We need to make a closed off web

Read this first, if you haven't already: https://forum.agoraroad.com/index.php?threads/dead-internet-theory-most-of-the-internet-is-fake.3011/page-14 No chan/forum/social network can really be immune from bots. Considering that the government and other very large institutions are the only known entities capable of operating botnets to such a large scale to the point of taking over the entire internet, they are probably capable of more than we know and have classified technologies we are uninformed about. Things like captchas are probably only there to fool you into believing they work (and to get rid of unwanted, small-scale bots). We could make invite-only chans and forums but that would rely entirely on the admin not lying. We need to create a closed off web, where there is an egalitarian system for adding and banning users, where anyone could easily create a forum/chan without having to worry about abuse.

>>80077 I agree with this and we need to stop relying on the internet for everything like dating and socializing and posting all events on social media. It feels like we need more real life 3rd spaces.

Wouldn't it be funny if the posters that keep mentioning this theory were actually bots? Anyways my answer is PGP based WoT, Debian style. Keys should be hard to authorize, easy to revoke, but the entire process should be relatively anonymous and offline (no need to verify identity, just that you're human). Though I imagine tons of pointless drama about about people being accused of botting.

Captchas arent only measure against bots out there and some sites make their own captchas instead of using ones that have automated ways around them. Other methods include honeypot forms which you have to be clever in how it is done to be not detected or automatically gotten around,

>>80082 I meant the bots could be so developed that they are able to bypass most, if not all, captchas

I don't think we can ever go back. Anything short of heavy surveillance, like a superhuman AI overlord observing everyone 24/7 to stop botters, can be worked around. It's not as simple as bot vs human. I think the general direction of propaganda or the best clickbait content was and still is made by human experts, perhaps now assisted by AI copilots. Troll farms to sway online opinion were always a thing, AI is just a tool to increase their efficiency and lower costs. As long as it's wildly profitable to do "public relations" (marketing), it will be done, think Gaza, Ukraine war propaganda on all sides, elections, climate change, tobacco, or legitimate product ads. Doesn't matter if a dishonest man or a bot creates the content. Same thing was done way before the internet existed. Amplification (likes, upvotes, replies, saturation by repeated same sentiment posts) would be the more automated part where AI and botting shines. Captchas that a cheap AI can't solve can be outsourced to Indians who solve them for let's say $0.002 each. Proof-of-work captchas could price these out, but captcha as a technology is no way to prevent botposting. >>80081 >PGP based web of trust Perfect technical solution for authenticating people, but the core issue isn't the authentication method or centralization. It's that people are for some reason attracted to and easily influenced by shortform feels based shallow messages, tailored to their personality type. Like the drug problem, it'll always be made and sold as long as there are buyers. >people being accused of botting Past a certain group size (in a https://lobste.rs type invite system) you'll be able to buy an account or invite at an affordable price. Radical idea, but we could try to limit social groups to 150 people living together physically, with their isolated social media. Return to monke. Small closed websites is a version of that. Neat, but challenging to get a critical mass of people, seeing how few active users open altchans have. A lot of the people you'd want to bring are trapped in big tech social media, or are smart enough to be offline. https://en.wikipedia.org/wiki/Dunbar%27s_number >Dunbar's number is the suggested maximum number of people that humans can maintain stable social relationships with, based on brain size and primate research. It is around 150, but varies depending on group cohesion, dispersal and language. We could invent a +50 IQ pill. Leaders and their corporations would fight tooth and nail against this being widely available, because they couldn't abuse the population as much without getting hanged. Sounds like a conspiracy theory, but relatively stupid, impressionable, docile masses are what any logical leader would choose to rule over, especially an exploitative psychopath. In fact corporations are responsible for a few points of IQ drop caused by adding lead to fuel in the 1900s, whether it was intentional or just a byproduct of greed. Best we can do in practice today is avoid slop content (regardless if it's human or AI made). 4chan, Facebook, Reddit, YouTube, Twitter, what's the general result of spending 1000 or 5000 hours on these? Set a time limit at least that you'll stick to. Read a book instead or something.

