Those people are completely missing the point in a way that is just funny.

The Arguments

I have read and have seen multiple arguments, so I’ll try to address the main ones:

“doing this will kill their business”

The answer to that is easy: Codeberg is not a “business”, it’s a non-profit organization that just does stuff for its members. Its members decided to ban AI, so that’s what they do. It’s democracy. The people who are mad about this literally are the problem in the world, Codeberg didn’t do that for any kind of business reason. They did that purely out of principle. This was purely a decision made out of moral principle, which for a lot of (mostly American) techbros is inexcusable, because they don’t understand how someone can make a moral decision (and stand by it) instead of following the money.

And again, this is an organization made to serve its members, so what external people think is irrelevant, they’re not members (and if they disagree, they should have been members).

“doing this isn’t efficient! It will make it so the developers using Codeberg are going to be left being”

The answer to that is: So? Not everyone cares about being the “best” or being “at the top”, some people just want to make stuff, to make stuff. It’s that simple. They don’t care if it’s not the most efficient way, and they don’t even disagree with the fact that it makes you more efficient, they disagree with it morally.

“doing this while saying it’s a fascist technology is doing fascism”

That one is kinda funny, because all the people I’ve read saying that are Americans, a country famous for “not doing fascism” (sarcasm).

A lot of people also took it personally by assuming they were called fascists… which is literally not what they said (though I guess them feeling targeted kinda says something).

But that’s not the point, the point is… they’re not forced to accept you. It’s their platform. It’s their decision. Don’t like it? Host your own or use the dozens of other platforms that welcome AI.

Conclusion

Codeberg has always been human-focused and mostly anti-capitalist. I mean, they even promoted and moved to “Forgejo” a protest-fork (that I disagreed with strongly) made because Gitea became a company (to sustain its development).

Point is: the only people who are surprised (like this guy) have discovered Codeberg a year or so ago (and he did). Codeberg has ALWAYS been morally/politically oriented. None of this is surprising.

In the end, having a reasonably big organization that stands by what they mean (and has a spine) is amazing (and I say that as someone who does use AI for some stuff - but not anything on this blog).

And it genuinely annoyed me.

I ended up having a discussion/rant about it, and this blog post is basically a summary of that.

For reference, I’ve been messing around with/learning about codecs and file formats for nearly 20 years, so my annoyance comes with reasons.

It’s Happening yet Again

For people not knowing exactly what AV2 is, I’ll try to explain it simply: AV2 is a video codec, a codec is basically a “format” / set of standards in which you store media.

Something stored using a specific codec can be played (decoded) or made (encoded) in 2 ways:

• Hardware accelerated (specific hardware coder/decoder for each format specializing in making it really efficient)
• In Software (take your CPU/GPU and just “bruteforce” it, which is highly inefficient and slow)

AV2 is an “enhancement” of AV1, and it’s a pointless codec that’s going to cause yet another environmental disaster.

AV1 was already kind of pointless, it’s VP9, but slightly better, slightly more compressed, while requiring far more CPU processing.

H.264 was the baseline that replaced all the really shitty ones like Xvid, MPEG2 and VC-1. Then came H265 which was basically a compressed version of H264 with the downside that low-bitrate files look bad. VP9 (VP8 was basically the same format, but it was replaced so quickly that it doesn’t really matter) was truly the last useful format, because it fixed the problem of H265 and low quality (albeit, with sometimes over-processing/over-shining).

A New Codec is More than a New Codec

Making a new codec means that to use it… you need to replace all your hardware.

It happened with the previous codecs already, and it took years. We don’t need another one.

We were fine with H.264 for a decade. Then we had H.265, which took ages to be adopted but was kinda a good thing (back when connections were slow), and then when things were almost fully set up for H.265, the industry almost instantly moved to VP8 and VP9, the adoption of them was faster but still took years.

Today, AV1 isn’t even adopted by the vast majority, and they are already publishing AV2. Like, don’t release your new format if it’s not good enough instead of making a new one every two years, and I wasn’t being hyperbolic when I said two years, it’s real: “AV2 development work started in 2020, two years after the release of AV1.”

Software Decoding is Not a Solution

People might reply to me that the solution is just software decoding, which, even if it’s slow, is usable. Sure, it’s usable on hardware that is actually overpowered, but on low-power devices, it’s unusable.

For example, I personally experienced the fact that AV1 at 2160p (4K) is barely usable on a high-end 2022 “low-power” device.

But It’s Optional… Right?

Some people might also say that “it’s optional, you don’t have to use it”.

Sure, it’s optional on principle. However, if you try to use YouTube today, you are basically forced to pick VP9 or AV1. Sure, you can technically choose to play an H.264… except Google destroyed their quality on purpose and made them so bit-starved that a supposed “1080p” looks worse than a DVD (not to mention the fact that 1080p is the highest quality they provide in H.264). 

Netflix has also done the same thing, albeit with lower bit-starving.

And spoiler for you: YouTube and Netflix are the main video sources for the vast majority of people.

The Point of This post

My point really is that they need to stick to a format for more than 10 years.

Give people time to actually have hardware to support it.

I know the people working on it are millionaire tech bros, but the vast majority of people aren’t, and even I, while being massively passionate about technology and media, only got my first device that could play AV1 last year and still have some devices that can’t.

Also, up until last year, one of my main devices couldn’t play 1080p VP9.

