lifthrasiir

Joined 16 years ago

Karma
14,217
Submissions
4,124
Member since
Feb 20
Kang Seonghoon aka lifthrasiir and many other nicknames. (Twitter @senokay, Bluesky @mearie.org, Github @lifthrasiir, https://mearie.org/) Views my own, I'm not a laywer (nor any other occupation that might require professional qualifications), your mileage may vary, yadda yadda yadda.

Brief bio: Born in South Korea. Was a wannabe programming language researcher until I realized I'm not fit for the academia. Went to several startups with various duties. Then became a gamedev---primarily server frameworks and infrastructure but did many others too---in a big company for nearly a decade until I finally burnt out. Then worked for an ML accelerator startup doing tons of systems programming. Still doing strange things over the course: one winning IOCCC entry, a full-fledged type checker for Lua, several influential Rust libraries including Chrono, compression stuffs like Roadroller and J40 etc. Known for needlessly big walls of text.

[ my public key: https://keybase.io/lifthrasiir; my proof: https://keybase.io/lifthrasiir/sigs/dGcjGsucdiIiTpeVc7WH5YfQNZnkypj-azYwkMJH60Y ]

View on Hacker News

Recent submissions

Comment · 15 days ago
Comment · 23 days ago
Presumably because Rust is more suitable than Go in agentic coding? (I'm not sure, but that seems the purported answer.)
Comment · 1 month ago
Comment · 1 month ago
The former also conveys that you are rude, which is not what you usually want.
Comment · 1 month ago
I was about to immediately suggest UCS-X, and then pleased that the author offers a fair comparison to UCS-X and many other alternatives. Great. Too bad UCS-X is possibly more memorable name.
Comment · 1 month ago
Comment · 2 months ago
Apart from the usual hypocrisy,

> Safety controls may be lost – models trained on Claude's Outputs won't have our safety measures, potentially leading to harmful or dangerous AI systems. We also have no visibility into deployment, meaning we cannot monitor how these distilled models are used or prevent misuse.

So was that why your models caused three real-world security incidents?

Comment · 3 months ago
Comment · 3 months ago
The dilemma is hypothetical (because we now know the exact precision requirement for those functions over all 64-bit doubles) but let's assume that we don't know that and also we happen to use decimals instead. What if, some f(x) is known to be, say, 123.4999...9997 with 1,000 fractional digits? You have to calculate at least 1,000 decimal digits of f(x) in order to correctly round, before that we only know it's between 123.4999...999 and 123.5000...001. There seems no mathematical theory that specially prevents this annoying possibility, hence the dilemma.
Comment · 3 months ago
Common sense testing only tells about the presence of common sense, that's it. It is clear from recent events that LLMs do definitely possess some sort of intelligence even though it might have been polarized to certain domains.
Comment · 3 months ago
It is a project-wide concern, so the stake is obviously high.
Comment · 3 months ago
They learned English or gave up.
Comment · 3 months ago
All of which have been challenged though. As the stake is high, any such reason (for or against LLMs, to be fair) has to be backed with Wikipedia-level evidences at the very least.
Comment · 3 months ago

Live from the official Hacker News API

A modern reader for Hacker News · SvelteKit + Skeleton