@glyph Did you quote post something?
Uncategorized
903
Posts
312
Posters
0
Views
-
it's truly amazing what LLMs can achieve. we now know it's possible to produce an html5 parsing library with nothing but the full source code of an existing html5 parsing library, all the source code of all other open source libraries ever, a meticulously maintained and extremely comprehensive test suite written by somebody else, 5 different models, a megawatt-hour of energy, a swimming pool full of water, and a month of spare time of an extremely senior engineer
@glyph If you had that engineer and only about 3kWh of energy and 12 liters of water, could the engineer do it in a week?
-
@glyph If you had that engineer and only about 3kWh of energy and 12 liters of water, could the engineer do it in a week?
-
it's truly amazing what LLMs can achieve. we now know it's possible to produce an html5 parsing library with nothing but the full source code of an existing html5 parsing library, all the source code of all other open source libraries ever, a meticulously maintained and extremely comprehensive test suite written by somebody else, 5 different models, a megawatt-hour of energy, a swimming pool full of water, and a month of spare time of an extremely senior engineer
This is something I really struggle w/, I explain to folks you can only get meaningful results of the problem you are trying to solve or something close to it is well or overrepresented in the training set.
Folks often eventually agree and then go onto to argue things that can't possibly logically follow once you accept this.
It is as-if folks see LLM and their brains just turn it into "magic" somewhere along the way.
-
it's truly amazing what LLMs can achieve. we now know it's possible to produce an html5 parsing library with nothing but the full source code of an existing html5 parsing library, all the source code of all other open source libraries ever, a meticulously maintained and extremely comprehensive test suite written by somebody else, 5 different models, a megawatt-hour of energy, a swimming pool full of water, and a month of spare time of an extremely senior engineer
@glyph But wait! What is this I see in my vibe coded content? Could this be....a fuck off variable? Seriously, people are naming their variables things like "theHTMLDohickey", and those are showing up in "innovative and original" vibe code from LLMs.
-
This is something I really struggle w/, I explain to folks you can only get meaningful results of the problem you are trying to solve or something close to it is well or overrepresented in the training set.
Folks often eventually agree and then go onto to argue things that can't possibly logically follow once you accept this.
It is as-if folks see LLM and their brains just turn it into "magic" somewhere along the way.
-
@RichiH @bornach @GroupNebula563 @glyph t being what here?
-
1. Or maybe the senior engineer would have simply written all the code and it wouldn't have taken that long. Nobody measures this. We don't even know yet if it's a "force multiplier" or a distraction. I've written at length about this phenomenon here: https://blog.glyph.im/2025/08/futzing-fraction.html
2. Or maybe they would have solved the actual social problem instead, i.e. that the original library is insufficiently maintained, rather than rewriting to move the locus of control closer to themselves.
1. We have (for a given value of we, also have). It is. With dragons and foot guns.
2. Great point. One of my favourite techniques for working with LLMs is telling it not to write a thing 😂.
I did enjoy your opening post though.
-
1. We have (for a given value of we, also have). It is. With dragons and foot guns.
2. Great point. One of my favourite techniques for working with LLMs is telling it not to write a thing 😂.
I did enjoy your opening post though.
1. [citation needed]. and not like, sarcastically. I have heard this claim from enthusiasts over and over, but it’s always in secret internal discussions where there’s no methodology (or even results) to evaluate.
meanwhile, in the public scientific record:
https://hackaday.com/2025/07/11/measuring-the-impact-of-llms-on-experienced-developer-productivity/
-
1. Or maybe the senior engineer would have simply written all the code and it wouldn't have taken that long. Nobody measures this. We don't even know yet if it's a "force multiplier" or a distraction. I've written at length about this phenomenon here: https://blog.glyph.im/2025/08/futzing-fraction.html
2. Or maybe they would have solved the actual social problem instead, i.e. that the original library is insufficiently maintained, rather than rewriting to move the locus of control closer to themselves.
-
@glyph It's pretty good.
Like, consistently better than any of the pre-LSP approaches.
-
@glyph It's pretty good.
Like, consistently better than any of the pre-LSP approaches.
@offby1 I am, thus far, pretty disappointed
-
@glyph My main experience of pyright is as a less buggy and broken MyPy.
-
@glyph My main experience of pyright is as a less buggy and broken MyPy.
@glyph That's damnation by faint praise, though...
