03 Apr 2025 by cthos 2973 words

---

breakermaximus @ Shutterstock #1832963167

😈 While this is a post about Big tech, it’s also an indictment of privately run social commons in general. So, prep for that.

This is a cross-site blog post, the other half is on Cthonic Studios.

I’m sure you’ve noticed the trend, if you’re even marginally online. Facebook is flooded with Generative AI bots, vying for views and clicks of the few human users who are still around. Instagram wants you to buy stuff, showing you an ad every five posts and an influencer every three. Twitter was destroyed and now X is masquerading in its corpse, filled with inauthentic content and fascists all arguing with each other all the time. Reddit sold out its moderators and went public, and is now being scraped for AI training data to benefit its owners but not the people who created the content.

Why is everything getting worse? Well Cory Doctorow talks about this all the time, but it’s a combination of capitalistic market forces combined with unchecked impulses of those that control them, and a series of authoritarian interests who want to control the social landscape to benefit themselves. It’s the Oligarchs standing atop a massive pile of wealth with an insatiable desire for more.

In short, everything that made the internet a great place for community is being eroded and devoured in the search of short term profit and autocratic power grabs. And yes, Generative AI is partly to blame.

Standing against those are Mastodon (and other Fediverse software) and Bluesky, both decentralized (though whether Bluesky actually qualifies as decentralized is an open question) services that offer an out for corporate controlled media, but there are challenges to their adoption and a number of less-than-ideal considerations around all of this that we have to grapple with, but let’s start with a history lesson.

How we got here, an Elder Millennial Tale

Permalink to “How we got here, an Elder Millennial Tale”

Come, young ones, and let’s take a trip back in time (to my elders, you’ve experienced even more of this and for that I’m sorry), back to the 1990s and early 2000s. The internet, when I was a child, consisted of a dial-up modem, screaming over the existing phone lines until you were able to log in to your America Online account to check your mail and participate in chat rooms. There were MUCs and MUDs and MUSHes you could play over telnet. You could make a website in raw HTML in Notepad. You could use CSS to style the scroll bars. "Social Media" as a common term just didn’t exist.

You could wait for your online friends to show up in AIM or ICQ and chat with them in real time, or put up a status so that your crush knew that you were thinking about them in cringe song lyrics. Persistent forums were a thing (and still are, if you know where to look), that you could post to and argue about the finer points of TTRPG theory (RIP the Forge, of which I was not clued into at the time).

All of that was decentralized. If you wanted to run a forum, you’d buy hosting from someone and run a forum. AIM controlled chat rooms, sure, but XMPP was a thing (still is, baby!) and you could run your own server. You could join other people’s servers. You can federate XMPP servers together.

Eventually, all of that began to change. First we got MySpace and Xanga. Then eventually came Facebook. And Google Plus (Circles?). Each of these places offered us convenience. Let the company take care of the difficulty of hosting and maintaining a site, and in exchange everyone and all their friends can join up, and you can all be together in a community square. Twitter came along, and let you SMS your thoughts to all your friends instantly. No need to be at a computer, you could pull out your phone and make it happen.

😈 I was working at a white-label social networking site at this time, so I feel fairly qualified to talk about all of this.

This was great for people who weren’t terminally online, like me. A lot of folks didn’t have the impetus to learn how to host and run their own forums, or they didn’t want to go out and seek out a niche community run by some random guy/gal/nb on the internet. More people who might not have been online in the first place flocked to centralized platforms.

The more tech-centric users, too, started moving from their decentralized online identities to the centralized platforms. After all, if all of your friends are in one place and all the folks making fun creative things are there too, why would you want to have to go to multiple places to meet them? Eventually you’ve got just a few places where you’re getting your group interactions. Want to plan a group event? Well Facebook or Google Plus can offer you a scheduling feature where all your friends can sign up. Don’t want to invite friends? Use Meetup.com! Centralized platforms became the norm. The older spaces began to atrophy, though many of those places still exist, many others moved to centralized platforms.

The great collapse and Enshittification

Permalink to “The great collapse and Enshittification”

I’ve talked about it before, and I’ll talk about it again, but Google loves to kill services. Even ones that are relatively popular. In this case, it killed Google Plus, and as a result scattered a bunch of TTRPG discourse to the wind. This happened in 2019, and by no means is the start of these platforms getting shittier over time (Facebook, arguably, has always had some level of garbage to it and indeed it helped fuel a genocide in 2017), but it feels emblematic of what I want to talk about. Vice did a decent article about this, but it includes this quote that I want to call out (emphasis mine):

Right now, itch.io and The Gauntlet, a throwback to the days of forums, are looking like contenders.

