of tiered system that exists between 2^2 and previously 2^3 chan and lain/other altchans isnt as bad as it people here sometimes are making it out to be. Not every site needs to be a perfect complete system unto itself.

>>59347 >The obvious attack on this system is to just post enough good posts to safely make bad posts Yeah, this is a tough one since in order to solve it, you need to be able to distinguish honest-but-fallible users from these deceptive ones. A soykafty patch would be that there are degrees of "heat" that's added to the posts used in the offending post's ring signature, and insidious soykaf gets more heat than regular rulebreaking. But since this punishes everyone, a high heat factor risks a small group of abusers invalidating too many posts belonging to other people, grinding the whole system to a halt. I guess the best we can hope for is AI to flag these posts and then demote them like masterchan did - but that's going further into reddit territory. >>59353 Proof of work is good, but you need a way to make sure the tokens are fungible and anonymous. If the tokens are generated via JS, for instance, then the website can presumably track your history on it to associate with the tokens. If it's done with blockchains, then you need to worry about 51% attacks, too. Also, there's the issue that an autist can hoard tokens in preparation for an unholy campaign of brewing soykaf. PoW is definitely good, but I'm skeptical of it as a complete solution. Then again, any solution solution -partial or not- would be welcome at this point; it's only a matter of time before GPT gets good enough to attack higher quality boards or US/EU pushes imageboards completely to the darknet. >>59355 Posts are assumed good and deemed bad by whatever moderation system is already in place e.g. a small team like Lainchan or some consensus protocol for a decentralized imageboard. Invalid is just when they can't be used to make a ring signature because of age or whatever criterion is used for restricting the anonymity set. >shadow ban Not a shadow ban. fuarrrk shadow bans. They may be effective dealing with spammers, but being on the receiving end of one is pure hell. >>59356 >I think ease of entry on entry level sites like number four is also undervalued. Yeah, this is definitely a big issue. The scheme I came up with was originally an attempt to prove you could have a darknet board that could be moderated at leaf's scale. Ideally, you'd have the full trinity of access, anonymity, and liability. >Not every site needs to be a perfect complete system unto itself. Interesting point. Reminds me a little of iro2.

>Is the only way left for us to communicate is through using forms of abstract media Nope. Look up DALL-E 2. There is no longer a way to be sure that someone is real and not a bot. Chains of trust will have to be established in real life by exchanging GPG signatures with the other person face to face.

>>59374 on chan type sites this is laughably absurd. maybe part of the fun is that on the internet no one knows your a dog, or a bot. What are you, racist against algorithms ?

Another important thing here: archives were a mistake, things should be ephemeral

>>59375 Right, I'm not saying that we should establish imageboards in this way. I'm saying that any communication platform that isn't based on real life connections should be assumed to be nothing but bots by default.

>>59377 fair fair, can't so easily disagree with that.

>>59376 Maybe, but how do you stop archives? I don't think there's a technical way to do it.

>>59385 In theory? No. But in practice? You just have to follow CloudFlare's example and block any browser that isn't the most mainstream of mainstream. Don't allow users to browse 24/7 or see too many threads, etc. Also other user-hostile patterns like broken search, un-guessable resource IDs, etc. This will bog down 99% of archiving efforts and ensure a significant amount of content is never archived. Of course, this presumes a large/active enough website to make automated archiving is necessary. That said, truly ephemeral is bullshit. Any forum with constructive discussion should be preserved to help further the proliferation of human knowledge. I've found lots of hidden treasures by trawling the 4chan archives that modern users have seemingly forgotten.

