zero_shift

How much of the work here was done by an agent?

(Being AI accelerated doesn't make this project low value. But it does mean _you_ didn't do the port so much as prompt it)

If all the languages together add up to 15MB that is a game changer for me.

It means the CLI I am working on can ship support for many languages whilst still being a smallish (sub 50mb) download

I shall definitely check it out!

Interesting. I have a similar usecase but intended to use CGo tree-sitter with Zig

Are these pretty up-to-date grammars? I'm awfully tempted to switch to your project

How large are your binaries getting? I was concerned about the size of some of the grammars

I agree. Whilst I think MS products are on a downward trajectory, I'm getting "Maastricht Planning Department switches to Kali Linux" vibes

I want to see (sincerely) a whole government ditch MS

Those are very large reductions - can I ask you for a source?

And why is the error bar so large?

I'm not sure how anyone can respond to that, without asking you to divulge that private data

Cost, debt, difficulty forming a moat, gap between what the product promises and what it can do, and the difficulty actually raising capital required.

His style is acerbic and (imo) excessive sometimes. But he's also one of a minority of journos actually looking at the numbers and adding them up. Which seems to be a rarity

I don't think so.

His argument is not "this tech doesn't work", but rather "these businesses aren't economically viable"

And that the smoke and mirrors accounting and perpetual thirst for more billions indicates just how unviable it is

Whilst he does dunk on LLM capabilities, the framing is the business angle - can Anysphere etc. actually form a moat and make a profit?

Typical shakedown tactic. I used to have a boss who would issue these ridiculous emails with lines like "you agree to respond within 24 hours else you forfeit (blah blah blah)"

Maybe my expensive university degree was worth it after all

I am being slightly sarcastic

One of the great benefits of AI tools, is they allow anyone to build stuff... even if they have no ideas or knowledge.

One of the great drawbacks of AI tools, is they allow anyone to build stuff... even if they have no ideas or knowledge.

It used to be that ShowHN was a filter: in order to show stuff, you had to have done work. And if you did the work, you probably thought about the problem, at the very least the problem was real enough to make solving it worthwhile.

Now there's no such filter function, so projects are built whether or not they're good ideas, by people who don't know very much

That's why he succeeded

Would publishing under AGPL count as poisoning? Or even with an explicit "this is not licensed" license

We are already seeing this in scams, advertising, spam, and social media generation

Not all AI pull requests, are by bad actors.

But nearly all pull requests by bad actors, are with AI.

I don't disagree that technology is less fun in an AI era. The question is, what other careers are out there for someone who wants to make things?

About a decade ago, I went through a career crisis where I couldn't decide what job to do - whether technology was really the best choice for my particular temperament and skills.

Law? Too cutthroat. Civil service? Very bureaucratic. Academia? Bad pay. Journalism? An industry in decline.

It is a shame, what is happening. But I still think, even with AI hollowing out the fun parts, tech remains the best job for a smart, motivated person who's willing to learn new things.

I wonder what Zig would be like as an ILR. Easy cross compilation, plus, you can compile with runtime checks to help debug your compiler output. Might be fun for a sideproject

One hundred? Did I read that right?

Ha! Nice. I never worked with CodeStar / CodeCommit. Was it pretty bad?

How much has the volume increased, from what you know?

So I work for a devtools vendor (Snyk) and 6 months ago I signed into Azure DevOps for the first time in my life

I couldn't believe it. I actually thought the product was broken. Just from a visual perspective it looked like a student project. And then I got to _using_ the damn thing

As an aside, God, Azure DevOps, what a total pile of crap that product is

My "favourite" restriction that an Azure DevOps PR description is limited to a pathetic 4000 characters.

But again, how much of that time is spent writing code?

I've done rewrites, replatforms, a bunch of times. The actual programming is not the tricky part, but instead (1) picking apart the legacy system, understanding what to build, (2) orchestrating the work to shift transactions from service A to service B without breaking anything.

Teams and especially developers love to think they can skip that phase and just crack on with the programming, invariably what happens is the same as described above: intoxicating velocity followed by a hard stop when you realise you haven't solved problems (1) or (2) above

I honestly think this would qualify as "ruinous empathy"

It's fine and even good to assume good faith, extend your understanding, and listen to the reasons someone has done harm - in a context where the problem was already redressed and the wrongdoer is labelled.

This is not that. This is someone publishing a false paper, deceiving multiple rounds of reviewers, manipulating evidence, knowingly and for personal gain. And they still haven't faced any consequences for it.

I don't really know how to bridge the moral gap with this sort of viewpoint, honestly. It's like you're telling me to sympathise with the arsonist whilst he's still running around with gasoline

something something Goodhart's Law

To which my reply would be, we can engage in the analysis after we have taken down the paper.

It's still up! Maybe the answer to building a resilient system lies in why it is still up.

> Labeling people as villains is almost always an unhelpful oversimplification of reality

This is effectively denying the existence of bad actors.

We can introspect into the exact motives behind bad behaviour once the paper is retracted. Until then, there is ongoing harm to public science.

Yes. It's one reason I've lost interest in OSS completely.

I'm doing a project with tree sitter right now

Any tips for keeping the grammar sizes under control? I'm distributing a CLI tool that needs to support several languages, and I can see the grammars gradually bloating the binary size

I could build some clever thing where language packs are opt-in and distributed as WASM, maybe. But that could be complex