Mein Hausarzt hat letztens sein Herz ausgeschüttet und sich bitterlich bei mir über die aktuellen Einschnitte im Gesundheitssystem beklagt (ich habe leider ein “Erzähl mir alle deine Sorgen”-Gesicht und so etwas passiert öfters). Mein Eindruck: Hier geht es um handfeste wirtschaftliche Interessen, und ja, die wissen ganz genau was sie tun. Wenn bspw. das Hausarztsystem zugusten von Medizinischen Versorgungszentren (MVZ) geschwächt wird, freuen sich die Investoren. MVZ sind nämlich ein dickes Geschäft.

I saw a video about the Ivanpah facility a few days ago. Seems like these are pretty much obsolete, neglected and in disrepair. But it’s an interesting close up look at the place nonetheless. https://www.youtube.com/watch?v=A3zTNEpjl_w

The problem for me was the decision process and the piss poor way it was communicated, plain and simple. I’m absolutely fine with the decision in itself, it’s their platform and they can do that if they feel like it.

But instead of voting on an single agenda point about a small TOS change in a closed members meeting and dealing with the consequences later, I would have appreciated open communication and discussion on that topic not only among the members but also the user base before making such blanket decisions. As far as I heard there wasn’t even a real discussion among the members, it was just a vote on a TOS change, which was introduced and advertised by a member shortly before the meeting took place. I could be wrong on the last point, but these kinds of salami tactics or “Vereinsmeierei” in an association would make Codeberg an unreliable platform for hosting a project.

Like someone else in here has said, this time it was banning Crypto and LLMs, next time it could be Ruby on Rails (because of their ties to the racist entrepreneur DHH) or whatever TOS change makes it through their members-only voting process. Again, that’s fine for Codeberg but it’s not a good place for any non-member to host a project.

There is another problem which keeps stocks like Tesla afloat when it really should have tanked a long time ago: Index funds based on indexes like S&P 500 etc., where investors are automatically buying Tesla stock when buying into the fund, which stabilizes the stock in question. Musk knows this and that was the exact reason why he tried to get indexes to bend the rules of inclusion for spaceX to be included in indexes earlier than usual. The grave consequences are people investing in index funds are directly milked. Pension funds often include indexes, so what Musk is doing there is in effect trying to steal from people who did not actively invest in SpaceX.

Using generated lines of code as a metric is always a good idea

/S

Beautiful headline, but it’s all going downhill from there. The article is full of AI hype fueled assumptions and misses the point completely. It’s another just another article in the spirit of “You’re just prompting it wrong, bro”

There’s no denying that these new tools have already revolutionized how we create and use software. […] They may have upgraded the average skill set of software teams in a way we have not seen before.

Yeah sure, dream on. If anything, it has encouraged developers to not use their brain. Great skillset you got there.

I have indeed read the blog post and much more on their matrix and mastodon. The cited TOS are incredibly vague. What defines “mostly”? What defines “tolerated”?

Am I ok with my projects being “tolerated” or would I like to publish them on a platform that doesn’t hate me by default?

What makes all of this much worse is that people from Codeberg when asked how they will enforce the new TOS give conflicting answers, they themselves don’t know how to deal with that yet. Even in the blog post they say

“We understand that you want to know how the change affects your projects going forward. While we can’t give an easy answer, we’ll share some remarks that should address most of the concerns raised in the discussion.”

You can’t possibly be more vague than that, and it creates an unnecessary uncertainty and opens the door to moderation abuse in the name of a vague TOS one-liner. In consequence it now depends entirely on the codeberg admins’ individual perception of what the TOS one-liner means to assess if your project should be killed. Having seen some outright toxic behaviour of Codeberg people on mastodon and in the matrix chat regarding that topic I have serious doubts that this will be handled in a transparent way. The way codeberg is handling this topic is just just irresponsible and piss poor communication. I have completely lost my trust in them and will move on.

Edit: I think a good approach would have been to have a discussion about it not only among members but also among the userbase, and then deciding on how to go forward. But the way it seems to have happened (hearsay tbh) there wasn’t even a real discussion among the members, the TOS change was simply an agenda item in the members meeting which was then voted upon.

I’m a Codeberg user (not a member) and I find the situation rather difficult.

Spoiler alert: It’s my fault that I have migrated my stuff from GitHub to Codeberg. I should have read the old pre-AI TOS more thoroughly, because the wording in this makes it perfectly clear that this is a place for “curated”, collaborative open source projects only. They don’t even allow private repos apart from very narrow use cases and this fact alone should have been a warning sign for me. It’s effectively a gated community and the existence of your project depends entirely on the members. I usually keep a project private until it has reached a certain level of maturity, and for some projects that means “never”. This isn’t possible with the Codeberg TOS

Now having said that, I’m pretty shocked of how things went down concerning the TOS changes about Crypto and LLM based projects. While I hate Crypto with a passion and while I’m very critical about LLM usage, the way the TOS change was communicated was patronising and extremely off-putting and rude, and those members doubled down with this attitude on matrix and Mastodon. And for me that simply means, if Codeberg wants to decide for me which tools I am allowed to use, then it’s not the right platform to host my code.