-
it's truly amazing what LLMs can achieve. we now know it's possible to produce an html5 parsing library with nothing but the full source code of an existing html5 parsing library, all the source code of all other open source libraries ever, a meticulously maintained and extremely comprehensive test suite written by somebody else, 5 different models, a megawatt-hour of energy, a swimming pool full of water, and a month of spare time of an extremely senior engineer
@glyph (I giggled) Worthy meme material 👍 .
-
@GroupNebula563 @bornach @RichiH @glyph also it's just called desalination with extra steps, which is already an expensive process, and I think operating salty pipes carries its own batch of problems. It's already nerve-wracking enough running water around tons of computers (eventually every pipe will need to be replaced, or it'll break...)
-
it's truly amazing what LLMs can achieve. we now know it's possible to produce an html5 parsing library with nothing but the full source code of an existing html5 parsing library, all the source code of all other open source libraries ever, a meticulously maintained and extremely comprehensive test suite written by somebody else, 5 different models, a megawatt-hour of energy, a swimming pool full of water, and a month of spare time of an extremely senior engineer
@glyph We're getting close to building a table of tasks that 2020s AI is good for, and tasks it's not
-
@neurobashing @glyph FWIW Ruff has mostly been waiting for ty to handle multi-file type-aware lint rules. We plan to integrate those bits of ty back into Ruff at some point in this coming year.
-
it's truly amazing what LLMs can achieve. we now know it's possible to produce an html5 parsing library with nothing but the full source code of an existing html5 parsing library, all the source code of all other open source libraries ever, a meticulously maintained and extremely comprehensive test suite written by somebody else, 5 different models, a megawatt-hour of energy, a swimming pool full of water, and a month of spare time of an extremely senior engineer
@glyph Psh, I can do it with *only* the full source code of an existing library.
That’s called efficiency.
-
it's truly amazing what LLMs can achieve. we now know it's possible to produce an html5 parsing library with nothing but the full source code of an existing html5 parsing library, all the source code of all other open source libraries ever, a meticulously maintained and extremely comprehensive test suite written by somebody else, 5 different models, a megawatt-hour of energy, a swimming pool full of water, and a month of spare time of an extremely senior engineer
-
@tuban_muzuru I have no idea what point you are trying to make here.
Gli ultimi otto messaggi ricevuti dalla Federazione
-
@twilliability@genart.social Thanks for sharing this. I didn't know about the bootloader issue—I've been on iPhone for a while now, so I'm a bit out of the loop on Samsung's recent moves.
-
@quinta Oh, bene grazie.
Perché sono anni che ci scalpellano l’uccello con faro culturale, la democrazia matura, gli anticorpi ma con la cultura del fare.
No, non le faranno.
Imploderanno e noi con loro, perché non abbiamo mai voluto fare l’Europa, preferendo la rappresentanza del Molise a Bruxelles, che utilizziamo da sempre come bancomat, e esilio per mezze seghe e antagonisti interni.E abbiamo Di Maio come appresentante speciale dell’UE per la regione del Golfo.
E Otelma muto.
-
@hongminhee Thank you! This is entirely in line which I believe must have happened around the EU's bootloader unlock ban. Ie it can only be the result of severe "lobbying" (to put it mildly) by Samsung in the background.
It's sad, because it removes consumers' right to repair, tinker, and extend device lifetimes.
-
@mcc oh wow I just looked him up and I feel so ignorant.
-
Current* conditions near Duluth, MN:
-
@mcc Charles Martinet? The voice of Bit Trip Runner 2 Future Legend of Rhythm Alien?
-
Siria: prosegue la resistenza dei quartieri a maggioranza curda. la testimonianza di un anziano in partenza per sostenere la resistenza
@anarchia
Quinto giorno in cui i quartieri autogovernati a maggioranza curda – Sheikh Maqsoud, Ashrefye e Bani Zeid – sono sotto attacco da parte delle milizie salafite ora
-
@mariosiniscalchi
«Ma Enrico Mattei è anche quello che finanzia i partiti con fondi neri (sono sue le parole «per me i partiti sono come taxi: ci salgo, li pago e poi scendo»), applica metodi discutibili per costruire gasdotti in breve tempo ed elabora strategie piuttosto creative per assicurarsi un accordo per il petrolio iraniano.»https://www.ilpost.it/2025/04/10/la-storia-di-enrico-mattei/
