Something that bothers me here is that Anthropic claimed in their blog post that the Linux kernel could boot on x86 - is this not actually true then? They just made that part up?
It seemed pretty unambiguous to me from the blog post that they were saying the kernel could boot on all three arch's, but clearly that's not true unless they did some serious hand-waving with kernel config options. Looking closer in the repo they only show a claimed Linux boot for RISC-V, so...
It's phishy because it's breaks the rules people are generally told for avoiding phishing links, mainly that they should pay attention to the domain rather than subdomains. Browser even highlight that part specifically so that you pay attention to it, because you can't fake the real domain. The problem with what GitHub does here is that while `github.github.io` might be the real GitHub, `foobar-github.github.io` is not because anybody can get a github.io via their username, that was part of why they made github.io separate. Additionally they could easily host this via GitHub Pages but still use a custom domain back to github.com, they just don't.
I would say that GitHub is particularly bad about this as they also use `github.blog` for announcements. I'm not sure if they have any others, but then that's the problem, you can't expect people to magically know which of your different domains are and aren't real if you use more than one. They even announced the github.com SSH key change on github.blog.
Without questioning the LOC metric itself, I'll propose a different problem: LOC for human and AI projects are not necessarily comparable for judging their complexity.
For a human, writing 100k LOC to do something that might only really need 15k would be a bit surprising and unexpected - a human would probably reconsider what they were doing well before they typed 100k LOC. Where-as, an AI doesn't necessarily have that concern - it can just keep generating code and doesn't care how long it will take so it doesn't have the same practical pressure to produce concise code.
The result is that while for large enough human-written programs there's probably an average "density" they reach in relation of LOC vs. complexity of the original problem, AI-generated programs probably average out at an entirely different "density" number.
Yeah I agree that one isn't very clear, perhaps the idea is to use `msync()` as a barrier to achieve consistent ordering of the writes without having to handle that yourself with more complex primitives. But then, they do mention some of those primitives at the bottom of the article, so it's hard to say what exactly the idea is.
Just because you believe X is going to happen doesn't mean you can make money in the market off of that information, that requires judging what _everyone else_ thinks will happen and thus how the market is priced. You could just as easily get stuck in the situation where the market as a whole was expecting it to be worse than it was and didn't move far enough for you to make your money back.
I would add, I've always felt the addition of the 3D screen also held it back because the touch screen became relegated to always being secondary (so that the main display of the game could be 3D). Many games that were sequels to touch-focused DS games did this, they had you touching the bottom screen to interact with something on the top screen and made those games feel a lot worse to me.
It depends what rules you're using, but there are the three-fold repetition and 50-move rules which allow a player to force the game to end in a draw. The catch is they both require one of the players to claim a draw under the rule, otherwise they can keep playing.
There is additionally the 75-move rule where the the game is forced to be over without either player claiming the rule (the arbiter just ends the game), that rule would give an upper bound regardless of the players knowledge of the rules.
As I understand it, the 50-move rule must be invoked by one of the players, lets assume our immortal players agree not to invoke that rule.
The 75-move rule is automatic, so that would be the limiting factor.
Note, that 75-move rule is only applicable after no pawn has moved or a piece has been captured. So our immortals can do a lot of shuffling things around.
I'm thinking that the number of moves of the longest game is going to be (16 pawns * 7 moves each + 16 pawns being captured + 14 other pieces each being captured, not the kings) * 75 moves for shuffling around = 10650 moves.
That's only 1 week at 1 move per minute! But given the permutations, it might take much longer to calculate the actual moves required to get to the end state :)
Pawns only get 6 moves :) But also they can't all make 6 moves because they can only move past each-other via capture, so half of them would get 5 moves instead (if you're counting all the captures), so that gives a maximum of ~8850.
How I'd put it is that there are two sets of stopping points under FIDE rules:
- After threefold repetition or 50 moves, either player may claim a draw.
- After fivefold repetition or 75 moves, the game is automatically drawn.
Most modern counts of the longest possible chess game, or the total number of possible chess games, are based on fivefold repetition and the 75-move rule.
Meanwhile, threefold repetition and the 50-move rule are still relevant in endgame tablebases, since they rule out certain forced mate sequences.
Endgame tablebases don't take into account threefold repetition; if so, you would have to basically be able to exclude any arbitrary position from the tree, which would seem impossible. The 50-move rule is respected by the Syzygy tablebases, though with the concession that they do not generally give the fewest possible moves to mate (they would rather delay the mate than delaying a pawn push or a capture).
Here's an example (adapted from the URL below): https://syzygy-tables.info/?fen=3R4/5R2/8/8/8/1K6/8/4k3_w_-_... — if you asked pretty much any player, even a child, how to win this, they'd show the staircase mate starting with Re7+ (mate in 4). If you asked a computer or the older Nalimov tablebases, it would say Kc2! (mate in 2). However, if you ask the Syzygy tablebases, they would argue that this is not optimal if we are extremely close to the 50-move rule, so the safest and thus best move is Rf2!! which forces Black to capture the rook on the next turn (they have no other legal moves), resetting the counter and giving a mate in 18.
There were a set of experimental DTM50 tablebases made at some point (though not made public); they store the shortest mate for all 100 possible zeroing counters in any position. See https://galen.xyz/egtb50/ for some discussion.
Comparing Chess AI to AlphaStar seems pretty messy, StarCraft is such a different type of game. With Chess it doesn't matter if you get an AI like Lc0 to follow lines it played previously because just knowing what it's going to play next doesn't really help you much at all, the hard part is still finding a win that it didn't find itself.
In comparison with StarCraft there's a rock-paper-scissors aspect with the units that makes it an inherent advantage to know what your opponent is doing or going to do. The same thing happens with human players, they hide their accounts to prevent others from discovering their prepared strategies.
> Presumably, there's a step in this process where money incentivizes the opposite of my suggestion, and I'm not familiar with the process to know which.
> Is it the university itself which will be starved of resources if it's not pumping out novel (yet unreproducible) research?
Researchers apply for grants to fund their research, the university is generally not paying for it and instead they receive a cut of the grant money if it is awarded (IE. The grant covers the costs to the university for providing the facilities to do the research). If a researcher could get funding to reproduce a result then they could absolutely do it, but that's not what funds are usually being handed out for.
I can't speak for the author, but I do often do this. IMO it's a misleading comparison though, you don't have to debug those things because rarely does the compiler output incorrect code compared to the code you provided, it's not so simple for an LLM.
It seemed pretty unambiguous to me from the blog post that they were saying the kernel could boot on all three arch's, but clearly that's not true unless they did some serious hand-waving with kernel config options. Looking closer in the repo they only show a claimed Linux boot for RISC-V, so...
[0]: https://www.anthropic.com/engineering/building-c-compiler - "build a bootable Linux 6.9 on x86, ARM, and RISC-V."
[1]: https://github.com/anthropics/claudes-c-compiler/blob/main/B... - only shows a test of RISC-V
reply