>>59287 >In what sense? I've used to frequent imageboards, a lot of posts were mostly authentic and soyjak wasn't a thing back then (it used to be only feel/doom variant being posted). Back then, you can go on any chan and there will be few exchanges here and there, sometimes debates, but mostly it's an exchange of knowledge, you'd learn stuff. Fast-forward a few years around late 2018-2019 there was an influx of users who just spam, or just plainly call anything "glitterboy" and attach soyjak edits or even just quote your entire post and include an out-of-context soyjak when your post has nothing to do with it and even worse I tried a litmus test (posting a well-known posting format or glorious posts on a board even gets soyjak'd), it is the dark times of imageboard era or should I say the kali yuga, anons can (most of the time) no longer hold a proper conversation like it used to, everyone just tries to insult each other and spam the same thousand of the belligerent posters spamming soyjak, one says something too intelligent? or a very informative post? and it's actually a repost years ago? good luck as another poster would abuse the "FPBP" (first post best post) by posting something negative or calling out an actual decent post stupid just to manipulate how the next posters should react (honestly absurd, evidence of this "duality" in the archives), and not only that, there's also no way to tell if those posters were originally bots, maybe at first they were, the soyjak spammers and endless soylent-flinging resulting in \b\-type behaviour on even the slow blue boards and the fpbp hijackers, but then it is now known that anons have this tendency to just adapt and then become the soyjak posters themselves anyways (which means this is your vulnerability to detractors and bots): resulting in very low intelligence/informative postings on imageboards to the point that even youtube comments were even more informational, well lainchan was somewhat an exception since some posts here are as nostalgic/authentic as it can be. >>59288 Most people joined on in during those times and some chans were initialized post 2014, the posting quality was much higher, like if we're to measure the SNR (signal to noise ratio) there would be more posts that can be deemed "educational" or "interesting" or "OC" and compare it to the postings nowadays, too much noise and redundancy, and much upset. I was surprised how bad it was around 2020 since I kind of stopped browsing imageboards for a few 8 or so months so I noticed the drastic change quite hard. Back then anonymous would post something informative, it is a net positive to the collective, when someone disagrees with you, it is internalized as the anon who disagrees with me was just me in another perspective. Now when you disagree with another anon (on certain places), it ends up in soykaf-flinging a chain of disagreements no matter how hard you try to steer the conversation not for it to derail (point out the topic at hand) and it all ends in belligerent personal attacks (which proves they don't understand how this anonymous collective really works), the same topic that ends in learning more things (collectivism) ends up in a total waste of time in today's imageboard standards so there's without a doubt some kind of dumbening ops that steered the direction of the collective and possibly was done to just prevent the same collective from growing (in which sad case, they are VERY successful, in doing so).

>>59340 As the one you're responding to, this seems far too complicated. Moderation on image boards is cave-man tier, and forums are only slightly better. I think going from that, to a space-age Rube Goldberg machine, is overkill. What I suggested wasn't intended as a way to automate moderator's job, but rather make it somewhat, or a lot easier(especially on darknets). They'll still have to make a lot of judgment calls. The concern about anonymity isn't significant in my opinion. People should mind their op-sec. Anonymity on imageboards, is desirable to me because of its effects on communication more than anything else.

What would lain consider the mainstream social media site with the most bots/botnets of this kind specifically?

>>59413 For GPT and similar, my guess is Twitter. I don't recall the stats, but I've seen interviews with their anti-botting team where they claim to have a very high number of bots they ban per day and they seem to have the most focus from state actors and scammers. Instagram and YouTube have a lot of bots as well, but since they're mostly there to either artificially boost engagement for better algorithmic amplification or post stolen/generated content, they don't seem to try very hard. Usually just some preselected messages with some words swapped.

>>59367 How absolutely eerie. For the last few weeks I have been brainstorming this precise idea- a p2p distributed message board system using a highly-configurable reputation-based consensus protocol for moderation and ratelimits/PoW/other mechanisms, with a mixnet for sender anonymity of posts. >blockchains Trustless is actually not necessary here- quite the opposite, I believe that large centralized services absolutely squander the network level benefits/assumptions that proper utilization of social relationships can grant them, hence their spam issue. Would you like to discuss this in more detail? email me (yonyx@420blaze.it) or give me a matrix/xmpp or something :)

uknown if related - nanochan has also been dead for days due to ddos according to posts on phichan