>>80084 But no one government or big corporation is going to bother with tailoring anything that isnt Facebook, Xitter, Reddit, YouTube, TikTok, 4chan or VK.

>>80091 That's the main content the masses see online though. Rest of the web is Google Ads which can also be effectively targeted if the ad client wants to. Even traditional media like TV or print do rudimentary targeting, "only old people watch TV at this time of day, let's put old people ads". Political profiling was used in the U.S. Facebook Cambridge Analytica scam. Russia's or India's leaders don't have backdoor access to those granular tools, or simply don't have to resort to paying for them because less freedoms: they kill off real media companies and saturate their own digital and physical media with the desired messaging instead. With infinite cash (pay with taxpayer money) and no opposition you don't have to target as much, just saturate everything with your messaging. If ads, propaganda would be more cost effective using hyper personalized messages, they'd totally do it, data brokers + big tech surveillance + text-image-voice analysis + AI image gen: "ur gf secretly wants u to marry her! click here to buy our blood diamond ring fitted to her finger! she's pregnant btw [wholesome image of you putting the ring on her]" This should be doable today, maybe the only reason they don't do it is a cost barrier, or that normal people would realize the extent of spying going on. It's nigh impossible to buy a car without an always-online tracking via built-in 4G and the maker selling your GPS history for $0.50, FFS.

>>80088 Perhaps the invitation system could work like this: The first group of people to get invited gets (for example) 5 invites. The people they invite get 4 invites, and the people they invite get 3 invites, and so on. It should be implemented in a way that rounds off to 150-200 people getting invited. And whenever a user invites someone it should be approved by a group of random users.

>>80088 I've heard of Dunbar's number, and its actually pretty believable. I would imagine 150 is a small village, prefect balance of retvrn to monke with economies of scale. Not sure how this would pan out with social media though, since you arent in a standard individualistic relationship with them (but rather the collective), and they only form part of your social circle regardless (so they cant take up the full 150). As for keeping the population stupid, I recommend reading "Dumbing Us Down: The Hidden Curriculum of Compulsory Schooling". I used to think the people hating school were retards who couldn't do multiplication or read books, but that book changed my mind.

you are basically dreaming about urbit/web3 you want closed off net? you already have it. phone verified, id verified, paid only, invite only, login to view, captcha-on-every-step, centralized websites with increasingly policed and moderated content+users with CIA monitoring as a bonus are a common place and make up most of the net traffic nowadays... honestly its like each generation of netizens keep getting worse and worse... there is nothing new about dead internet theory its been known for decades bots are fuarrrking native to internet of course they will be the dominant over here anonymity of the net its a feature not a bug the fact that you might be taking to a russian bot, escaped AI, 40 year old who pretends to be a 16 girl, a very smart dog or a mix of all of this is what makes the internet not what breaks it. what breaks the net are attempts to move away and reject its fundamental properties. if you dont have the balls to embrace its chaos then that is your problem.

>>80129 Urbit is a train wreck: https://wejn.org/2021/02/urbit-good-bad-insane/

>>80133 yarvin is back and sounds like he wants to change some things make it better address possibly a few problems in this article I havent heard much but it sounds perhaps promising so stay tuned

>>80077 Doesn't I2P fit this descriptions pretty well?

maybe nothing is immune from bots, but imo they are not *that* big of a problem. they're mainly recycling existing content with minimal supervision or the content they make can be spotted at a glance. if you can't tell a markov chain made that, this web thing isn't for you

>>80129 >you want closed off net? you already have it. phone verified, id verified, paid only, invite only, login to view, captcha-on-every-step, centralized websites with increasingly policed and moderated content+users with CIA monitoring as a bonus are a common place and make up most of the net traffic nowadays... Yeah great, except the part where none of the bad stuff is what these people want in their solution to the bot problem. While there is some adventurous chaos lost in seclusion, a single entity spinning up a ton of bots is the opposite of chaos. I would prefer a variety of reputation devices on a very open platform.

>>80129 Your phone manufacturer is unhappy with you.

