20 条评论

  • LukaD 1天前
    This is delightfully insane! I don't think I would say it doesn't play _entirely_ terrible though ;) It's playing really bad, but it could be worse and it's already super impressive that it can even generate legal moves.
  • Kaliboy 1天前
    This is amazing. I'm at loss for words.

    During my CS years I remember being fascinated by NFA's, as opposed to boring single universe DFA's.

    For some reason I internalized that I would never see something like an NFA implemented beyond text books.

    Then came Carlini.

    • bigdict 1天前
      But... they are equivalent?
      • Kaliboy 23小时前
        Yeah I know, but I thought I was doing purely theoretical excercises.

        And we always changed the regex NFA to an equivalent DFA and that was the implementation.

        So somehow I managed to internalize the idea that an NFA is purely theoretical and can't be built.

      • xpon 1天前
        Modulo an exponential blowup! That’s like saying P is equivalent to NP.
        • tgv 1天前
          Depends on what you mean by that. You can convert every NFA into a DFA. That's a NP complete (IIRC), but running the DFA is O(n). Running the NFA without converting it is also NP complete. One isn't better than the other, but the costs vary for different expressions and usages.
          • Running NFA is O(nm) not NP.
            • tgv 1天前
              Sorry, you're right. Capturing worst case was much more expensive, I believe, but I'm no longer sure.
            • So it is NP (in fact P)
        • froh 1天前
          The blow up is exponential for carefully crafted academical regular expressions.

          im practice is a good idea to build a DFA from your regex, up front (re2) or lazily (ripgrep)

        • pkal 1天前
          No, because you can compute the optimal automaton (as in least number of states) that recognizes the same language: https://en.wikipedia.org/wiki/DFA_minimization
          • IsTom 1天前
            And there are language families where minimal DFA is still exponentially large compared to NFA.
  • zelphirkalt 1天前
    It would be different, if somehow all those 84688 regexes were coded by hand. Then it would be a piece of art.

    It would be different, if the number of regexes was maybe below 300, and it still plays acceptably. The sheer number of regexes kind of defeats the purpose.

    At that code size, a much better engine can be written, or other kind of code for an engine be generated. Regexes themselves are not really something we should strive to use more either. Maybe its intentional badness kind of makes it art?

    • colonCapitalDee 23小时前
      What a depressing way to view the world. Sorry this great project wasn't to your taste, but there's no reason to be a dick about it.
    • matja 1天前
      I was also thinking along the same lines. Interesting, but I'm not sure in which aspect it is an achievement, considering the loop isn't a regex.

      Meanwhile, 1K ZX Chess takes fewer bytes of memory than the first four paragraphs from the post.

    • senfiaj 1天前
      > Maybe its intentional badness kind of makes it art?

      I guess it's the whole point of such type of blog posts. Similarly, some people write complicated interactive web pages without using JS, like this https://benjaminaster.com/css-minecraft/. But if you look at the HTML / CSS code size, it's usually huge, but still requires creativity to do that because of constraints. Obviously, it's not something practical or even optimal.

      • latexr 1天前
        > Similarly, some people write complicated interactive web pages without using JS (…) Obviously, it's not something practical or even optimal.

        There are people who navigate the web with JavaScript turned off, so those experiments do have practical applications.

        There are entire projects around not using JavaScript.

        https://theosoti.com/you-dont-need-js/

        https://github.com/you-dont-need/You-Dont-Need-JavaScript

        • senfiaj 1天前
          > There are people who navigate the web with JavaScript turned off, so those experiments do have practical applications.

          This is practical (and necessary) for relatively basic stuff, such as text content, navigation, basic form / input validation, and things like that. But when people write more complicated things (requiring state management, logical branches, etc), like games, 3d programs, etc, it's much more challenging (also can be sub-optimal) and requires more creativity. I mean they are more of a demo art rather than some strong necessity.

    • LatencyKills 1天前
      This is a quintessential, crazy idea that used to be adored on HN. The author, obviously, didn't intend this to be a serious engine.

      I wish more submissions began with, “This might be a bit wild, but I wanted to see if it could actually work.”

      • Timwi 23小时前
        > “This might be a bit wild, but I wanted to see if it could actually work.”

        That is how esoteric programming languages start.

      • Out of curiosity, why wouldn't it work?
        • LatencyKills 1天前
          Oh, I didn't mean that this specific project wouldn't work. I just wish HN were a little friendlier towards projects that are primarily thought experiments.

          Some of the best things I've ever created started from, "I wonder what would happen if I tried this crazy approach..."