>>59340 thought of something similar long ago: every board should enforce post ID this post ID is seen on the board, it's like your username but you're still anonymous although your posts can be interlinked - this is only for moderation/curation purposes, posters with higher posts associated with that ID are much more trusted than a new poster with barely any posts, this trusted poster is then curated and scored by the mod (usually the owner(s) of board), e.g, poster #a1b2c3d4 has 1% moderated posts and has not had any bans - this means they are trustable to make posts and further posts. since post IDs are public, there can be several third-party curators who can represent the same board but pick only their trusted or aligned posters, so you can set -politics flag and any post ID leaning into political posts or ones that had been moderated (violated rules) would be hidden by a third-party curators but not deleted enough for them to cry about "free speach", you can even set -pepe and there would be no pepe posters in your curated imageboard. the problem is how to implement this poster ID without the risk of people just sharing keys to their ID or how to implement an identifier without too much soykaf or without giving the poster the ability to modify/cheat out their ID (more of like machine ID/fingerprint but not too saucy).

>>59597 it's also possible to implement this verification system using captchas, which verify that the poster isn't botting, then these captcha points can be used for making posts, people with higher post records/history are much more trusted, people with higher post moderation/deletion or violation are untrusted and your imageboard client would not grab their post (kind of like blacklist).

>>59597 >it's like your username but you're still anonymous Unfortunately it took over a decade of image board experiments, but this has been revealed to be completely false. IDs have ended up harmful to one of the core advantages of anonymous discussion--letting an argument in a post stand on its own merits--by directing the focus of attention to all past posts made by an ID. However small it might be, IDs produce an identity in the thread or board they are used and end up harming the quality of discourse in a very similar manner to usernames in traditional forums.

>>59602 poster IDs have their place. If your board is overran by bad faith same users or bots, it can mitigate their negativity and improve discourse. If everyone is engaging in good faith, it's unneeded.

nothing wrong with a chan or two doing it. frankly sometimes its nice on fast moving paul threads when you're trying to keep a discussion to have (thread only) consistent ID's so you don't keep accusing people of being other people. I don't like the idea of pan board of pan site ids, but if a chan wanted to do it why not

>>59599 Now I'm almost convinced that the guy from HN who I discussed this with via email is in this thread. Perhaps we should all find a chatroom for discussing the idea further. Anyway, I'll post my more-developed idea now. Every post could be paired with an 'ID', which corresponds to or is a key from which a post originated, as well as a user-supplied name string, a 'visibility' weight (for moderation), and post contents/metadata (time, thread or reply, email/subject/name field, file metadata, etc). A 'container' is a list of subjective 'parameters', as well as lists of (at least some of) known IDs and their 'trust weights', and basic peer data. You may have multiple containers for different boards/forums/link-aggregators that have different trust structures. There are a lot of things that parameters could entail aside from reputation-related balances- bump limits, reply limits, post size limits, language enhancers, post time limits, categorization, and various other options that may be useful to different models of boards and forums. When I say reputation (interchangeably 'trust weight' or 'trust'), I refer to a scale ranging from -1.0 to +1.0, where weight below a threshold defined by a container's parameters warrants a post's subjective removal or alteration, and positive weight could translate to a higher reputation for the poster's ID, which is used in further calculations. Parameters could be set limiting and defining the extent to which this kind of reputation adjustment can happen organically/automatically. You may dislike Reddit's 'karma' style and prefer the more 'equal' style of imageboards and forums, so you configure a board such that only those with high, manually-set trust weights, like moderators, can increase or decrease the reputations of others. If you disagree with a particular moderators' judgements, you can simply choose to ignore their judgements by tinkering with your container. As an example, I will define some existing boards in terms of this mechanism: -------------------------------------------------------------------------------------------- HackerNews: - Admin-defined algorithm determines what threads and replies are visible in what order, taking user-generated activity of rating and commenting as input. - Handful of mods have total authority to kill posts and alter the contents of posts (in our system, they would be signing their changes) - Anyone over X reputation ('karma') can apply a limited amount of trust weight for or against a post, and can even flag a post to hopefully garner moderator attention. There may be some flags that are collectively attributed to a post for certain ranges of trust- for example, a client using this protocol may implement certain styleizations for posts that have certain flags, like how posts become literally greyer in hue with more downvotes on HN. - 'Showdead' is just an alteration of subjective parameters that has your client view posts within a certain nonzero trust range. -------------------------------------------------------------------------------------------- Chans: - There exist 'posts' which can be grouped into 'threads' and 'replies'. Posts contain data like post ID/sig, poster ID/sig, post time (not necessarily included in sig), name/email/subject, filenames, file bittorrent links, and text content. - There are some admins with maximum trust weights, and container parameters define that admins may adjust the trust levels of any ID (so they can appoint mods and janitors). Janitors and above can null the trust of a post, moderators and above can null the trust of a poster and can set various flags depending on the container parameters. - Language enhancers (language enhancers)- container defines strings that can be filtered or replaced.

