Hacker News: theLiminator

New comment by theLiminator in "Show HN: Graphicle, a Rust port of Graphify with memory for coding agents"

theLiminator — Wed, 14 Oct 2026 18:06:02 +0000

Has anyone actually gotten good results out of this?

For the evals I've run (albeit, not super scientific), these memory systems did worse than the raw harness given ripgrep.

New comment by theLiminator in "GPT-6 Astra helped me jailbreak my Kindle"

theLiminator — Tue, 13 Oct 2026 02:07:49 +0000

Yes for most devices people care to jailbreak though.

New comment by theLiminator in "I turned my unused Kindle into a photo frame with GPT-6 Astra"

theLiminator — Tue, 13 Oct 2026 00:41:55 +0000

Not exactly imo, you could spend a crapton of money with white box pen testing with a bunch of frontier llms, and it might become a bunch harder to break into (it generally costs more to attack than it does to defend). Not to mention formal verification might become possible for more devices which might make it nearly impossible to find exploits.

New comment by theLiminator in "Supabase is acquiring Turso"

theLiminator — Fri, 02 Oct 2026 20:09:26 +0000

Yeah, I've seen a lot of technology killed by acquisitions...

For example, https://github.com/Mooncake-Labs/pg_mooncake

New comment by theLiminator in "GPT-6 Astra, looped transformers, and hidden reasoning"

theLiminator — Wed, 09 Sep 2026 17:15:25 +0000

I wish someone ran some sort of representative benchmark suite every X days to see if this occurs.

New comment by theLiminator in "We found a division by zero bug in FFmpeg with a vibecoded fuzzer"

theLiminator — Thu, 27 Aug 2026 20:47:18 +0000

Perhaps coq/agda/idris/etc.

New comment by theLiminator in "We found a division by zero bug in FFmpeg with a vibecoded fuzzer"

theLiminator — Thu, 27 Aug 2026 20:47:05 +0000

Imo, formal methods like more expressive/stricter type systems are key to making LLM generated code successful. Of course models will get better, but trusting the output will become much easier with a type system that proves more properties.

New comment by theLiminator in "2027 memory capacity is reportedly sold out"

theLiminator — Fri, 07 Aug 2026 23:30:33 +0000

Yeah, local LLMs are probably an order of magnitude less efficient at a fixed level of "intelligence" if not more.

New comment by theLiminator in "New Mexico court orders Meta to pay $567m over harms to children’s mental health"

theLiminator — Fri, 07 Aug 2026 17:46:30 +0000

That's probably going to lead us to trusted attestation age verification.

New comment by theLiminator in "Show HN: Capy – A Git-style platform for managing your team's secrets"

theLiminator — Wed, 05 Aug 2026 21:00:03 +0000

Looks pretty cool, what steps are you taking to ensure that the backend is secure?

New comment by theLiminator in "Algorithms on billion-scale graph using 10GB RAM: I love DataFusion"

theLiminator — Fri, 31 Jul 2026 17:05:58 +0000

DataFusion is really cool, it's kind of like the LLVM of the OLAP world.

New comment by theLiminator in "Claude Opus 5"

theLiminator — Fri, 24 Jul 2026 18:26:22 +0000

It depends on your quality bar. At a fixed level of quality, given a high reasoning sonnet vs a low reasoning opus, the low reasoning opus tends to be pareto optimal.

It's only when you need even lower levels of cost than opus at zero to low reasoning when sonnet starts to make sense at all.

New comment by theLiminator in "Bonsai 27B: A 27B-Class model that runs on a phone"

theLiminator — Tue, 14 Jul 2026 20:25:56 +0000

This is useful research, but this particular model itself is likely absolutely useless.

New comment by theLiminator in "Separating signal from noise in coding evaluations"

theLiminator — Mon, 13 Jul 2026 18:36:00 +0000

What makes something `True AI`?

What's your test for it?

New comment by theLiminator in "Separating signal from noise in coding evaluations"

theLiminator — Wed, 08 Jul 2026 23:45:56 +0000

Please define AGI first.

New comment by theLiminator in "Rewriting Bun in Rust"

theLiminator — Wed, 08 Jul 2026 23:35:24 +0000

Maybe something like Hylo? But personally I don't see anything displacing rust for the next few years, as I think there's enough rust in the training data for it to be the best "serious" language for agentic systems-level development.

It's really the only systems language in its exact niche.

New comment by theLiminator in "Rewriting Bun in Rust"

theLiminator — Wed, 08 Jul 2026 23:21:59 +0000

That's the power of a strong test suite. LLMs excel when you have verifiable rewards. I imagine we'll get a lot more rewritten in rust projects in the future. Rust is also an ideal target for such rewrites as it offers a lot of verification (via its type system) and is low overhead with zero-gc. There's less and less reason to use GC'd languages in the agentic coding era.

I think Rust is a locally optimal target for LLM coding, we might see a better language in the future, but I think Rust will dominate for quite some time.

New comment by theLiminator in "Why skilled workers come to Germany and then leave again"

theLiminator — Tue, 07 Jul 2026 19:45:00 +0000

> Much more than any other country on Earth.

What about Canada?

New comment by theLiminator in "No LLM Code in Dependencies"

theLiminator — Thu, 02 Jul 2026 20:29:46 +0000

If anything I think discipline and rigor will go up.

I think it will force us to adopt stronger type systems, formal methods, and more automated verification.

New comment by theLiminator in "Claude Sonnet 5"

theLiminator — Tue, 30 Jun 2026 18:11:07 +0000

Seems like the way to go for any smaller models is to only use the low reasoning levels, and for anything where you'd want it to reason harder, to just use a larger model.

In effect, high reasoning only makes sense when you're using the frontier model and need extra performance (higher levels of reasoning are never pareto optimal unless you're at the largest model size).