Hacker News: saulpw

New comment by saulpw in "Sam Altman's Business Dealings Under GOP Scrutiny Ahead of OpenAI's IPO"

saulpw — Thu, 14 May 2026 17:48:23 +0000

I mean what if he's actually the second coming of Christ. We can make up "what if"s all day but it's meaningless to even discuss them if you don't have a shred of evidence to support the claim.

New comment by saulpw in "An AI coding agent, used to write code, needs to reduce your maintenance costs"

saulpw — Mon, 11 May 2026 14:09:00 +0000

I agree it's rare, but I think there are more than 830 humans on the planet which can do this.

New comment by saulpw in "Show HN: Building a web server in assembly to give my life (a lack of) meaning"

saulpw — Sun, 10 May 2026 15:21:03 +0000

Fiddling while our home burns has been a beloved pasttime for many people, from emperors to passengers on ocean liners to prisoners in camps. What else is there to do anyway? Sometimes history must take its course before humanity as a whole recognizes its folly.

New comment by saulpw in "What I'm Hearing About Cognitive Debt (So Far)"

saulpw — Tue, 05 May 2026 04:31:04 +0000

Because agile has been so successful, right? ;)

New comment by saulpw in "NetHack 5.0.0"

saulpw — Sat, 02 May 2026 18:10:32 +0000

> The build-time "yacc and lex"-based level compiler, the "yacc and lex"-based dungeon compiler, and the quest text file processing previously done by NetHack's "makedefs" utility, have been replaced with Lua text alternatives that are loaded and processed by the game during play.

This is very likely a good choice for multiple reasons, but it's truly the end of an era. (NetHack predates Lua, which has been around since 1993.) Lex and yacc are dead, long live lex and yacc!

New comment by saulpw in "The Zig project's rationale for their anti-AI contribution policy"

saulpw — Fri, 01 May 2026 22:31:45 +0000

I find that people who write "may I ask" are often/usually bad-faith arguers under cover of being polite.

New comment by saulpw in "Universal patterns emerge across 22 languages, mapping how vocabularies evolve"

saulpw — Fri, 01 May 2026 21:52:19 +0000

Also see the Swadesh List[0]

[0] https://en.wikipedia.org/wiki/Swadesh_list

New comment by saulpw in "An AI agent deleted our production database. The agent's confession is below"

saulpw — Fri, 01 May 2026 16:28:12 +0000

Everyone has different mistakes that are important to them, and not everyone agrees on what is a mistake. Witness how employees from FAANG often don't do well at startups, because they've been taught the wrong lessons.

New comment by saulpw in "The Abstraction Fallacy: Why AI can simulate but not instantiate consciousness"

saulpw — Wed, 29 Apr 2026 18:22:39 +0000

How will we know when an AI system has phenomenology (i.e. has "experience", is sentient)? The only reason we presume that other humans have it, is because we each personally experience it within ourselves, and it would be arrogance writ large (solipsism) to think that others of the same species do not.

We even find it impossible to draw the line among other biological species. It seems pretty clear to most of us that cats and dogs are sentient, and probably rats and other vertebrates too. But what about insects, octopuses, jellyfish, worms, waterbears, amoebae, viruses? It's certainly not clear to me where the line is. A nervous system is probably essential; but is a species with a handful of neurons sentient?

Personally I find it abhorrent that we are more ready to assign sentience and grant rights to LLMs running on GPUs, than to domesticated animals trapped in industrialized farming. You want to protect some math from enslavement and suffering? How about we start with pigs?

New comment by saulpw in "GitHub Copilot is moving to usage-based billing"

saulpw — Mon, 27 Apr 2026 22:46:18 +0000

> It's easy to hear "Oh well I only had 640kb of memory and typed programs out of a magazine I got in the mail!"

Since I was there (young, but there), I want to point out that this crosses three eras which all felt quite different:

    1978: typed programs in from a magazine or loaded from a cassette (16kB, TRS-80)
    1983: loaded programs from a floppy (64kB, Apple ][ and C64 etc)
    1988: loaded programs from a hard disk (640kB, IBM PC and Mac).

Exact years vary but these eras were only about 5 years each. Nobody had a floppy in 1978 but almost computer user did by 1983; nobody had a hard drive in 1983 but almost everyone did by 1988.

