iainmerrick

Oh, I guess you have to ensure the inputs aren’t correlated, or they’ll cancel out?

And "the west" isn't really homogenous for these purposes. Most western countries don't have a school shootings problem, only the USA does. The parts of the west most heavily criticising Israel are not the USA, it's places like Spain and Ireland (both of which have interesting historical parallels!) I think the USA has a higher birth rate than the EU, so maybe you'd prefer them on that basis, but then you don't want school shootings, so maybe you prefer the EU. You can probably find a reason to dislike everyone if that's what you want to do.

In the most charitable interpretation, the filmmakers chose to make this film as Western liberals trying to influence the Israeli government to adopt more Western liberal values

That's hardly the most charitable interpretation! Maybe they're trying to influence Israeli citizens, or maybe they're trying to influence Zionists abroad? There's no need to assume some insidious outsider conspiracy.

Surely people are allowed to disagree non-violently with their current government! If that's the kind of "foreign ideology" that you're dismissing out of hand, well, that certainly clarifies things.

I don't think many people would buy your "more logical for them to just go on the record" claim -- in this situation, people are scared! And they have conflicted loyalties as well. They want to blow the whistle, but don't necessarily want to directly accuse their immediate colleagues to their faces.

Another viewpoint is that it's partly true but misleading -- the filmmakers and whistleblowers have blown things entirely out of proportion. That sounds entirely possible. But wouldn't the "logical" response then be more transparency, rather than accusations of treason and a witch-hunt for the collaborators?

And another viewpoint is that it's all completely fabricated, malicious lies. Possibly an extreme crackdown might be an appropriate response in that case? That seems very counter-productive to me, as an outsider, but again cold logic doesn't always apply.

Which of those viewpoints do you find yourself agreeing with?

To me, the broadly true one seems most likely from the available evidence; the hiding of evidence and the attempted crackdown support that one too. It also seems like some people genuinely believe (or pretend to themselves) that it's all malicious lies. To me that smacks of sheer paranoia. Understandable after the October 7 attack, sure, but not rational or productive.

Hey, let's present our story as a series of soundbites over emotive imagery and ominous music, that definitely evokes serious journalism.

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 only one implying those things!

It seems like an unlikely long-shot that Microsoft saved Apple.

Why unlikely? They got into financial trouble due to bad decisions. They needed a bail-out from a former arch-rival to dodge bankruptcy. They eventually achieved huge success due to better decisions. There's nothing contradictory there.

Jobs was booed when he announced it! It had very much the air of Lando Calrissian announcing Cloud City's strategic partnership with Darth Vader.

I always felt it's basically the same argument, and Lucas's was never really convincing; Hofstadter rebutted it very convincingly in GEB long before Penrose's books came out.

To me, camp C's claim is possible but I don't buy the argument that says it's necessary. So I guess I'm camp A. (D is religion and B seems to me incoherent.)

- intelligence: ability to solve problems

- agency: ability to be self-directed, to choose what to do

- consciousness: ability to experience things; "having a self"

The "intelligence" part is largely solved now (that's a very big statement, but I can't think of a better way to phrase it!) Although really it's just moving the goalposts -- computational stuff like calculating trajectories or playing chess was solved long ago, this is just more and more things moving into the "solved" column.

That leaves agency and consciousness as the hot topics that nobody has really figured out. Do they always go together, does one require the other, does one create the other? Who knows?

"Free will", there's another one. That must surely connect with agency, with consciousness, or both, in some manner we haven't figured out.

To me, agency seems like a solvable problem via existing approaches and technology. I have no idea what that means for consciousness. It's tempting to conclude that consciousness is just a mirage, but then we all do feel like we have it, so it seems like it must be something. Maybe consciousness arises automatically once you have sufficient intelligence and agency; but how would you ever determine that?

Other people might think that true agency requires consciousness, and that consciousness requires some magical new ingredient that LLMs don't have yet. The problem with that approach is, you have to identify what it is that true agency can do that software can't do; and every time you do that, it turns out LLMs can do it, so you have to keep moving the goalposts. Trying to make your argument rigorous immediately makes it self-defeating; I think that's why so much of the philosophical discussion around this is impenetrably vague. All the arguments that aren't vague just turn out to be wrong.

Oh, I guess there's a question of whether you can separate intelligence from consciousness. His writing was mostly done in the long period where things that are very easy for humans were still intractably difficult for machines -- identifying objects in pictures, following simple written instructions, etc. So I think it's clear that he saw "intelligence" as being able to solve those simple-yet-intractable common-sense problems; the kind of thing LLMs now excel at.

Nothing to do with latency, I think, it's all about convenience and safety. I don't have to worry about running an agent on the computer that knows all my admin API tokens and SSH keys, and the agent isn't interrupted if I close my laptop.

The one thing that might be useful is the ability to run on a beefier machine when you need it, e.g. make sure it has direct access to a GPU. But then you're back to thinking about the host VM, not fully abstracting away from it.

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 amounts of gameplay. Every human insight can eventually be rediscovered and made rigorous by data (in chess, at least!) In the short term, those insights are useful, but in the longer term, they add nothing at all.

Note also that "raw gameplay" here can mean new games -- you can generate as much data as you need, you don't need to rely on real recorded games.

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 of chess games.

If you can draw any lessons from chess commentary, I think it’s very reasonable to call it “hand-crafted heuristics.”

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 use the data.

Thinking that an LLM might be able to improve on purely “big data” machine learning seems to me to be the same incorrect idea. Its “intelligence” is no more useful than human intelligence. The LLM is based on a massive data corpus, sure, but the amount of data specifically about chess in there pales in comparison to just playing billions of games of chess.

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.

- There’s always a risk of over-fitting to your specific problem, but if everyone is now making bespoke optimizations maybe that isn’t actually a problem.

You said:

LLMs are terrible at optimizing memory utilization. There is just too little training code that does it well and far too much that doesn't.

There’s probably something in that, but it can be mitigated by testing against a local benchmark. LLMs are good at iterating tirelessly and finding incremental improvements. And as noted above, it doesn’t necessarily matter if your benchmark isn’t fully general.

You can argue back and forth about whether LLMs are actually “conscious”, whatever that means, but it’s clear that they can be tremendously effective and useful. They can use language to get stuff done.

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 clean slate each time because the LLM’s baseline general knowledge is so good.

In terms of accuracy and “lying”, I don’t really see a huge difference. Most LLMs are unfortunately a bit sycophantic and over-confident, but you sometimes see that in people as well.

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 do exactly the same thing! Haven't you ever wasted a lot of time chasing down a blind alley?

To say the LLM has immutable limits because it only predicts the next token and can't backtrack is like saying we have immutable limits because we can't travel backwards in time. It's a true statement but not particularly relevant or helpful.

And the Daring Fireball article does complain that watermarking will reduce quality. If that's what you're trying to check, "which is better?" is the right question.

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 other Google services. So Anthropic and Mistral get my money instead.

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 error messages nearly free.

That doesn't sound like much, but having hand-written plenty of recursive descent parsers, it's most of what you need for good error messages. Just being able to pinpoint where the error occurred is usually 80% of the battle; but keeping track of lines and columns in a hand-written parser is a pain.

Sure, for something like Rust, you need vastly more than that, but parsing is a tiny fraction of what the Rust compiler is doing -- type-checking and borrow checking is much more complicated and much more important.

A tiny library like this is a great fit for something like an INI file parser.