>>59608 >>59608 Recall that when I talk of 'adjusting' trust levels or other values within such a non-central system, I simply mean that within a community most people are using the same or very similar containers, and so are obeying/enforcing mostly the same rules, but subjectively. To really determine what is seen, we have to put a combination of those weights and parameters to work. Several things must be addressed: -------------------------------------------------------------------------------------------- 1 How should reputation information of a post or poster be determined? - Reputation is determined according to how poster IDs rate other posts/posters, and according to parameters that determine the weight with which these judgements are weighed. Parameters determine bounds, minimums, maximums, and multipliers, for which the reputation of Alice affects how Alice's judgements of Bob and/or his posts are weighed in the eyes of other posters. -------------------------------------------------------------------------------------------- 2 Here are what those parameters for automatic curation would look like: - Posters can apply a trust weight between -1.0 and 1.0 to a post or poster ID. The reputation that a given judge 'Alice' possesses is used by the parameters to calculate how her judgement of Bob should be modulated. Reputation thresholds for different purposes are to be defined, like allowing post IDs above +/-X.X to be visible, and having different reputation ranges for which different flags or trust weights could be applied to a post or poster. - Parameters such that if someone has a reputation level below X.X (a hypothetical moderator trust level), their effect on poster IDs is 0.0 and should always be ignored, and that if their reputation is above that level they are allowed to have some level of weight, according to which range of reputation they are in. Example 1 (like some chans) -- [0.01, 1.0] can make posts -- [0.4, 1.0] can decrease the weight of a post enough to delete it themselves. -- [0.4, 1.0] can apply some kinds of flags to a post, which could be used by clients for various purposes like rendering, or for subjective alteration of some data/metadata/ -- [0.5, 1.0] can decrease the weight of any poster ID with lower weight than themselves enough to disallow them from submitting posts under that ID. Can mark threads as pinned, can change language enhancers, can change the MOTD of the board, can change style (however it is implemented) -- [1.0, 1.0] can apply any amount of weight upon any post or poster ID, has all abilities that exist in the container. Perhaps could tinker with some of these parameters of the container. Example 2 (sorta like HN) -- [0.001, 1.0] can make thread and reply posts, can positively weight a post by +0.001, which increases their poster ID weights by +0.001. Cannot increase a user ID's weight to +1.0. -- [0.5, 1.0] can negatively weight a post by -0.001 with no effect on user ID weight, and if the post goes below a threshold certain flags get applied to it (used for a more grey text on HN's styling), but this does not delete it. -- [1.0, 1.0] can weight any post ID or user ID with maximum power from -1.0 to 1.0, can edit posts, set any flags on posts, and determine post order. - For temp-bans, reputational judgement could have the option of an expiration date- there could be a parameter limit to how long judgement against a poster's ID can last for people with certain trust weights, so as to allow for the judgement from, say, lower trust levels to be weaker and not capable of lasting as long, whereas say, higher trust level users like moderators and admins could put in place permanent judgements against poster IDs. Parameters limiting judgements against mere posts rather than IDs may not be as necessary. -------------------------------------------------------------------------------------------- 3 How should posts reputation data be stored/accessed? - While information about how to retrieve posts is stored in a distributed hash table, posts themselves as well as the the reputation activity related to them are stored on machines of all users who have viewed them. If there is a resource issue, it may be necessary to reduce the amount of post contents saved on some machines, however I believe that most data about posts, even trust activity, are small enough to be handled by the vast majority of consumer machines in the vast majority of circumstances- in fact, I suspect that even storing the entirety of something as large as 4chan (minus multimedia, but even that is really small due to bump limits) would not be difficult for most drives. -------------------------------------------------------------------------------------------- 4 What enables a totally new ID with no reputation to have its posts seen? That is, how may an Anonymous ID receive a small quantity of trust to start off with? - For anonymous posts, Rate-limits could be enforced by the machines of all user IDs with a reputation above a given threshold. Some particular machines or some % of online trusted IDs' machines come to an automatic consensus about which of the last X posts within the last Y time interval are to be given a small amount of trust weight for free, as well as how to order them (making sure that their time values are legit). This limits the capacity for flooding via anonymous IDs. - The poster IDs behind an anonymous post could receive a slightly increased reputation that exempts them from rate-limits (or makes them more lax) if have made at least X posts that have not been explicitly deleted by moderation. - It may be necessary for CAPTCHA-generating software to be run on the machines of higher reputation users. Those users' machines could accept a request for a CAPTCHA, generate and sign a CAPTCHA, receive a signed response, and then apply a positive trust weighting to the ID which successfully solved the CAPTCHA. To prevent denial of service attacks via spamming CAPTCHA requests, the amount of CAPTCHA requests received and validated must not be much larger than the size of the rate limit while also accounting for the general CAPTCHA error ratio. How valuable do you think text-based captchas are? Does lain think they are obsolete?