>Radical idea, but we could try to limit social groups to 150 people living together physically, with their isolated social media. Return to monke. Small closed websites is a version of that. Neat, but challenging to get a critical mass of people, seeing how few active users open altchans have. A lot of the people you'd want to bring are trapped in big tech social media, or are smart enough to be offline. https://en.wikipedia.org/wiki/Dunbar%27s_number >Dunbar's number is the suggested maximum number of people that humans can maintain stable social relationships with, based on brain size and primate research. It is around 150, but varies depending on group cohesion, dispersal and language. Not a bad idea, something like the WELL. Anyone familiar with captchas should be aware that they've become a scam used to justify more data harvesting. Just saw this article about how an AI run on regular consumer hardware can solve them. https://arstechnica.com/ai/2024/09/ai-defeats-traffic-image-captcha-in-another-triumph-of-machine-over-man/

>>80077 Just saying, everyone's been saying "dead internet" for the past 10 years, get over it and move on new user. Also >agora road

If my interest is have a healty comunity I could ask to everyone trying to enter to speak directly with me or mods first. If you want only make profit of media your have to sell it by number of users.

>>80150 Those images glow. Extreme glow detected. You are beginning to glow. State: Glowing. Degree: Faint but apparent Notes: Non-glow is \creating or proliferating glow content.

>>80077 >No chan/forum/social network can really be immune from bots. Can I introduce you guys to Hyphanet(Formerly known as Freenet)? It's decentralized and provides high anonymity. It's open source, web 1.0 style and you can easily create your own forum or participare in one. You can view forums in Thunderbird which I found very comfy. https://www.hyphanet.org/index.html

>>80088 >We're living in a police state with the most restrictive censorship ever seen Not even remotely close. Americans think that the US is the only place that exists

>>80460 No one cares about your third world country, sorry. All western countries are a puppet of the US anyway. Welcome to 21st century globalism

>>80077 >No chan/forum/social network can really be immune from bots. Indeed, though with hashchans we can make those big institutions pad anons investments if they want to flood, slide, wage slave, groom and heaven bann

>>81780 This model requires no hosting other peoples images, so it cant be flooded with illegal content, there is no central server to ddos so the cloudflare mafia needs to attack ethereum to shut it down- as it can run on localhost and ones own archive node.

>>81780 why do you keep advertising this so much? This chan has some big flaws, like it fully depending on the will of somebody to pay just to post. Not everybody is. Not everybody has the crypto, or wants to waste it to reply to some ragebait, for example. Then, there's also the proof of stake (aka, your system). A government entity can still generate a lot of eth and still spam, albeit at a much higher cost. Why not use Proof-Of-Work with an ASIC and GPU resistant hashing algorithm? Everybody does the same work, ASIC's and GPU's are useless, no crypto involved.>>81780

>>80077 >Things like captchas are probably only there to fool you into believing they work >probably Captchas IMO are fit for purpose. Let's think of it this way: If you can make the cost of machine-posting reach parity with human-posting, then you rob the machine operator of any advantage that using the machine grants. If you do that, then the machine stops being worth it, and you basically devolve back to the status quo of a pre AI world. Now then, right now on sproutgig (a micro-work site), there's a nigerian fella who's offering to do 1000 backlink spam posts for $5, which works out to $0.005/post. ChatGPT's paid models work out to $0.002 per 1000 "tokens", which roughly means a 1 paragraph response to a 1 paragraph input. So for brevity we can say GPT works out to $0.002/per post. That cheaper than human labour for posting alone. Captcha can be trivially broken, but at a cost. 2Captcha, for instance, can solve most common captchas for around $3/per thousand or $0.003 per captcha. Our nigerian buddy solves the captcha as part of his per post fee. However, with AI, we have to add a new machine per mitigation, so GPT does the posting, but we need to now stack on 2captcha's fee for solving the captcha, taking us to price parity with Mr.Nigeria, since the cost per 1000 tokens for text is $0.002, and the solve price per captcha is about $0.003, totalling out to our prince's price of $0.005/post. We could possibly stack other technologies up to increase the marginal cost even more, and make human posting to be cheaper than machine posting as well. All of the sites we see where mass botting with machines is happening either don't do captcha-per-post, or else they have bypasses, e.g. like 4chinz's pass system. With these sorts of situations, you don't have stacked up mitigations in place to keep the price of machine-posting higher than human posting. Generative AI can be detected pretty easily too with machine learning solutions, so if you already have a decent system in place for triaging posts for moderation, you could add bot-checking the posts to the moderation workflow, and then ban any bot posters. >>80082 >Other methods include honeypot forms which you have to be clever in how it is done to be not detected or automatically gotten around, If you're targeting a particular platform, these work out to be a security by obscurity mechanism -- all it takes is one manual post with burp attached to see what you're real post fields are. If you're developing a bot to spam a particular site or a particular software suite, that should be your first course of action anyway. The only thing this mechanism stops is automated bots that work more like spiders. >>81818 >Why not use Proof-Of-Work with an ASIC and GPU resistant hashing algorithm? Proof of work is just a convoluted rate-limiter. What you're doing with proof of work is wasting CPU cycles to increase the amount of processor time wasted per post, which with rented infrastructure theoretically increases the cost per post. Point one against it, is simply that you can use programmatic rate limiting (i.e. like a timer) to equally waste time for the spammer. Like the whole point of proof of work is to waste cycle time, so you can equally waste it with something simple like a timer. Point two is that compute cost is so low that wasting it via time at all requires pretty absurd timeframes, whether you use a timer or a proof of work system. This nerd kinda goes over it here https://news.ycombinator.com/item?id=37056504 >why do you keep advertising this so much? well, he makes pay per post :3