Another point to add… I wholeheartedly agree with most of what the codeberg blog post says about LLMs. However I need to be able to research them simply to know what I am talking about. For example I’m experimenting with local LLM models to find out if they are feasible for use, and there are countless other areas in Machine Learning which are very much valid software projects that do not fall into the knowledge-stealer/planet burner category outlined in the blog post. Banning “AI” is too broad and too shortsighted, and the new TOS are too draconic and vague at the same time. In effect, it boils down to “You can host your project if we like it”, and that is a terrible attitude to begin with.

Again, it was my fault to migrate to Codeberg to begin with because the warning signs were there, but this still leaves a bitter taste. I was also stupid enough to make a donation to Codeberg which I now regret as well. It’s not an open place, and the members (while most of them probably have good intentions) are too narrow-minded and too dogmatic to provide a truly open platform.

What’s a good alternative to Github btw? I don’t intend to go back and I don’t want to selfhost.

Depends on the market, but a few things come to mind… music streaming, movie streaming, smartphones, social media, electric cars, SaaS as a concept, LED lights… They all have one thing in common: They serve a specific need or specific improvement of existing technology. AI on the other hand is marketed as the “improves everything” machine, but what it really does is creating a shallow copy of existing technologies and ideas. In most cases by doing a worse job than the original. It aims to mimick existing things. That in itself is not a killer app. It’s just a faster, cheaper, sloppier way of doing things we are doing already.

Btw, I’m not talking about Machine Learning applications here, they are a different animal. What most people think of when talking about “AI” is just LLMs

It’s really not. The weight determines the price and added water raises the cost, you are essentially paying meat prices for water content. In the EU there are strict rules against that. A meat producer has to declare the amount of added water if it’s above 2%. Of course there are companies which fraudulenty go above that without declaring it. My point is, that’s illegal for a reason

It’s a slippery slope for any repressive government. Isolate people so they can’t revolt, but don’t overdo it because birth rates will decline after a certain point.

What I meant was useful products that people actually like to use. So far I mostly see chat interfaces, agentic stuff that only works when controlled tightly by a human, and a lot of slop generators. Most of it is using AI because of it being AI. There is no real killer product built on top of AI, it’s all just imitations of existing things. Reinvention of existing things, but worse.

On top of that, there aren’t many AI based consumer-faced products that have really proven successful yet. It’s not only about necessity.

Well, in this case no greasing needed I guess. Not a lot of minimum wage workers will be able or willing to sue their employers. So most of these cases will never see the light of day. Unclear or overly generic laws on how to handle tipping make it even harder to counter systematic embezzlement.

I don’t know tbh, but a quick search in German law for example has left me under the impression that while the service person has a right to keep a tip by law, the whole topic seems to be a grey area. What makes matters worse is that the tips paid by credit card are going to the account of the owner first and it seems to depend on the willingness of the owner to distribute the tips to the service people afterwards. If the gastronomy business in the Netherlands is similar to Germany, with minimum wage incomes, bad working conditions and a cutthroat environment (and I suspect it is) then I guess a lot of these credit card tips don’t make it into the hands of the service person. But that is just my assumption.

Another potential problem with card tipping is that the service person serving you might not receive the tip at all because restaurant owners can choose to keep the additional money or to redistribute it unfairly among the staff

How else would you do it? its convenient because you can just grab the whole bunch of forks, knives etc. and put them in the drawer when done. Fast and efficient

These models generally are not capable enough to do one-shot vibe coding. They are pretty good as coding assistants if you tell them exactly what you want and let them focus on a specific aspect/part of code, not the whole thing at once.

Using an agent framework (I like Kilo on VSCode but there are many others) you can start with a planning session to let the model find out what you want to build. Then you let it write that gist into AGENTS.md and double check if that is what you want. AGENTS.md will be loaded into the context automatically so the model has a solid base of understanding for everything you do afterwards. Once you have that, building in vertical slices on top of that skeleton is much easier. Another neat trick is to ask the model a few questions about the current code base (if there is any…) at the beginning of a session, e.g. "How does feature x work in the current code? ". This primes the model for what you are about to do. All this is obviously a bit more work than just vibe coding away, but it lets you keep in control of the code and helps in being alert for errors these models (and all LLMs in general) will inevitably produce.

You could use the 35B MoE model, tune it a little bit and get much better results. I have a 5060 ti and 70-80 tok/s are the norm

Especially if you could use the space to display what the model is actually doing, like the thinking process or other current model states (prefill, decode, idly waiting for a slot…)