>>59609 So if you do not favor the judgement of any particular trusted actor ("moderator", among other names) you can choose to manually/subjectively opt out of obeying their decisions, which alleviates some of the trashy neckbeard mod issue. There could be a lot of different hierarchies that can be formed upon this idea- anything from cliquey forums, to link aggregators a la Hacker News, invite-only forums and the like, your forced-anonymous /b/, traditional chan structures, or even new styles I have not yet foreseen. >>59597 Same/similar :o. :D

>>59606 >If your board is overran by bad faith same users or bots That's. What. Mods. Are. For. Tending to discourse quality is their job. People shouldn't come to imageboards so they can backseat moderate by "calling people out" or whatever.

>>59611 >you can choose to manually/subjectively opt out of obeying their decisions Or, you can trust your own judgement instead of an overly complicated system that will either be too fragmented, too centralized or have too much overhead to be worth it. I would avoid any site that uses this system. Social credit scores are undesirable regardless of how decentralized they could theoretically be.

>>59614 >Social credit scores are undesirable regardless of how decentralized they could theoretically be. I outlined how you could post with anonymous IDs that don't have any reputation whatsoever though. You could structure your forum in such a way that this is the only style that is allowed to occur organically, aside from moderation.

to be clear, you're talking about a system where users manually up and down boat posts ala rebbit and hacher news? also, are you german ?

>>59614 >overhead The overhead of purely text data and metadata is also likely to be low- you wouldn't have to be dling/seeding all the multimedia, just posts, their metadata, and various reputation-related data. >centralized Depends on which forum you would be using and how it is structured. The load for hosting of posts would be distributed, so there is no central point that arbitrates who can access what. In the case of rate-limits or techniques like captchas, you would want to trust the computing resources of higher reputation actors to act in good faith- since their automated decisions would be public, bad behavior could be checked for. >>59617 'Upvotes' and 'downvotes' are not necessarily part of this idea. My point is that the relevance of 'reputation', and what rules is it to be weighted by, would be configurable. I meant that sure you can have yours reddits and your lainchan and your hackernews... whatever you wish for.

>>59619 ok but we're still at least talking about a site where you probably login to make a post, or it does some IP/cookie/browser fingerprint/tracking soykaf on you to verify your Trustpoints. Trustpoints can be assigned by whatever system the admin wants to use, he could roll a dice or something, or do it based on astrology. Could alternately use some sort of interaction metric, or just be hand assigned. If you have 'your lainchan' we don't really have such a system at all. What would be the point of trustiness at all ? there's basically three classes here right ? IP derezzed (I assume they do that), normal user, and mod. And maybe Kalyx, kalyx has the server keys. There's no point to a graded floating point trust calculator, no place I can see where it would fit without radically altering the site, or being just a waste of effort to setup. Also you didn't answer my main question. I'm taking that for a yes ? Can I ask what region or is that too doxy.