          • solumunus 1天前
            I think it’s because of agent involvement. It takes away the coolness.
  • Rendello 21小时前
    This reminded me of Tom7's video where he made a bunch of ridiculous engines and pitted them against each other (and against "diluted" versions of Stockfish):

    https://www.youtube.com/watch?v=DpXy041BIlA

    https://tom7.org/chess/

  • evilsnoopi3 1天前
    The technical write up is worth perusing but I played a game before reading and accidentally found a winning strategy immediately. I'm not sure if this is a result of the 2-ply nature of the engine or if the mentioned deficiencies account for this but the computer did not act to prevent checkmate in 1 (without any intervening check); the game I played was (in algebraic notation): 1. e4 e5 2. kf3 kf6 3. kxe5 kxe4 4. d4 kxf2 5. Kxf2 a5 6. Qf3 b5?? 7. Qxf7 1-0
    • EvgeniyZh 1天前
      Yep, the scoring function is just piece value difference, so it can only detect checkmate in 0 (i.e., when king capture is available).
    • zelphirkalt 1天前
      Nitpick: In chess usually "N" is used to mean "knight", because "K" is already taken by "King".
    • FergusArgyll 1天前
      Hey! I had a very similar game
  • strenholme 1天前
    For people who are interested, here is the solution. In standard PGN, the solution is:

    1. e4 e5 2. Nf3 Nf6 3. Nxe5 Nxe4 4. Qe2 Nxd2 5. Nc6+ Ne4 6. Nxd8 Kxd8 7. Qxe4 a6 8. Bg5+ Be7 9. Qxe7#

    In the Stockfish notation this engine uses, White’s moves are:

    1. e2e4 2. g1f3 3. f3e5 4. d1e2 5. e5c6 6. c6d8 7. e2e4 8. c1g5 9. e4e7

    Here is a Lichess analysis of this game:

    https://lichess.org/WnMF3LpX

    (In terms of Regexes, Javascript has a very rich Turing complete Regex library; it’s an open question whether Lua 5.1’s regexes are Turing complete, but they are good enough for the text processing I do)

    • zzazzdsa 20小时前
      I won with 1. e4 e5 2. Qh5 a6 3.Bc4 a5 4. Qxf7#. I wonder if you could implement a stronger engine in regex (stockfish classic at O(1) nodes is plenty strong already)
    • wavemode 1天前
      I won faster than that:

      1.d4 d5 2.c4 dxc4 3.Nc3 Qxd4 4.Qxd4 a6 5.Bf4 a5 6.Bxc7 a4 7.Qd8#

  • This is like a fever dream.
  • deviation 1天前
    Not sure it's completely accurate. I played a standard queen's gambit accepted, took black's queen which it immediately blundered, then tried to move my queen from c5 -> e5 and the game ended immediately showing:

      *Illegal Move*
      You Lose.
      Game over.
    
    A little disappointed, since it's of course a valid move.
    • Someone 1天前
      “then tried to move my queen from c5 -> e5”

      Are you sure you typed “c5e5”?

      It’s very picky about how you specify a move. “e2e4” is fine as a first move, for example, but auto-capitalized “E2e4” is losing immediately. Quite weird, given that there are guardrails against “e2-e4” and “E2-E4” (an alert pops up telling you how to write moves)

    • Yeesh, one illegal move attempt means you just lose? That's harsh...
  • userbinator 1天前
    Upon reading the title, this is one of those "I know that's possible, but I'd never bother to implement it" things, although this particular implementation isn't exactly what I had in mind.
  • asplake 1天前
    And now you have 84,689 problems
  • explodes 1天前
    2025
  • dtj1123 1天前
    Brilliant. The Chinese room thought experiment as a chess engine.
  • devanshp 1天前
    This is absurd. I did not realize you could do nearly this much computation in regex.
    • tgv 1天前
      It's not just regex. The regular expressions are used to select and perform an action. There's a loop around it with controls the stack. That has more power than the regex.
    • karlgkk 1天前
      It’s turing complete so you could compile almost any language to regex. You might have to build a vm for some languages, also in regex. The point is, it’s regex all the way down.
      • Patryk27 1天前
        Regular expressions are not Turing-complete.
        • 0xffany 1天前
          True in the CS Theory space, but most modern regex engines implement a few niceties which make their "regex" turing complete. https://blog.poisson.chat/posts/2024-06-18-turing-regex.html
        • Javascript/PCRE/etc regexes have additional features (like backreferences) that give them strictly more computational power than a regular DFA/NFA. (Still not Turing complete though without external control flow to support arbitrary iteration/recursion, like is done here)
  • carlsborg 1天前
    "Memory plus search is all you need"
  • casey2 1天前
    Alternate title:

    Compiling Python to a Branch-Free SIMD Virtual Machine via Extended Regular Expression String Rewriting

  • mashijian 1天前
    [flagged]
  • kudesnikz 1天前
    [dead]
  • [flagged]