The author of that Vice piece considers forums a symbol of a "bygone era" just in that one distilled section. Which, I think is kinda true. Google Plus, arguably, was a type of Forum presented as something "new and different", a true "community". It wasn’t, but it was centralized. A "town square". And people bought the line and went there in droves.

Then Google killed it, and scattered those people to the winds.

Facebook bought Instagram, partially because users were leaking to it in droves from Facebook, because Facebook had steadily gotten worse - slowly turning the screws to get more advertisements in front of its massive user base (in the US) and presenting itself as the "whole internet" (in other countries where internet access was less accessible).

Twitter, too, has undergone a massive transformation since a certain billionaire purchased it and made it into a "free speech utopia", but only for hate speech, apparently. Since his acquisition and running the company into the ground, it’s become a haven for those 4chan kids who want spout as much hate as they can from the safety of their basements, while their glorious leader whines about how oppressed he is. Couple that with the "Anti-woke AI bot", and now the user base consists primarily of edgelords and phishing bots. Attempts to both monetize the site via "pay for a verified badge" and "pay users to post vitriol" are both emblematic of late-stage capitalistic cash grabs and an utter disaster. Twitter’s valuation has dropped by nearly 80% since its acquisition.

A mass exodus occurred there, to Bluesky, and well, the results are predictable. I still know folks who are on there for one reason or another, but having an "X" account is now a social marker that conscientious folks are wary about having.

But that’s not the latest salvo in "every internet social commons is getting worse", that, dear friends, is Generative AI.

Generative AI is poison, being fed to every tech company

Permalink to “Generative AI is poison, being fed to every tech company”

That’s right, I said it. Your favorite "dumb intern" technology is poison, and you should feel bad for using it.

The past couple of years have seen an acceleration of LLM and Diffusion Models poisoning the social media and community ecosystem. This is manifesting in several different ways.

The first is slop being generated and disseminated across any social platform where it’d be useful as a slop vector. For example, here’s a garbage image from Facebook:

Why is this happening? Facebook pays people for viral content. So, people are spending money to generate fake images to try and get fake likes, to try and get money out of Facebook. Bots for bots for people who want to make a few dollars. Meanwhile, Facebook isn’t really doing anything about it, and has decided to pivot to "community notes" because moderation is hard apparently? That’ll fix it for sure! (It won’t).

Along the same lines, you can see Polite Disagreement Bots on Bluesky, "Ignore all previous instructions" on Twitter/X. This is all done for the same reason, to boost artificial engagement, usually for monetary reasons.

There’s a whole other category of using LLMs maliciously in the same way: Disinformation. LLMs can spread plausible looking disinformation at record speed, and the existing social platforms don’t have much interest in stopping them. Cause all engagement is good engagement, baby.

Meanwhile, the only weirdness we’ve seen on Mastodon in recent memory (that I’ve seen) is "Nicole the Fediverse Chick", which appears to be a revenge campaign of some sort? It’s not been particularly effective, most moderators block the bots at the network level as soon as they show up, and if an instance is flooding the network with LLM garbage they tend to get network blocked pretty quickly.

AI Content Scrapers are Terrible Internet Citizens

Permalink to “AI Content Scrapers are Terrible Internet Citizens”

The AI companies think that the internet’s data belongs to them and that they can scrape whatever they want as fast as possible. They’re sending massive amounts of traffic at Open Source Projects pushing communities to close off their content behind auth walls, threatening the open web.

These scrapers are terribly written. They hit the same page over and over again, regardless of errors or if the content hasn’t changed. They’ll obfuscate their User-Agents. They’ll utilize botnets to evade bans. They’re pushing tons of bandwidth charges onto the content creators with no regard for their continued existence. They’re a plague on the open web, and they’re slowly strangling it to death.

Why? Money. LLMs are greedy for data. They’ve hit their plateau and need exponential amounts of training data to improve even marginally. We’re at the point where they’re eating the seed crop, all to chase something that has barely any utility.

Meanwhile, Facebook Pirated millions of books to train Llama 3, flagrantly violating copyright because they were desperate for data. Meanwhile, OpenAI and Google want the US to waive copyright. Ridiculous.

Service providers are selling your content to line their own pockets

Permalink to “Service providers are selling your content to line their own pockets”

So, let’s talk about Reddit and Stack Exchange. Both of those services decided that all the content their users had created for them over the years were a treasure trove of data for Generative AI.

Did they decide to reward their users by sharing in that bounty? Hell no they didn’t. They decided instead, if you objected to this monetization of your work, you’d catch a ban, sucker. In Reddit’s case, they did offer some mods the "opportunity to buy stock" ahead of their IPO, but if you were an average user... well, nothing for you.