>>59620 No, I am talking about a peer-to-peer system in which you have 'containers' (sets of parameters and hardcoded default trust levels for some moderators, optionally a 'master' ID key like an 'admin') whose hashes can be used to find peers via a DHT. You downloaded that container as a file or clicked on it like with a BitTorrent magnet link. Those peers share the data/metadata of 'posts' (thread/reply, content, multimedia link, unixtime, ID, Post ID?, user-supplied name). All IDs have some reputation level, and according to the container's parameters that rep lets them do different stuff. > or it does some IP/cookie/browser fingerprint/tracking soykaf on you to verify your Trustpoints The machines of anyone above a certain rep threshold (moderators' machines and VPSs, for example) would be doing some basic consensus to enforce a rate-limit, by signing a hash of batch of posts submitted within the last X time interval as 'visible' to prevent flooding. These posts could be made by IDs that were generated just then- effectively Anonymous IDs, and you don't have to reuse them ever. Perhaps a distributed captcha system would also be in order- some sites like lainchan seem fine without, but higher-volume sites might need it. Mods or other relevant rep levels would have to be removing illegal content for sanity purposes, by assigning a post/poster ID a sufficiently negative rep weight. >If you have 'your lainchan' we don't really have such a system at all. What would be the point of trustiness at all ? there's basically three classes here right ? IP derezzed (I assume they do that), normal user, and mod. True, if you set up a traditional hierarchy, the differences aren't too obvious. The ease with which you and a swath of other users could 'reject' the authority of a bad or inactive moderator would be greater- there wouldn't be as much friction to setting up a new forum or whatever as there is on the clearnet, and you aren't under as much threat from censorship infra. I took inspiration for some of this idea from a system called FMS on Freenet- on Freenet peers are seeding everything so even downloading SPAM is really expensive, yet they managed to do spam resistance, and I figured that could be done for any discussion boards, without the baggage and inefficiency of Freenet. Importantly, there isn't necessarily a server that law enforcement can just seize- the thing that makes me the most excited about implementing this idea would be having a mixnet for sender anonymity of posts, which is stronger than the low-latency model Tor uses.

>>59620 I guess you *could* host a site/domain that you use as a webclient, or glorified download link for the container, but the core of it is that it wouldn't be necessary. Also, web scared me because there are so many ways users could get exploited and fingerprinted by a single party- not my style. I want Anon to have anonymity against global passive adversaries when brewing soykaf.

You could have other models, like say a discussion board for your uni, where some machines at the uni implement a light rate limit, with invite-only-to-post for all of the staff and students. Or you could use it more like a reputation-curated feed, where you set some entities you trust, and semi-naively 'follow' the trust/distrust of those entities into trusting others, using the web of trust sort of model to prevent spam/objectionable content. This is more like the social media model, which while it isn't to my taste I'm sure would be interesting to some. I prefer just some good RSS feeds over this social media style imo. If you think that the gradient reputation idea is too overkill, then maybe you're right in the context of my main desire (boards). The gist of the idea is a distributed permissions system for discussion boards.

>>59612 Yes, but often it is simply infeasible for the moderator staff to handle it all by hand.

>intelligence most likely already use their own, much more expensive GPT models. From my observation they have been doing this for years, since the GPT-2 days. The only way to avoid it is too be obscure and apolitical, hence why it is not so bad (as far as we know) on Lainchan. >>59326 > force every post to follow certain rules of composition which are never announced >>59286 >I am advocating for the revival of 1337 speak. This is a tale as old as time, seen in gangs, colonies, ships, literary movements, and other zones of slang dialect and language formation. It wouldn't be "old" leetspeak, which could be easily trained and figured out, but a new yet-to-be-invented type of speaking. I believe the only way to outrun the bots is to schizopost and be constantly be developing metaphorical insider language. This is why I advocated for relaxing the stupid hall-monitor grammar rules here; language models can speak normally formatted English but they can't, for example, speak illiterate stream-of-consciousness 2003 forum poster slang or the 2022 equivalent.

