Hacker News: SonOfLilit

New comment by SonOfLilit in "Rise of the Triforce"

SonOfLilit — Tue, 17 Feb 2026 12:07:08 +0000

"Heck" is to "Hell" like "Darn" is to "Damn" (or "Freaking" to “F--ing") - a word that sounds similar but is more polite, to be said in public in more religious places and times.

New comment by SonOfLilit in "Ask HN: What did you find out or explore today?"

SonOfLilit — Thu, 15 Jan 2026 11:24:14 +0000

Gotta say, I assumed this is some sort of virtual/imaginary thing, but it seems like there's a point in the optical system where if we placed a screen, we'd see the FT of the image coming in! And before we had digital image processing people used to place masks there to filter out low frequency or high frequency details in the image. Which is absolutely insane and I have no comprehension of how the physics works out!

New comment by SonOfLilit in "Debian's Git Transition"

SonOfLilit — Tue, 23 Dec 2025 11:04:41 +0000

The hacker in me is very excited by the prospect of pypi executing code from my packages in the system that builds everyone's wheels.

New comment by SonOfLilit in "You are the scariest monster in the woods"

SonOfLilit — Thu, 16 Oct 2025 09:53:49 +0000

It's an argument that we don't need to understand LLMs to build them. But mostly it's just a statement of fact to promote correct understanding of our world.

New comment by SonOfLilit in "You are the scariest monster in the woods"

SonOfLilit — Thu, 16 Oct 2025 09:49:31 +0000

Have you tried giving it basic logic that isn't in its training data?

I have. gpt-3.5-instruct required a lot of prompting to keep it on track. Sonnet 4 got it in one.

Terrence Tao, the most prominent mathematician alive, says he's been getting LLM assistance with his research. I would need about a decade of training to be able to do any math in a day that Tao can't do in his head in less than 5 seconds.

LLMs are suffer from terrible, uh, dementia-like distraction, but they can definitely do logic.

New comment by SonOfLilit in "You are the scariest monster in the woods"

SonOfLilit — Wed, 15 Oct 2025 16:12:58 +0000

Evolution took millions of years to reach the minimal thing that can read, write and solve leetcode problems, so by similar logic we're at least a few million years in...

New comment by SonOfLilit in "You are the scariest monster in the woods"

SonOfLilit — Wed, 15 Oct 2025 16:11:48 +0000

Oh, boy, are you in for a surprise.

Our theories of how LLMs learn and work look a lot more like biology than math. Including how vague and noncommittal they are because biology is _hard_.

New comment by SonOfLilit in "How Apple designs a virtual knob (2012)"

SonOfLilit — Sat, 11 Oct 2025 23:26:32 +0000

This is not an Apple original design, this is standard fare in DAWs and VST plugins and has been since at least the early 00s. In the beginning of the article he talks about context menus as something that is not one GUI's but just standard in the industry - these knob interactions are like that for the audio industry.

New comment by SonOfLilit in "Under the hood: Vec"

SonOfLilit — Thu, 09 Oct 2025 12:17:25 +0000

When I was young I peeked into the C++ stdlib, which is probably the best thing to compare this to. It was orders of magnitude worse.

New comment by SonOfLilit in "The React Foundation"

SonOfLilit — Thu, 09 Oct 2025 12:02:31 +0000

Are you implicitly complaining that React is not moving fast enough? What the JS ecosystem needs is for some big players to CHILL a bit and take backwards compat more seriously.

New comment by SonOfLilit in "We bought the whole GPU, so we're damn well going to use the whole GPU"

SonOfLilit — Thu, 02 Oct 2025 14:16:38 +0000

This used to be less true at the time Carmack said it :>

New comment by SonOfLilit in "Qwen3-Next"

SonOfLilit — Fri, 12 Sep 2025 10:14:31 +0000

If you ask me to guess an answer, I'll _usually_ produce the same answer as if I had time to think about it deeply, but not always...

Show HN: Testing Masterclass

SonOfLilit — Thu, 11 Sep 2025 17:59:43 +0000

I ran this yesterday at PyCon IL.

It has some very nice exercises that you can try at home, especially the LLM evals exercise.

Posting it as a Show mostly because I'm proud of the Excalidraw aesthetics which many of you might appreciate.

If there's interest I can run it in English online. Shoot me an email if you're interested.

Comments URL: https://news.ycombinator.com/item?id=45214333

Points: 1

# Comments: 0

New comment by SonOfLilit in "Almost anything you give sustained attention to will begin to loop on itself"

SonOfLilit — Thu, 04 Sep 2025 17:13:36 +0000

Mandarin Chinese: 注意 (zhùyì) - "note/record intention" Spanish: prestar atención - "lend attention" English: pay attention - "give/spend attention" Hindi: ध्यान देना (dhyaan dena) - "give meditation/focus" Arabic: انتبه (intabih) - "be alert/awake" ...

https://pastebin.com/3ghPnjb9

New comment by SonOfLilit in "Almost anything you give sustained attention to will begin to loop on itself"

SonOfLilit — Thu, 04 Sep 2025 16:15:33 +0000

Huh, I never placed my heart to it

New comment by SonOfLilit in "Almost anything you give sustained attention to will begin to loop on itself"

SonOfLilit — Thu, 04 Sep 2025 16:13:17 +0000

I enjoyed your words and pixels!

New comment by SonOfLilit in "A Linux version of the Procmon Sysinternals tool"

SonOfLilit — Mon, 01 Sep 2025 08:52:51 +0000

They are celebs in windows security. Mark's name is synonymous with windows internals, he wrote the definitive textbook.

New comment by SonOfLilit in "Eternal Struggle"

SonOfLilit — Sun, 31 Aug 2025 20:49:20 +0000

There are stableish equilibria that are not 50-50, e.g. one color having a donut around the other color that has a donut hole.

New comment by SonOfLilit in "Ask HN: Best codebases to study to learn software design?"

SonOfLilit — Sun, 24 Aug 2025 07:25:32 +0000

To answer your actual question:

The codebases I learned from the most are Git, Postgres, CPython. Not saying they are perfect designs, but they are well maintained, solve hard problems, have seen many years of evolution, and are very easy to get your hands on.

New comment by SonOfLilit in "Open Lovable"

SonOfLilit — Sun, 10 Aug 2025 20:18:51 +0000

Registering your trademark makes enforcement easier, but there's a reason both TM and R exist - unregistered trademarks still hold power.