Meanwhile, Stack Exchange users revolted, redacting their previous content or deleting their accounts. Stack Exchange couldn’t let the cash cow go to waste, so instead they banned users en masse.

For these commercial products, your labor is their gain, whether you like it or not.

Beyond AI, legislation is threatening open content

Permalink to “Beyond AI, legislation is threatening open content”

Another factor currently playing out are a couple of high profile legislative actions taking place across multiple jurisdictions.

Over in the UK, they’ve implemented the Online Safety Act which requires websites that accept user generated content (which they define as U2U) must comply with a series of vague demands, and for a good while ahead of implementation there were very few guidelines for what individual service providers must do. This has lead to several services simply shutting their forums down rather than having to deal with the requirements, for example Gaming on Linux. Other services have resorted to geo-blocking UK users altogether.

Note: Lobsters rolled back the block after clarification from Ofcom.

This also applies to comments on blogs, too, as The Register reports. Namely, it’s a broad-reaching law, and it’s affecting small communities disproportionately. Facebook can afford to throw money at this, or pay fines. A random blogging community in the UK might not be able to (or, outside the UK with a non-zero number of UK users).

Look, there’s some horrendous stuff that gets posted to websites that can and should be moderated / stamped out. I get that the UK is trying to combat those things, but legislation should be careful of the consequences.

Meanwhile, over in the US, there’s a movement to kill Section 230 of the Communications Decency Act. For the uninformed section 230 ensures that the website owner is not responsible for what the content one of their users posts on the site. Proponents of repealing section 230 point out that Twitter and Facebook host tons of illegal content, and should be doing more to moderate their content. Other folks think it provides too much cover to "censor speech" that they don’t like. Regardless, repeal of 230 also endangers community differently, depending on what winds up happening in the aftermath. While the proposals I’ve seen so far don’t impact smaller sites nearly as much as the OSA, odds are good that it would result in a more chaotic user-generated content landscape.

Here’s the EFF’s take on the matter, and I want to call out the following quote:

If Congress deletes Section 230, the pre-digital legal rules around distributing content would kick in. That law strongly discourages services from moderating or even knowing about user-generated content. This is because the more a service moderates user content, the more likely it is to be held liable for that content. Under that legal regime, online services will have a huge incentive to just not moderate and not look for bad behavior. This would result in the exact opposite of their goal of protecting children and adults from harmful content online.

Big Tech can deal with Section 230 going away, but smaller sites will likely be in a lot of trouble unless there’s a specific carve out. Regardless, they’re not going to moderate better, they’ll just make their services worse.

Towards a Dark Forest Internet

Permalink to “Towards a Dark Forest Internet”

If you’re not familiar with the Dark forest hypothesis, it boils down to "what if there’s alien life, but everyone’s hiding because something out there is wiping out other intelligent life".

The Dark forest theory of the internet hits along the same lines. Communications are moving further into those isolated places of the internet where we try to escape various enshittified processes. We congregate into Discord servers (which, btw, will ban you for using alternative clients), private Slacks, private forums, and the like.

We communicate most often with those we trust, as the wider internet becomes more hostile, and becomes more weaponized, as our services degrade, we cling tightly to those places where our community is. Until yet another platform becomes unbearable and we’re scattered to the winds.

What do we do about this?

Permalink to “What do we do about this?”

I wish I had a good answer for this. I’m going to put out some ideas on what I’m doing, but there’s no one-size-all fits answer for community. I think part of the way we get out of this is cultivating a DIY mindset for community. Teach folks how to self-host their tools, and protect them. Use less proprietary platforms that can go garbage on you when they feel like it. Don’t spend your dollars on services that don’t meet your needs.

So, towards that goal, I’ve done this:

• Using Mastodon as my primary social media, on a server I host, federated with friends servers.
• Self-hosting infrastructure for my RP groups, for hosting our notes, and our virtual tabletop.
• Moving all of my business stuff over to open-source projects, or things I can at least change if I need to.
• Having folks who collaborate with me use my Nextcloud instance rather than Google or OneDrive.
• Exploring federated chat alternatives like Matrix and XMPP.
• Moved over to Linux on most of my machines, and try to use only things that’ll work on Linux.
• Running a Peertube instance and making content for it, because I want to see cooler content on Peertube.
• Helping any of my friends do the above, to get off corporate controlled stuff.

I don’t think this is actually practical for a lot of groups, but I’m willing to give it a go for anyone who wants to come with me. It takes effort and money, I’m not being subsidized by a YouTube, for example, to host my videos or live streams.

But I think it’s worth it.

Part 2 is over on https://cthonicstudios.com, where I’ll talk about TTRPGs over the internet as community in a degrading web. Hope you enjoy.

community rants social-media tech