iainmerrick

4 ms·

1.
by iainmerrick 10d ago
I don't think we are ever going to see eye to eye, but thanks for trying to explain.
2.
by iainmerrick 10d ago
You're hanging a lot of weight on birth rates to assert that they are wrong and therefore we are right . A birth rate above replacement isn't endlessly sustainable either, right? And "the west" isn't really homog
3.
by iainmerrick 10d ago
Okay, one viewpoint is that it's broadly true , maybe with some exaggeration by the filmmakers and/or interviewees. It's very understandable that soldiers would want to stay anonymous in that version. I don't think many
4.
by iainmerrick 11d ago
Yes - this is a very important story, but the "Reuters Connect" page is completely awful. Hey, let's present our story as a series of soundbites over emotive imagery and ominous music, that definitely evokes serious journalis
5.
by iainmerrick 15d ago
this did not imply that the technology driving them was inferior Does this imply that with $2 billion today, plus or minus $1 billion, one can achieve all these accomplishments and grow into a large technology company? You're the o
6.
by iainmerrick 16d ago
And don’t feel you need to go straight to GEB, either. I think his collection of Scientific American columns, Metamagical Themas , is also great and much more accessible.
7.
by iainmerrick 16d ago
Right. It was convenient for Microsoft, but essential for Apple. MS was the last partner they'd have chosen if they weren't desperate. Jobs was booed when he announced it! It had very much the air of Lando Calrissian announcing Cl
8.
by iainmerrick 16d ago
40 years ago, maybe. The Microsoft investment that saved them was in 1997. The writing had been on the wall for most of the 90s, at least as I recall.
9.
by iainmerrick 23d ago
Do you distinguish between Penrose's argument and the older one by Lucas, invoking Gödel's proof that certain facts are non-computable? I always felt it's basically the same argument, and Lucas's was never really convinc
10.
by iainmerrick 23d ago
When you put it like that, I think there are actually three separate things we're arguing about: - intelligence: ability to solve problems - agency: ability to be self-directed, to choose what to do - consciousness: ability to experien
11.
by iainmerrick 23d ago
Yes -- I can't remember the exact words, but there's a striking bit in GEB where he directly addresses the question of "will a machine ever create art?" His answer is yes, but only after it has really lived life, experie
12.
by iainmerrick 27d ago
Anthropic has Claude Code Web -- sorry, looks like the official name is "Claude Code on the web" -- which I think is basically this? I've been using it for a while and I really like it. Nothing to do with latency, I think, it
13.
by iainmerrick 28d ago
Yeah, "hand-crafted" is a bit of a stretch; I mean their value is in the human insight they contain. The key point I was trying to get at is that the human insights don't contain anything that can't be mined from vast am
14.
by iainmerrick 29d ago
I was replying to this: The conclusion of the bitter lesson would be that a large language model trained on chess commentary as well as being trained on millions of chess games would outperform stockfish which is only trained on millions
15.
by iainmerrick 1mo ago
I think you have it backwards. The common mistake is to think “maybe if we use a blend of raw data and hand-crafted heuristics, we’ll get the best of both worlds!” But the bitter lesson says no, beyond a certain point it’s better just to
16.
by iainmerrick 1mo ago
They mentioned Fenway Park, which is a baseball stadium. Baseball scoreboards are quite complex with a lot of detailed stats, similar to cricket scoreboards.
17.
by iainmerrick 1mo ago
I think you’re talking somewhat at cross purposes to the original article. The points I take away are: - Good optimization is difficult and slow work, hence expensive, but LLMs can do it so we should be able to afford it more often now. -
18.
by iainmerrick 1mo ago
Right, that’s why it costs $45,000 rather than just $25,000.
19.
by iainmerrick 1mo ago
Well, hang on, there are several different layout algorithms in CSS, right? They keep the old ones for backwards compatibility, and they have to interoperate which makes things more complicated; but flexbox and grid are relatively new and q
20.
by iainmerrick 1mo ago
LLMs aren’t just blindly parroting human speech, though -- they’re not trivial Markov chains any more, copying words and phrases at the surface level. You can argue back and forth about whether LLMs are actually “conscious”, whatever that m
21.
by iainmerrick 1mo ago
Hmm, what specific differences does it make in practice? For me, the huge one is that LLMs are currently bad at learning from experience. I don’t trust any kind of automatic MEMORY.md or whatnot; in fact I greatly prefer starting from a cle
22.
by iainmerrick 1mo ago
I had a similar thought -- rather than fixed benchmarks, you want dynamically-generated tests, specifically designed to exercise newly-exposed corner cases. So the way forward might be antagonistic benchmarks generated by another LLM.
23.
by iainmerrick 1mo ago
I don't see how "it's just token prediction" is relevant to that at all. Sure, the model can go wrong, but sometimes it's able to realise that and correct its course. Stronger models are better at doing this. People
24.
by iainmerrick 1mo ago
That's true, but you don't typically explain what you're testing in this sort of (presumably) randomised trial. And the Daring Fireball article does complain that watermarking will reduce quality. If that's what you&#x
25.
by iainmerrick 1mo ago
They do exactly that "which is the better answer?" test -- I've seen it pop up a few times.
26.
by iainmerrick 1mo ago
How exactly do you propose they should keep track of quality, then, if not by A/B testing?
27.
by iainmerrick 1mo ago
Yep, I use Gemini for this too and it’s great - very fast and high quality. I’d be very willing to try it out as an API, but it’s far too complicated to set up payment, and I don’t want to risk taking a wrong step and being locked out of ot
28.
by iainmerrick 2mo ago
It actually does touch on this: Built-in line and column tracking. Any movement across a newline updates the line number, including a backwards seek. getLine and getColumn are always available and both are one-based, which makes decent err
29.
by iainmerrick 2mo ago
I really like that categorization, but I think you've missed a step (or several): designing and implementing all the extra stuff needed to make it work, but aren't directly related to the problem. For example, you have an idea for
30.
by iainmerrick 2mo ago
Well, you can still have breakpoints, and switch between different structures for e.g. mobile vs desktop, or landscape vs portrait. But I don't think you can cover all requirements with a set of fixed layouts, they do need to be somewh

More ›