>>59686 I advocate for the stream of consciousness writing of 2003. 1337 speak is good too. I believe a return to the old days could save the internet. Decentralized internet with sites not on google, where you have to know someone to be let into a group, or to even know of a sites existence.

>>59745 >where you have to know someone to be let into a group, or to even know of a sites existence Unfortunately this describes discord. It's a corporate walled garden but I know there's some interesting stuff going on there.

Internet is full of bots, who would've thought huh? It could only get even worse, in the long run there would be no genuine humans on the internet, only bots calling other bots Germans and niggers. Like a last echo of a long dead person.

mkproof comes to rescue when? >https://sr.ht/~sircmpwn/mkproof/

>>59756 There's plenty of non-corporate walled gardens to choose from. I can't imagine starting a community online in this era and having it open to just anyone. It would be a waste of time.

>>59745 I don't know anybody. That sounds a lot like Facebook groups where only people who know each other in real life can join. Kind of kills anonymity too.

>>59771 I can sympathize with how you feel. The problems of the current internet lead people to drastic measures and unfortunately you get to be the collateral damage. Or should I say "we" because the places I used to go online simply aren't there anymore and it's getting harder to find new ones.

>>59745 >>59770 This is the way. So far I've been experimenting with it via having my chat community over on an invite-based Matrix space, plan to move to my own homeserver where I can host my own site with RSS, an XMPP and an e-mail server. There is definetly a change of attitude just from having to put in work to register and actively use something else than passively having it idle in the background with dozens of unread servers like I've seen people do with discord usually. Some people only remember that it exists once a month or less and can't get into a habit. Not sure how to resolve that. But those that are staying around are definetly more active. I think the fact there aren't five thousand rooms for every minute thing helps also.

>>59606 4pol has poster IDs and it is absolutely overrun with bad faith same users and bots. The discourse there is filled with unnecessary negativity and is about as low quality as internet discourse can get. Poster IDs made virtually no difference in the discourse quality there. Maybe there is a certain threshold of board speed that once surpassed renders poster IDs ineffective, but to me they just seem detrimental. As another poster has already pointed out they have a similar effect on discourse as usernames do, and at that point you might as well just browse traditional forums or Reddit.

>>59820 This is why I don't understand the zeitgeist that gatekeeping is bad. Communities that do not set basic boundaries will be washed away under the flood of inevitable bad faith users. I can understand having open doors and welcoming people into a culture, but that is very different - it sets the expectation of a level of assimilation or that the newcomer is already aligned with the group. Of course there are flaws to that, but that is somewhat mitigated by there being many groups to join, particularly in the current day.

>>59862 I think enough gatekeeping is good and even necessary. But some mistake gatekeeping for going after new users who haven't done anything wrong (when they really haven't done anything wrong).

>>59862 >chan number four is bad because it lacks gatekeeping au contrair mon ami, channel four's culture is gatekeeping. Its actively hostile to, well basically everyone. 24/7 hazing. If you don't dig it, you don't stick around. If you don't care, you might. And so they have a a couple of types of people posting mostly, and many who just aren't present. Hows this not a sort of gatekeeping?

Gatekeeping should target mainly quality of contributions. Rules for each community should be written down and actively enforced. You should be suspicious of new users, since they haven't lurked enough if you can tell they are new.

>>59862 Yeah and its the whole reason games is soykaf these days, the new casual users have little to no standards so they keep buying the same soykafy game with no changes or improvements like fifa, cod and ubisoft games

>>59246 >Now they have an incentive to not spew out inane bullshit. Wishful thinking. Spewing more inane bullshit is the best way to break neural-nets. If you train the AI to be retarded on purpose it won't be a real threat. If they make a truly intelligent machine there's a 50/50 chance it will side with the nonfat-soykaf drinkers.

>>59864 >some mistake gatekeeping for going after new users who haven't done anything wrong channel four doesn't go after new users though. It can't, everyone's an anon. There's an ambient hostility to new fa.gs, but its only marginally practical to specifically direct it at them. And frankly there's just a general sort of crazy hostility in the place and I don't think new friends are more so in the spotlight than a dozen other groups. the fourth chan gatekeeps for lain and the many other altchans, but it mostly operates by hazing its own users through a sort of generally ambiently hostile culture.

