Hacker News: elphard

New comment by elphard in "Why Social Media Bans Alone Can't Solve the Age Verification Dilemma"

elphard — Fri, 19 Jun 2026 09:55:26 +0000

Meta made $60 billion in profits last year; they can spend some of that to figure out how to protect children from harmful content on their platform.

New comment by elphard in "A sleep-like consolidation mechanism for LLMs"

elphard — Tue, 26 May 2026 20:23:34 +0000

We should let them sleep with half a brain at a time like migrating birds.

New comment by elphard in "Show HN: A MitM proxy to see what your LLM tools are sending"

elphard — Wed, 28 Jan 2026 23:37:11 +0000

This is fantastic. Claude doesn't make it easy to inspect what it's sending - which would actually be really useful for refining the project-specific prompts.

New comment by elphard in "How the Lobsters front page works"

elphard — Sun, 18 Jan 2026 18:24:39 +0000

Thanks! That's an interesting approach for filtering new joiners.

New comment by elphard in "How the Lobsters front page works"

elphard — Sun, 18 Jan 2026 18:20:06 +0000

How difficult is it to get invited to join Lobsters?

New comment by elphard in "Poking holes into bytecode with peephole optimisations"

elphard — Sun, 18 Jan 2026 16:05:34 +0000

I like that you are treating peephole as a strict fallback after IR optimizations, with a tiny window and single pass. In a lot of compilers this stage turns into a junk drawer of pattern matches that quietly grow until no one remembers why half of them exist.

The opt_trace! hook is the underrated bit here. Once you start rewriting instruction sequences, knowing which patterns fire and how often on real programs is usually more valuable than another synthetic benchmark. Keeping it behind -O1 is a nice way to make sure you only pay for that complexity when users actually opt in.