>>81830 >What you're doing with proof of work is wasting CPU cycles to increase the amount of processor time wasted per post partially true, proof of useful work also exists, but afaik that isn't as ASIC resistant as Scrypt/Argon2 hashing is. the link you provided also mentioned that the time needed to "mine" is too high for a regular web application, but since we are talking about image boards, which already have a decent latency (as in posting speed), does this really matter? also, a regular time-based ratelimiter isn't as effective because all you have to do is wait. with proof of work, you need to put effort in before you can post something. it's like comparing apples and oranges.

>>81832 >also, a regular time-based ratelimiter isn't as effective because all you have to do is wait. with proof of work, you need to put effort in before you can post something. it's like comparing apples and oranges. I mean, I'm not really experienced in implementing proof of work, so that could explain why I see it this way, but I don't see a difference. The way I'm seeing it is that the point of making them put in effort is simply to rate limit them. Like you give them a challenge that on XYZ configuration takes ABC number of CPU seconds to solve. It seems like an over-complicated timer to me. Instead of having a script somewhere on the server that simply counts off some number of seconds, you're forcing them to do a task that is expected to take some number of seconds. The ultimate result of both is that the spammer's time is wasted. It could just be that I'm not seeing the difference due to unfamiliarity, though.

>>80077 no worries, even though I have implemented a few PoW systems, I am not a pro on this field lol. while it does waste time, it also uses up a lot of resources. for your average poster, who posts a message every 3-5 minutes, having to wait 20 seconds before posting isn't too bad. but for a spammer who wants to post a new message every second, he now has to wait 20 seconds and use a lot of processing power to post a simple message. and if you combine PoW with some more advanced anti-bot techniques, such as preventing too much messages from a single IP (defeated by proxies and the like, but still) or limiting spam-like messages (crude filters, blacklists, idc) and you got a pretty decent system

>>81830 > All of the sites we see where mass botting with machines is happening either don't do captcha-per-post, or else they have bypasses, e.g. like 4chinz's pass system. With these sorts of situations, you don't have stacked up mitigations in place to keep the price of machine-posting higher than human posting. I've seen sites where Cloudflare's CAPTCHAs was absolutely required for every post, and it still got spammed by thousands of bot posts. Who cares if it is more or less expensive to the attacker, they still flooded us to DoS levels anyway. > Generative AI can be detected pretty easily too with machine learning solutions But for how long? An eternal arms race? And what about custom made models by bigger actors? They might not be detectable because no one has access to enough output data to analyze them. >>81830 > Point one against it, is simply that you can use programmatic rate limiting (i.e. like a timer) to equally waste time for the spammer. Like the whole point of proof of work is to waste cycle time, so you can equally waste it with something simple like a timer. >>81834 > I mean, I'm not really experienced in implementing proof of work, so that could explain why I see it this way, but I don't see a difference. But rate-limit by what key? Their IP? What of IPv6? Also, hiring botnets as a service, of hundreds of thousands of hacked home computers is not that expensive, which can avoid rate-limits. Those botnets can constantly attack a list of sites in turn. But those same computers would grind to a halt if they faced proof-of-work protections everywhere they tried to attack simultaneously. Thus making it more expensive to botnets by a few factors to attack proof-of-work sites. This is probably why Cloudflare introduced proof-of-work; it might not protect one site much, but harms the botnets collectively. Plus, one important factor, is that the attack bot needs to now be smart enough to execute scripting languages to solve proof of work. Most script kiddies and cheap gun-for-hires are basically curl posting, and would have no idea where to start to solve your custom proof of work in their scripts.

