133-line chess engine with annotated source | Hacker News Reader

133-line chess engine with annotated source

home.hccnet.nl

91 pointsstarpilot19 comments

Open articleView on HN

ZX81 - chess ...only 672 bytes. source code:

http://users.ox.ac.uk/~uzdm0006/scans/1kchess/

Micro-Chess, from 1976, which ran on a KIM-1 (6502) machine and used 1.1K of RAM.

http://users.telenet.be/kim1-6502/microchess/microchess.html

Toledo Javascript Chess Game

http://nanochess.110mb.com/chess4.html

+ TOLEDO JS1K orking demo: http://js1k.com/2010-first/demo/699

The article states the goal to be under 2000 characters, to state it in terms of lines of code (as the submitter has) is kinda misleading when the following is considered to be one line:

  if(*c-10){K=c[0]-16*c[1]+C;L=c[2]-16*c[3]+C;}else  /* parse entered move */

edit: can't see a way to escape asterisk characters, so this outputs slightly wrong (edit2: fixed)

very true. It would be way more impressive to be in 1000 expressions.

I don't think there is (without what follows)... I've tried a couple experiments, but without adding spaces I don't think it works. (though if you have one * on its own, followed immediately by a space, it keeps the * there).

The correct way, though: add two spaces to the front. Then you get mono-spaced font, reproduced verbatim.

  Kinda like this: **************

Ah thanks, fixed.

use to find that way of thinking mostly in perl hackers.

A pity that people choose to sacrifice a lot of readability just to save a few characters in the source... it's not even justifiable in a interpreted language... in a language that you have to compile it's criminal

The source is not intended to be readable. The glory is in the 1433 characters total (or 133 lines, if you prefer that metric). Everyone knows that it is artificial, but it does not matter. The spirit is similar to the 10K Javascript contest (see http://10k.aneventapart.com/ )

If you want to try to understand the chess code, the author provides an expanded version with comments that is more readable: http://home.hccnet.nl/h.g.muller/maximax.txt

In order to read the code, it is helpful to understand the basic algorithm (recursive negamax search). Here is a good explanation of minimax, negamax and alphabeta: http://www.hamedahmadi.com/gametree/

According to the developer, Micro-Max has an Elo rating of about 2,000 (http://code.msdn.microsoft.com/huochess). A 2,000 Elo rating corresponds to "expert" per the USCF, or about the top 1% of players (http://en.wikipedia.org/wiki/Chess_expert).

I've played at least a hundred games against it and lost every one. :P

Without attempting to read the obscured code, the reason for the middling high Elo lies in the algorithms implemented. One of the AI discoveries is the sad fact that chess does not need intelligence, just horse power and a good per move check list. The algorithms listed provide both...assuming a decent implementation--- I can't tell because I'm not about to translate the code :)

Would you tell me how to enter my moves so that I can play it? I've been trying to figure out the correct format but can't seem to get it.

I played it in Windows via Winboard (http://www.open-aurec.com/wbforum/viewtopic.php?f=19&t=5...), which has Micro-Max set as the default engine.

just enter source and dest, e.g.: d2d4 to move the pawn at d2 to d4.

This is what I assumed all code looked like before I started programming.

This is what all code looked like when I started programming.

Code golf for chess.

Sir, you are a hacker.

The article says the source code measures 1433 characters !

I am reminded of 1K ZX Chess:

http://en.wikipedia.org/wiki/1K_ZX_Chess

i'm finding the documentation extremely interesting... but wonder how much it would be improved with properly named variables :)

http://home.hccnet.nl/h.g.muller/maximax.txt

Reply on news.ycombinator.com