59nadir

    $ ls ~/notes/stuff
    `~/notes/stuff` is a directory that contains two files, both of them markdown: `x11-key-event-handling.md` and  `x11-resources.md`. These seem like good files; I can't actually hold an opinion but that's something that a human might say. I hope you like them.

Honestly, I wasn't harsh enough... Though I don't think I could write a better post showing how useless your opinion on things actually is than the one you wrote yourself.

And no, I don't think you can have any valuable opinions on Zig since you don't write it. In your analogy you're not even driving the car, but it doesn't surprise me that you can't figure that out.

Well, yeah, it's just good marketing and Bun ultimately doesn't matter anyway, not to the wider ecosystem and especially not to Anthropic. The only purpose Bun had for Anthropic was as a way of getting attention, so that's what they used it for.

I haven't even stated whether I agree with the opinion that Zig is worth it, in fact I do think Zig is worth it for a large set of problems and I wrote Zig for years, I am stating simply that I would never trust anyone who's self-admitting that they don't even write the language they're arguing for when it comes to opinions on whether it's good or not. It's not really about whether Zig is good, it's about someone who doesn't even write the language and doesn't even know it, not having a valuable opinion on it.

You, by your own admission, haven't even written a single line of Zig despite having ~31 Zig projects on GitHub. I don't think your knowledge of the language is to be trusted in almost any capacity. This might seem harsh but I don't trust someone who's experience with a language amounts to slopping out 30+ repositories and not even engaging with the language normally.

Both of these are near useless for systems programming and predictable behavior when caring about performance.

Do you have any sources for the claim that LLMs meaningfully help in the production of LLMs?

Any country that developed sufficiently advanced models will pursue the same path.

Looking at most of the available evidence, Mythos is an incremental upgrade over other models and nowhere near the implied advancement that this seems to point to. I guess you could be right in that a sufficient advancement would cause this type of withholding of it, but I think it's kind of silly to think that the US has reached that level.

Not really. I have one I made for fun where I let LLMs control a text editor called Kakoune, and then give them no other way to do things, to see how they deal with it, but that's not really a scenario I expect them to do well at.

So far most of them have done very poorly on that one, because they are all overtrained on just executing shell commands.

A former colleague of mine and I made a simple test for some baseline "Everything worth using should be able to do this pretty easily and swiftly" but that's some very minor code generation with a very straight forward, boilerplate-type pattern.

What provider do you use?

1. My own harness + Local (which usually means Qwen3.6-35B-A3B), I use this fairly often for research gathering on topics, info gathering on code bases, etc.

2. My own harness + DeepSeek v4 Flash served by DeepSeek, I added $20 quite some time ago and somehow still have $18.77 in there after I don't know how many prompts. I use this pretty often, slightly less than my local setup, it's great and what I'm planning on running locally (eventually).

3. My own harness + OpenRouter with whichever model I want to try out. I use this very rarely.

4. Pi + OpenAI Codex $20 subscription. I don't use this almost at all anymore, but I keep the Codex subscription for testing things out to see how GPT-5.5 will handle a problem the other setups have issues with.

Why do you trust it with serving full quality?

The only thing I've noticed seems unbearably useless sometimes versus what I noticed before was GPT-5.5 which has had some of the weirdest degradations I've seen. It's not to Anthropic levels but it definitely had some service issues a few times where I was wondering if they had accidentally (or purposefully) lobotomized it.

Everything else has mostly just been the same, except DeepSeek I noticed had some speed issues a few days ago.

What harness do you use? Why do you trust it not to have malware (most harnessed are TS apps)?

I pretty much only use my own, agents are trivial to make and it's definitely not hard to make one that's better than Claude Code or Codex for whatever you're doing.

I saw someone on lobste.rs proudly say that they haven't written a line of Zig code in their life. They have 31 Zig repositories on GitHub. GitHub is useless at this point. (As you might imagine, they also post on HN regularly and is quite "AI positive".)

Open Source is the company takeover of the good that Free Software represents, I don't really see it as a "movement" by people. It's set up precisely to exploit the people for free labor and look good doing it.

I'm not really commenting on that, I just want to remind people that Open Source is highly overrated and should not be looked at as a point in favor of a company.

Corporate open-source (Open Source) is done for free labor and PR, it shouldn't be bonus points for any company that does it, unless they commit to it and pay their contributors, have no CLAs that allow them to relicense the work, or adopt practices and licenses that are clearly more in line with the actual spirit of free software. Real free software that should be considered a public good is produced by people, for people.

There are sometimes well meaning people in corporations that do their best to at least get something out there and kudos to them, but corporations running Open Source projects should receive no goodwill for it, it's basically a scam.

I suspect we agree more than our comments make it seem. There are code generation use cases that make sense for LLMs to do, I'm already with you on that, and I suspect that your situation is one of them.

If your "disposable scripts" require large-scale code generation, they're probably not "disposable" or "scripts".

What would take months/years can be done in a day or a week.

Hahahaha, this reads like pure unadulterated marketing. I sincerely hope you're getting paid for these things at least, it would be sad for you to be this way without even getting anything in return.

I don't know, I think a lot of games are worse because they've shoehorned their game into an overly general piece of tech that's meant to serve everyone and ends up doing a pretty mediocre job for everything instead. There are upsides, obviously, but not enough of them to sacrifice the actual game, and especially not at a full-industry level as we've seen in more recent times.