VP9 had barely started to be fully accepted and available when we moved to AV1. Now AV1 is barely available on anything that releases AV2. Like, if it’s not good enough, take your time.

All about Money

Obviously this is all about money, money talks and a new codec:

• make Google save storage/bandwidth for YouTube
• make Netflix save storage/bandwidth for their service
• push/force people to upgrade perfectly working stuff because technology/computers have plateaued

Forcing new codecs is basically Intel/Nvidia/ARM/AMD (and basically all the companies that sell hardware) way of making you (us) buy new hardware, and this is a way for Google/Netflix to save money.

CPUs have been good enough for almost everything for at least ten years. The only thing that got better realistically is power consumption, but outside of that, it’s only codecs pushing (forcing) people to upgrade.

An Environmental Disaster

Technology aside and buying your hardware aside, this is a massive environmental disaster. This is forcing people to upgrade perfectly working hardware just to play videos. This is all about making people spent more money on new hardware just to play videos. Hardware that ends up replacing the previous hardware, even though it was perfectly working. This creates a massive amount of e-waste for literally no reason other than corporations making money.

A Social Disaster

A lot of countries aren’t as rich as the first world, some people struggle to get even a reasonably good phone, they can’t get the best and greatest.

Forcing new codecs down everyone’s throat is just a way to keep having them behind.

Made by a Non-profit, Sure

Some people might reply to me “but AV2 is made for the greater good by a non-profit”, and to that I’d reply, look at the members’ list of said non-profit when I used Intel/Nvidia/ARM/AMD/Google/Netflix… I wasn’t using them randomly, they are core members and  they’re the ones pushing the “non-profit” to make something that makes all previous hardware unusable.

All these because that’s what you do when you are a big company that has money. You build a supposed non-profit and make a thing supposedly for the greater good that massively benefits you.

The fact that governments aren’t stepping in to manage standards is telling.

We truly are living in a corporatocracy and this is hell.

The Project

Trakt (the movies/shows tracker) has recently gone the enshitification route:

• Their new UI is awful: It’s missing specific views (like season view for TV shows), the list of your comments/reviews is gone, ratings are now out of 5… for reasons… seeing your watch history is now done 1 week at a time, you can’t rate seasons anymore… and so much more.
• People are complaining about it, but the staff isn’t listening (and actually banned a lot of people from their forums… and often from the actual website too)…
• It also removed features and became hostile to free users (then again, paid users also lost features, like automatic data export - and some of the banned people were paid users).
• This is probably the (second) worst UI redesign I ever saw. It feels (and surely is) vibe coded…

So I fixed the UI by only using them for scrobble and making my own UI from their data export, in pure markdown: https://github.com/TheFrenchGhosty/trakt-export-to-markdown

I didn’t want to depend on any plugins or anything that wasn’t portable - and I wanted the file to be human-readable even “un-rendered” - which is why I went this route instead of anything that depends on specific software.

AI Acknowledgement

This was made with the help of LLM, a lot of it was still “human” made.

The project was written and refined with two different LLM “AI” models: Claude Opus 4.6 and GPT-4.1 running in “GitHub Copilot Chat”.

I can’t personally write Python from scratch, but I made those LLM do exactly what I wanted, and then I tweaked a lot of it by hand. There was a LOT of back and forth, and a lot of “human” work, but this is, at its core still a project made using LLMs.

It took me more than 10+ hours of work (minimum).

This was for personal use first and foremost, I just decided to release it.

Consider this provided as is, as the LICENSE says.

AI sucks, but I’m not a developer, have no interest in becoming one, and I’m too poor to hire a contractor. Blame capitalism, not me.

My stance on AI

This blog post is fully written by a human (and so will any blog post I make), I do not use any kind of “AI” (spellchecker aside) for creative work.

My stance on AI is this:

• AI/LLMs have a lot of issues, mostly ethical ones (and resource usage ones in countries where things are mismanaged and/or nuclear power isn’t a thing), like the fact that they did use almost all the open-source projects that exist for training, effectively ignoring the licenses.

• I’m not personally for them, I do use them sometimes, but purely when I have no other choice… which is starting to happen more and more since search engines have become so bad.

• When doing scripts, they are useful purely as a way to syntax things (and in this case, for me to be able to do Python even though I don’t know Python: having the LLM output as a starting point meant I could just read the code and edit what it gave me to make it better).

What does this mean for PussTheCat.org

In short, PussTheCat.org will remain as is… partially (see what’s next below). I started to fix longstanding “issues” of it recently (moving Rimgo to another server, updating/enhancing the homepage…) and it will continue (I also plan to deploy at least one more privacy frontend).

What’s next, and what does it matter for you?

I have had a plan in my head for more than a year now: this plan being to make a “PussTheCat.org on steroid” (I’m not sharing exact details on my plan yet).

This is now what I plan to do: I’ll open my own company (don’t worry, I don’t plan to turn it into a shit one), and go try to do this project.

Obviously, something like that won’t allow me to have a complete salary (unless hundreds of you decide to become a customer), but it’s a stepping stone toward being able to work full-time on what I want: IT stuff (except actually interesting), FOSS and anti-big-tech stuff.

I plan to spend at least 3-6 months full-time on it at the beginning.

Thank you

I’d like to thank, again, all the people who encouraged me and donated over the years, it means a lot to me.

Stay tuned! I love you all! <3

PS: to be clear, the PussTheCat.org ownership won’t be transferred to the company, it will remain completely independent - just with the same person behind it