Hacker News: vbarrielle

New comment by vbarrielle in "πFS"

vbarrielle — Thu, 11 Jun 2026 08:14:40 +0000

Chaitin's constant does not count? Depends on your definition of constructed, but contrary to "easy" normal numbers such as Champernowne's constant, it's not defined by its sequence of digits.

New comment by vbarrielle in "πFS"

vbarrielle — Thu, 11 Jun 2026 08:04:30 +0000

This would be easier using the Champernowne constant (https://en.wikipedia.org/wiki/Champernowne_constant) which is guaranteed to be normal, not just conjectured.

New comment by vbarrielle in "Who owns the code Claude Code wrote?"

vbarrielle — Wed, 29 Apr 2026 15:40:55 +0000

I thought fair use was decided on a case by case basis, and could not be guaranteed? If true, wouldn't that mean that in other cases it could be ruled differently?

New comment by vbarrielle in "Acetaminophen vs. ibuprofen"

vbarrielle — Wed, 22 Apr 2026 14:56:14 +0000

No but you can get it for free with a prescription.

New comment by vbarrielle in "F-15E jet shot down over Iran"

vbarrielle — Fri, 03 Apr 2026 15:44:53 +0000

Sounds like it indeed. The balance was... interesting, a single tank could not win against a dozen cavemen.

New comment by vbarrielle in "How the AI Bubble Bursts"

vbarrielle — Mon, 30 Mar 2026 14:39:52 +0000

Openrouter is a startup, what's the indication it serves token at a profit? It could be serving them at a loss to show growth.

New comment by vbarrielle in "ARC-AGI-3"

vbarrielle — Thu, 26 Mar 2026 16:47:31 +0000

> For example, in a variant of environment TR87, Opus 4.6 scores 0.0% with no harness and 97.1% with the Duke harness (12), yet in environment BP35, Opus 4.6 scores 0.0% under both configuration

This is with a harness that has been designed to tackle "a small set of public environments: ls20, ft09, and vc33" (of the arc-agi-3 challenge), yet it looks like it does not solve the full arc-agi-3 benchmark, just some of them.

New comment by vbarrielle in "Is legal the same as legitimate: AI reimplementation and the erosion of copyleft"

vbarrielle — Tue, 10 Mar 2026 10:47:14 +0000

No, APIs fall under copyright, but the Supreme Court found that Google's reimplementation of Java's API was falling under fair use. Fair use is decided case by case, one cannot use that decision as a precedent.

New comment by vbarrielle in "Is legal the same as legitimate: AI reimplementation and the erosion of copyleft"

vbarrielle — Tue, 10 Mar 2026 09:15:55 +0000

Not quite in my opinion. The output of an LLM from a simple prompt falls into the public domain, but if you also give a copyrighted work as input, the mechanistic transformation performed will not alter the original license (same as encoding a video does not change its license).

New comment by vbarrielle in "Is legal the same as legitimate: AI reimplementation and the erosion of copyleft"

vbarrielle — Tue, 10 Mar 2026 09:05:24 +0000

Well I'm implying that someone who's been reading a codebase for 10+ years is the worst person to claim an "independent reimplementation".

New comment by vbarrielle in "Is legal the same as legitimate: AI reimplementation and the erosion of copyleft"

vbarrielle — Mon, 09 Mar 2026 21:44:03 +0000

There's a difference between "I've read a LGPL code once, maybe I could do something similar" and "I've been reading this LGPL code for 12 years and now I'm going to do exactly the same thing".

New comment by vbarrielle in "Is legal the same as legitimate: AI reimplementation and the erosion of copyleft"

vbarrielle — Mon, 09 Mar 2026 21:34:31 +0000

Everyone writes as if he just fed the spec and tests to Claude Code. Ignoring for now that the tests are under LGPL as well, the commit history shows that this has been done with two weeks of steering Claude Code towards the desired output. At every one of these interactions, the maintainer used his deep knowledge of the chardet codebase to steer Claude.

New comment by vbarrielle in "Is legal the same as legitimate: AI reimplementation and the erosion of copyleft"

vbarrielle — Mon, 09 Mar 2026 21:28:23 +0000

Google vs Oracle ruled that APIs fall under copyright (the contrary was thought before). However, it was ruled that, in that specific case, fair use applied, because of interoperability concerns. That's the important part of this case: fair use is never automatic, it is assessed case by case.

Regarding chardet, I'm not sure "I wanted to circumvent the license" is a good way to argue fair use.

New comment by vbarrielle in "AI and the Ship of Theseus"

vbarrielle — Fri, 06 Mar 2026 09:20:48 +0000

The test suite was also licensed under the LGPL. The reimplementation can be seen as a derivative work of the test suite, and thus should fall under the LGPL. This does not even mention the fact that the coding agent, AND the user steering it, both had ample exposure to chardet's source code, making it hard to argue that the reimplementation is a new ship.

New comment by vbarrielle in "Relicensing with AI-Assisted Rewrite"

vbarrielle — Thu, 05 Mar 2026 14:21:19 +0000

Even the most permissive open source licenses such as MIT require attribution. Releasing as open source would therefore benefit the author through publicity. Bein able to say that you're the author of library X, used by megacorp Y with great success, is a good selling point in a job interview.

LLM ripping off open source code removes that.

New comment by vbarrielle in "British Columbia is permanently adopting daylight time"

vbarrielle — Tue, 03 Mar 2026 13:11:59 +0000

I'm in strong agreement with this. Even though I'd prefer winter time all year round, I would rather enjoy permanent summer time over switching twice a year. And I'm living in France, so my "winter time" is actually already a DST compared to the standard time (France's timezone should be UTC like the UK, but WW2 changed that to UTC+1 and we never switched back), so the "summer time" is actually a "double DST".

New comment by vbarrielle in "GrapheneOS – Break Free from Google and Apple"

vbarrielle — Tue, 17 Feb 2026 14:59:07 +0000

Unfortunately it needs push notifications to authorize online payments.

New comment by vbarrielle in "GrapheneOS – Break Free from Google and Apple"

vbarrielle — Tue, 17 Feb 2026 14:57:21 +0000

TOTP not accepted, because the confirmation for payment must include the amount to be paid, which cannot be done under TOTP as far as I know.

New comment by vbarrielle in "Thinking hard burns almost no calories but destroys your next workout"

vbarrielle — Tue, 17 Feb 2026 09:55:49 +0000

Not running, but in cycling we have power meters, and some workouts (eg 2 x 20' threshold) will definitely burn in the range of 800 calories in an hour. The energy measured by the power meter for this workout is 800 kJ for me (my threshold being around 260W). Now it turns out the conversion factor from kJ to calories is 1/4, but the body is only 25% efficient when producing calories for cycling, meaning one has to burn 4x the amount measured by the power meter. So that's 800 calories for this kind of workout, for me. I wouldn't be surprised if runners of similar fitness doing similar workouts had the same energy expenditure.

New comment by vbarrielle in "GPT-5.2 derives a new result in theoretical physics"

vbarrielle — Fri, 13 Feb 2026 20:29:27 +0000

I' m far from being an LLM enthusiast, but this is probably the right use case for this technology: conjectures which are hard to find, but then the proof can be checked with automated theorem provers. Isn't it what AlphaProof does by the way?