>>81840 ratelimiting by IP is just one example I gave, you could use cookies, user agents, SSL fingerprints, whatever. also, introducing scripting languages to the mix is a nice idea, but once you use JS you've already lost a lot of users who want to stay anonymous. maybe some simple scripting languages, just for solving some computationally expensive things? A simple interpreter would be passed/implemented in the browsers/clients and could then be used for PoW. the language needs to be just small enough to do simple arithmetic, maybe some bitwise operators, and that's it. keeping it small and light will make the attack surface relatively small (only things I could think of that could be exploited would be potential interpreter bugs, DoS because of hard challenges) idk just me ranting lol

>>81837 ah ok, I wasn't thinking about how you'd need to identify them for a conventional post timer. Actually forcing them to waste the CPU time does make sense in that case.

>>81841 >but once you use JS you've already lost a lot of users who want to stay anonymous. Not phone posters though.

>Everything is all fuarrrked up and I want a new place where things aren't fuarrrked up and can't be fuarrrked up. This is one of the biggest and most constant themes in the history of life on planet Earth/ It even prefates human life. The usual remedy is to migrate to some better place that isn't already overpopulated and full of conflict and competition. The problem with that is that the world itself is now overpopulated and in constant conflict and competition. That includes the internet. Build a walled garden and it immediately becomes a target for jerks with nothing better to do, mentally ill people with delusional paranoia about it, control-freak governemts that can't stand not knowing what's going on there, business interests who want to exploit it for profit, criminals who see it as a new hunting ground, fanatics who see a fresh chance to convert an isolated community to their flavor of insanity... Can it be done and work? Maybe, for a little while, with a lot of effort and ingenuity by some extraordinary, highly motivated people. But it never lasts. It never has in four billion years of life on Earth, and it never will. That's just the way life works.

>>80088 That first image is interesting but I can find no evidence on 4plebs that such an incident actually happened. The earliest references to it are from July of this year, and the first and only mention of "Dutch Day" was Sep 21 2024 Also notable is that the July posts claim the event happened in 2018 while the September posts claim it happened in 2021. They can't even get the story straight.

>>81830 > well, he makes pay per post :3 hashchan has no token or takes no cuts. there isn't any hashchan fee. the profit comes for people holding crypto when >>81860 pic related have to use crypto to pull this and it drives alot of network demand >>81818 >why do you keep advertising this so much? This chan has some big flaws, like it fully depending on the will of somebody to pay just to post. Not everybody is. everyone is paying to post, even right now, just in their data. Once anons realize their all being heaven derezzed by botswarms they'll realize that posting actually doesn't exist anymore, and they'll go to places where posting is freer of botspam >. A government entity can still generate a lot of eth and still spam, yes, though now for their social containment algorithms they have to balance it with making every crypto holder richer. so its better at the margin for government propaganda initives >hy not use Proof-Of-Work with an ASIC and GPU resistant hashing algorithm? ethereum classic is first on the list of supported chains for this reason. and I find it preferable to piggy back on existing chain security guarantees than trying to fund a new network

>>81996 this still stops a lot of people from using the site. not everybody has cash nor the ability to gain ethereum anonymously. the best way to do this is probably a que for posting or make a throttle for users so floods of bot spam and illegal content becomes way harder to do. you could even make a pass system where people could bypass the que or throttle.