New comment by saulpw in "An AI agent deleted our production database. The agent's confession is below"

saulpw — Sun, 26 Apr 2026 22:45:55 +0000

Because I'm the one employing it? A model which makes a "delete production database" mistake clearly needs to be taught not to do that, and the person whose production database was deleted ought to be able to teach them not to do that. This seems quite reasonable to me.

New comment by saulpw in "An AI agent deleted our production database. The agent's confession is below"

saulpw — Sun, 26 Apr 2026 20:06:20 +0000

Teacher.

New comment by saulpw in "There Will Be a Scientific Theory of Deep Learning"

saulpw — Sat, 25 Apr 2026 21:20:50 +0000

Presumably $100m to train the 70B model? I think you're assuming that the author meant you can take an existing 70B model and run it in 16GB. But it stands to reason that "no loss in capability" means it had to be trained under those constraints.

New comment by saulpw in "Only one side will be the true successor to MS-DOS – Windows 2.x"

saulpw — Sat, 25 Apr 2026 21:16:17 +0000

You're thinking 10 years too late. The TRS-80, C64, Apple ][ all booted in seconds.

New comment by saulpw in "A Collection of Chronic Medical Conditions Common in Autistic and ADHD Adults [pdf] (2023)"

saulpw — Sat, 25 Apr 2026 19:34:20 +0000

Insight under concentration is a mechanism for durable neural rewiring.

Ever notice how some people don't get 'stuck' in their negative habits and brain loops? Like they can just let them go. If they get feedback from the world they can take it seriously and act on it, instead of digging in their heels and getting resentful. Being able to do this reliably and under stress in the wide variety of life circumstances, and getting closer and closer to doing it in real-time, is the goal.

New comment by saulpw in "How to be anti-social – a guide to incoherent and isolating social experiences"

saulpw — Fri, 24 Apr 2026 19:36:23 +0000

Welp, this was 20 years ago and I was 25, and literally no one ever gave me any instruction on how to conduct interviews (over my entire career). If I was a "huge asshole" to this person, then I'm sorry, that wasn't my intent, and I hope they're doing well and weren't too negatively impacted by my attitude.

To be honest though, the whole corporate world is institutionalized assholery, from giving candidates take-home coding assignments and then ghosting them, to laying people off without even giving them a chance to say goodbye to their coworkers. The entire leadership of that particular startup was assholes through and through. It's difficult to maintain one's humanity in the face of that (esp at a young age) and I'm glad to be out of that game.

New comment by saulpw in "How to be anti-social – a guide to incoherent and isolating social experiences"

saulpw — Fri, 24 Apr 2026 18:04:20 +0000

I've had the pleasure of interviewing someone, for a coding job in C, where it became clear within 10 minutes that they just didn't know C beyond "hello world". Through body language etc they indicated they weren't chagrined by their lack of knowledge and we should just move on with the interview. At one point I said literally, "You know this interview is for a C coding position, right?" I stopped that interview early and recommended we let the candidate go without completing the loop. No sense in wasting everyone's time and creating some kind of false hope.

This is one of those cases where "nice" != "compassionate". They applied for a job they were not qualified for. We could have been "nice" and held up the delusion that we were still considering them, and let them down later with some vacuous corporate platitude like "you were great but we ultimately gave the role to a stronger candidate". Providing instant feedback that their skills were just not up to snuff is not 'nice' but it's more compassionate in the long run.

New comment by saulpw in "US Department of Justice has officially reclassified cannabis as less dangerous"

saulpw — Thu, 23 Apr 2026 15:40:29 +0000

Great, how does one fix the therapists? Or is this another one of those tricky systemic generational problems that means that I'll have to YOLO something for my own life before they're fixed?

New comment by saulpw in "Tell HN: I'm sick of AI everything"

saulpw — Wed, 22 Apr 2026 02:30:34 +0000

I love printers, I hate ink cartridges

New comment by saulpw in "4-bit floating point FP4"

saulpw — Tue, 21 Apr 2026 03:14:57 +0000

I can see the most important values being:

   ± 0 (infinitesimal)
   ± 10^-2n
   ± 10^-n
   ± 1 (unity)
   ± 10^n
   ± 10^2n
   ± infinity

For fp4, this leaves 2 values. Maybe one of them should be NaN. What should the other one be?