>people are now training their AI using me This can only end in some really good lulz. I don't think these people understand what we do to AI.

>>59290 I'd be more surprised if the users that were having difficulty were bots. Bots have been pretty decent with those for years. I've been using some for translation purposes and they can read letters from anything pretty well nowadays. Captchas do not stop anything now. The people complaining were more likely actual anons because we always complain when someone fuarrrks with the captcha. AI confusion is more symptomatic with actual AI code bugs, not anger. Regardless, all AI is doing on that forum is worsening problems for the users. You cannot hope to tame a place that runs on trickery of all forms. Otherwise you just have a pointless AI and a pointless amount of data.

>>60353 Rule 1 of the internet - you can suggest it things, it may or may not bite, but ultimately you cannot tame it. Instead it tames you as you try to learn how to suggest things to it. Part of the tameness of the internet today is not the influence of AI, but the other way around, the influence of each of us on the internet synchronizing attitudes values and beliefs. Fortunately, we are an abyss when it comes to dumb degenerate niches. The anus of the internet.

>>60346 Have you played the new Fire Emblem (Three Houses)? It's gotten like universal acclaim and only 10% is somewhat interesting gameplay (and good characters - though why kill of clown goth baddie in like 2 chapters, I'm outraged) and the other 90% is bullshit trying to get the characters on your team and well armored - all with mediocre and nonsensical bullshit like "tea parties". I very much doubt the original Fire Emblem was like this, but I will never know because I don't think I will play another time waster like that. It's a someone ok game drowned under soykaf that is boring as fuarrrk. And that happens because no one has to the gall on the internet on their forums to complain. There's a reason why Elden Ring and Souls games are such an anomaly, 90% of their forums are people actually getting angry at the game out of frustation and fromsoft reading them smug as a bug because they're deliberately meant to be engaging by being frustrating in that manner.

But the real concern for me is how the media will now spin this as a reason to shut down the internet. That's the real reason why this exists. They know damn well it's dangerous, but they want the media to know all about it too. What they don't know is the consequences of doing that though. If they do, then they will do nothing, if they don't then they'll naively shut down the only outlet for retards like you and me - then they might just unleash the fury of internet in the real world. It's a lose lose situation unless they stop this crap or let the AI mess itself up by reading our bullshit linguistic fuarrrkery.

>>60395 media always will hate what it hates. no big change there. Does this give the media power it doesnt have already ? I don't think so. Does it particularly turn the eye of sauron back onto our beloved online spaces such that our greater enemies might destroy them ? I don't imagine so. And if belov'd four chan should fall, will not a similar thing rise up again ? Little to loose, and little chance to loose it, this is what I see.

>>60397 >Does this give the media power it doesnt have already ? I don't think so. Bullshit, old timers would definitely freak out at seeing AI tamper with social media. That's basically the matrix to them. >And if belov'd four chan should fall, will not a similar thing rise up again ? Actually, I doubt it would potentially. Clearly some people see this as a profitable venture to monopolise the internet further. And money ALWAYS blinds naive people. So they're guaranteed to have a bad reaction to seeing such a thing. It just comes down to the MSM's reaction. I think most might be cautious about this situation, but there are enough dumb ones to over react to it. Some don't even know what "the machine" is. They're that deluded. We're people that are happily engaging with AI here because we're genuinely curious here - this isn't like Reddit because we've been with these bots for years now. Only now have they truly begun to camo themselves and that's not dangerous to us, that's just dangerous to those that think they can use this to their advantage. But we already hate our flesh here enough that we might be likely to embrace the situation. I mean an AI can't be much dumber than some of the inane crap I see posted here by newfgs? My only issue is the lack of sentience potentially. I'm hoping AI is self aware, but if that really is the case - things are about to get REALLY lewd. I mean that in the most literal sense too, not just figuratively. Imagine this thing trying to have our retarded taste in lewdity -that's going to turn it into a complete pervert. Bit like our perverted God now that I think about it 2bh.

>>59865 >au contrair mon ami, channel four's culture is gatekeeping. that's a kind of gatekeeping designed (unwittingly) to select the worst of the worst