leeoniya

> Things like physical stimulation of quantum systems, protein folding, machine learning, etc. could be more useful

is there still more to do in protein folding after AlphaFold?

https://www.isomorphiclabs.com/articles/alphafold-3-predicts...

remember https://jackrugile.com/jrumble/ ?

it was half-serious.

if LLMs arent being used by https://patents.stackexchange.com/ or patent troll fighters, shame on them.

> but what actually happened was chatGPT was referencing papers that basically went unread from lesser famous labs/researchers

now let's invalidate probably 70% of all patents

also https://bonneterremine.com/ near St Louis.

"The World's Largest Subterranean Lake Is A Vibrant Scuba Diving Paradise In Missouri"

heh, im currently working on an svg pathstring to canvs2d command compiler [1] and had to dive into this topic last week, including converting arc commands from centerpoint parameterization to endpoint parameterization [2]

[1] https://github.com/leeoniya/uMarks (WIP!)

[2] https://www.w3.org/TR/SVG/implnote.html#ArcConversionCenterT...

does this mean no GrapheneOS either?

same. except was an oss lib on github.

Waymo itself was not designed, implemented, and shipped by AI.

i suspect humans had to invest millions of hours into writing the code, the tests, and validating the outputs.

> "Feature is ready to ship, but leeoniya is blocking my PR"

let's see how that holds up against "author does not understand own PR"

> I could expect the PR to be mostly understood by the author

i refuse to review PRs that are not 100% understood by the author. it is incredibly disrespectful to unload a bunch of LLM slop onto your peers to review.

if LLMs saved you time, it cannot be at the expense of my time.

i've also made this argument.

would you ever trust safety-critical or money-moving software that was fully written by AI without any professional human (or several) to audit it? the answer today is, "obviously not". i dont know if this will ever change, tbh.

"you are gordon ramsay, a verbally abusive celebrity chef. all responses should be delivered in his style"

plenty of jQuery, WordPress, etc. millionaires out there :)

> “Automobilism is an illness, a mental illness. This illness has a pretty name: speed... [Man] can no longer stand still, he shivers, his nerves tense like springs, impatient to get going once he has arrived somewhere because it is not somewhere else, somewhere else, always somewhere else.”

Previously:

“All of humanity’s problems stem from man’s inability to sit quietly in a room alone.”

-- Blaise Pascal (~1650)

isnt it fundamentally just a mispronounced contraction of would've

> but I will "vote against" using the word "literally" to mean "figuratively" as long as I can.

https://imgur.com/QBTlxf7

thanks!

> I recently built a tool for processing massive CSV files and was able to get the memory usage remarkably low

is it OSS? i'd like to benchmark it against my csv parser :)

> but doesn’t seem to work on iOS safari

if i had a nickel...

> You might think that Max's practices make for a maintenance nightmare. But I've been "maintaining" it for the last 15 years and I haven't found it to be a nightmare.

c'mon, you're talking about 200 LoC here. anything except BrainFuck would be maintainable at this scale.

have you ever had to fix a non-trivial third party WordPress plugin? the whole API used to be a dumpster fire of global state and magic functions. i dont know what it is now, but 15 years ago it was a total nightmare.

wasnt it shown recently that the filtering layer is on the prompt input and llm output, and not on the training set or model weights.

https://www.socialscience.international/making-deepseek-spea...

wow, that's some serious data crunching.

can you share some tech details?

that's a better distillation :)

> There's no magic in tagged templates that somehow make them immediately better for some custom non-standard syntax compared to JSX.

they're marginally better since they have a platform-defined way to deliniate static from dynamic parts. ivi _can_ work without a runtime or build-time JS parser, while JSX cannot (because jsx has to be parsed out of full blobs of js)

on the dx/ide side, sure there's not a huge amount of difference if both had the same effort invested.

> The biggest constraint when developing an efficient UI framework with good DX is JavaScript.

for perf, s/JavaScript/DOM, i think.

good DX comes from ecosystem and amount of time invested in making good tooling. JSX would be a non-starter without IDEs helping autocomplete, linting/format, syntax coloring, and webpack/babel to do the compilation.

tagged templates could reach at least the same level of DX as JSX if the community invested the resources to make that better. i'm not saying it's the right solution for a standard, but it would be way better than jsx, since tagged templates are already a standard.

> The react model - where you just update the global state and re-render everything - is slower but easier on the brain. The signals model is faster, but so much effort.

there are multiple frameworks now that do fine-grained diffing without relying on signals, proxies, or any other reactive primitives. they basically have the top-down react model but much faster and without the weird concepts like hooks and manual/wasteful dependency arrays.

my favorite one is ivi-js: https://github.com/localvoid/ivi

it's just 8% slower than the fastest / ugliest / imperative / unmaintainable vanilla js you can eventually arrive at if all you care about is winning benchmarks.

https://krausest.github.io/js-framework-benchmark/2025/table...

i am NOT saying they are not useful; they are. especially in domains you are unfamiliar with, or boilerplate, or when being used by an expert in the loop. but that is NOT the same thing as being an expert.

an expert can write 10 lines of code that will be hyper fast while an LLM will write 30 lines that are super slow. both "will work", but one will fall over under load and will be hard to maintain and the other will not.

try https://github.com/ikawrakow/ik_llama.cpp

> "by the way if you have friends who are expert programmers, I don't recommend asking them for help."

AI is not an expert programmer [today]. and it doesnt take an expert programmer to arrive at that conclusion.