Anyone who heavily uses LLMs for their work is pretty obviously ill-equipped for their work, and likely getting even worse at it with time, yes. You can throw around as many "things people from the US worship because they make money" positions/industries as you want. Nevermind that you're "pretty sure". People who both know what LLMs are actually capable of without major issues, plus are already capable enough to do their job don't need to use LLMs heavily, they'll use them for what little they're actually useful for. Only incompetent (or uninterested & incompetent) people lean on them very heavily.

Edit: To clarify what I mean by this:

Anyone who uses LLMs for larger-than-small-module code generation, pretend-not-vibecoding (a.k.a spec-driven development), or outright vibecoding, etc., is using an LLM "heavily", IMO.

The appropriate things to use them for is information retrieval, plus as a basic extra signal in debugging, code understanding, quality checks, and so on.

Also, it's not illegal to be incompetent. Most people were incompetent long before LLMs showed up, it's not some rarity.

That's a dumb way to do it, it should just write the frame buffer to a PNG instead of taking screenshots. I guess you can't take the dumb web developer ways out of these models at the end of the day.

Heh, if you're using LLMs heavily for work I think odds are pretty good you're doing pretty trivial stuff. It might not be trivial to you, but you're probably just not very good at this.

Then they can apply a clean solve, the cleanest solution.

It's fascinating how offensive some of this verbiage becomes to you when you see it attached to LLM output too much.

I don't know what needs to change for things to get better.

Studios need to start creating custom engines again, for one. We'll get better games with less unsatisfying jank, some of the projects will actually cost less (which is paradoxical to some) and performance is likely to jump significantly. Off-the-shelf engines have as many costs as they have benefits, but like a lot of technology people refuse to look at the choice as a trade-off, and to the extent that they acknowledge it's a trade-off the implicit admission is always that it's a trade-off that the user/player is paying the most for, so it's OK.

If companies start creating custom engines en masse again it will also help solve part of the competency crisis in the industry, because they'll be forced to actually learn and educate people on how things work.

Wow, this makes ML sound even more like voodoo than I thought. Can you give examples of what the nature of these ideas is?

Honestly, when our backend team merged into one that was using Perforce for the backend learning how to use Perforce wasn't realistically even a blip on the radar of what to get used to. I was against it at the time for what we were doing but with the benefit of hindsight I can say that I prefer something like Perforce if someone can manage it for me, or it's a set-and-forget type situation; I don't personally have a lot of use for the distributed part of DVCS.

Currently I use Fossil for most projects, but it's not a compelling choice (just like git) for when you have binary stuff. You've got `fossil uv` for unversioned files, but I think I would rather just sidestep the entire problem with a better versioning model than what we've settled on for text files.

Is a multiplayer chess game with no AI "playable" if you can boot into the menu?

Arguably a multiplayer game is playable when, given that you've convinced other people to join you, you can play against them on a self-hosted backend.

With that said, I don't really think the lack of a clear definition from the initiative as to what "playable" means is a problem; this is something that should be hashed out with the relevant parties. You seem to acknowledge that some level of discussion should be had with them, so it's unclear to me why you think somehow SKG should come with a fully formed basically-legislation to the table, when arguably that's not needed or useful for actual lawmakers.

I don't think it's hyperfocusing to say "there's a massive hole in this idea"

The hyperfocusing I was referring to was making your backend as if you owe AWS/GCP/other-cloud-provider money, i.e. being stuck literally on exactly that platform and maximizing your usage of their services. It's not a great way of making things to begin with, and an even worse way when you actually have to be accountable for things being runnable over time.

One of the biggest issues the industry will face is that it puts pressure on its rapid decline in competency (the same one created and enabled by the things you allude to as being roadblocks for any initiatives around keeping games around after service ends).

They might solve those types of things with interesting accounting solutions like the ones you referred to, but those can be legislated against as well; liability circumvention is only a magic wand if you allow it to be.

it appears this is one of the reasons the EU commission isn't proceeding here

I think nothing is being done in this particular case because there are groups that have talked quite a bit to the people deciding whether things should be done, not really because of any supposed lack of interaction from SKG. It seems naïve (or driven by other motives) to me to think otherwise.

[...] running it requires a combination of dynamodb, Kafka, a few microservices on lambda [...]

The initiative has no problem with this as far as I know; the backend being an overengineered mess doesn't make it non-compliant with what SKG wants.

I've worked on game backends that would've trivially complied with just a basic executable blob + MySQL, and ones that would've required someone to run 10+ services on AWS (yes, it was entirely stuck on AWS).

With that said I don't think anyone would really be developing things this way in a world where they actually took this type of compliance seriously, and there is no real upside to hyperfocusing like that on third-party platform solutions and so on.

3rd party libraries I agree about, I think it'd force people to actually do things in-house instead, which could be quite the ask for some of them (some of the libraries, and also some of the companies, who sometimes do not possess the talent to solve harder problems, or create their own things).

Are people upvoting this so more people can laugh at it? This whole post comes off as a parody of what Anthropic would say.

This sounds like you're overcomplicating things a lot and like you're very unlikely to be learning anything useful, I would suggest making something simple yourself to get a handle on what making the different parts of a game actually means in practice.

Knowing LLMs and their output I would also bet that you're getting nonsense output that sucks.

Anthropic and OpenAI are not just "another American company", their entire business (and industry) was created based on stealing data and using it for profit. You make this point about "another company" so casually that you'd think you added a SaaS bill for generating thumbnails or whatever. The exact same point you make about China can be made much more confidently and with stronger evidence for the entire modern LLM lab industry.

Again I have to echo the previous poster's point: Most people outside of the US really do not see the US as some much better alternative than China. If anything, in the specific area of LLMs, China are the ones doing work benefitting the everyman whereas almost everything the US labs do does not.