Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The Kilobyte’s Gambit

vole.wtf

Post reply on HN
Print view View original

121–130 of 158 posts

Re: The Kilobyte’s Gambit

#121

Post by dandanua »

angry_octet wrote: post #116
GCA10 wrote: post #32

Nice work! I'm about 1850 in Lichess blitz, and it played at the level of an earnest 12 year old, which is actually meant as a compliment. It got crazy with its kingside pawns in the opening (no book knowledge), lost material pretty steadily and was checkmated down a queen, bishop and two pawns by move 25. But it developed some pieces sensibly, made obvious recaptures and was not reckless about king safety. It avoide…

What you're saying is I'm getting crushed by an earnest 12 year old. Oh dear.

There are 12 years old grandmasters in the history of chess.

The truth is you've been crushed by 1kb javascript code :)

Re: The Kilobyte’s Gambit

#122

Post by timonoko »

Incomprehensible scheiße code. I looked around and I like this one because it has "meta-level" definition of movements and liitle bit of strategy. You could implement context-free chess games with varying rules for us congenitally lazy and dull-witted. https://github.com/thomasahle/sunfish/blob/master/sunfish.py

Re: The Kilobyte’s Gambit

#124

Post by YeGoblynQueenne »

register wrote: post #110
chki wrote: post #35

Earlier quoted context omitted.

I agree. I'm at a similar chess rating level, I would add that it is very weak in the endgame. I guess if it only looks four moves into the future it's difficult to see pawn promotions.

It looks one-two semi-moves ahead at maximum.

"ply" :)

The "How it works" link says it looks ahead 4 moves, that's probably 2 ply.

Re: The Kilobyte’s Gambit

#126

Post by rob74 »

cout wrote: post #93
andrelaszlo wrote: post #8

I'm a novice player but I reached a stalemate in my first game. I am not a big fan of the graphics, they're really really cool but I made at least two big mistakes because I misread the pieces - and now I have a headache :D

I thought the graphics were brilliant. Reminded me of playing Battle Chess on 4-color CGA.

Ok, maybe if you have fond memories of playing CGA games... for everyone else, those colors (especially the cyan/magenta palette) are just a garish eyesore. But I guess the engine needs every advantage it can get ;)

Re: The Kilobyte’s Gambit

#127

Post by Someone »

schoen wrote: post #99

Sharing other people's amazement that such a tiny amount of code plays so well (I'm an inexperienced, infrequent player and it beat me once and drew once), I was pondering that there's no absolute reason to think that small programs inherently must play games poorly. For many abstract strategy games, you could write an explicit game tree search in a relatively small amount of code and it would, in principle, play per…

“but would for small games like […] Nine Men's Morris”

http://library.msri.org/books/Book29/files/gasser.pdf:

“We describe the combination of two search methods used to solve Nine Men’s Morris. An improved retrograde analysis algorithm computed endgame databases comprising about 10¹⁰ states. An 18-ply alpha-beta search then used these databases to prove that the value of the initial position is a draw”

So they built a database containing all all 7,673,759,269 possible positions in the endgame phase (i.e. after all stones are on the board), and then did a full-depth search for the “place stones” phase.

Paper is from 1996, so one might think current hardware could do a full search, but they show (page 10/110) a position with “White to move and win. Mill closure in 187 plies.”

So, I don’t think this game is small enough to search the full tree. They made it small by first creating an endgame database with 7,673,759,269 endgames, but that database requires over a gigabyte of data (“The hash function we decided to use maps the 7,673,759,269 states into a range of 9,074,932,579 indices.”)

Re: The Kilobyte’s Gambit

#128

Post by galkk »

lucb1e wrote: post #90
strken wrote: post #39

Earlier quoted context omitted.

For those of us who have no idea how to play chess, what level does an earnest 12 year old play at?

I joined my girlfriend to a chess tournament of the local club once. Old vs. young. They were an adult short, and I know the rules, so I volunteered to play - otherwise there would be a kid without an opponent every round (my plan had been to write some code in a corner, but entertaining the kids, why not right?). I lost every game. The kids were something like 10 years old. Admittedly, I know little more than the ru…

Sergei Karjakin became grossmeister at age 12

Re: The Kilobyte’s Gambit

#129

Post by h0l0cube »

bonzini wrote: post #113
h0l0cube wrote: post #36

Earlier quoted context omitted.

Yeah, the pieces are a bit illegible. I accidentally moved my Queen instead of castling, which is when I quit. Otherwise pretty impressive

The pieces were meant to be displayed on a composite (NTSC) display rather than with 4-color CGA graphics. Check out the Twitter thread at https://mobile.twitter.com/pinot/status/1330544777847332879 .

Don't get me wrong, I dig the retro art style, but at that resolution, dithering gets in the way of legibility. I think for the chess pieces themselves it makes more sense to trade off some of the realism (i.e., shading, specular) to better communicate the shapes, particularly as it's so important to game play

Re: The Kilobyte’s Gambit

#130

Post by ejolto »

The_rationalist wrote: post #45

If I remember correctly the size of DNA is ~100MB

This statement doesn't make much sense to me. Is there a theoretical limit to the length of DNA, or do you mean that human DNA is ~100MB?

The human genome is about 6.4 billion base pairs, where each base pair is represented by the letters: Adenine (A), Thymine (T), Cytosine (C), and Guanine (G), i.e. 2-bits. 6,4 billion is then 6.4E9*2 = 1.28E10 bits = 1.28E10/8 bytes = 1600 MB. Or 800 MB if you count the bases as pairs of two.

Polychaos dubium's DNA is over 200 times as long as human DNA[0].

[0]: https://en.wikipedia.org/wiki/Polychaos_dubium

Post reply on HN