Los Alamos Chess

[keid]

I knew one of the authors, Mark Wells. After he retired from Los Alamos National Laboratory he moved, temporarily, to Las Cruces, NM, to be head of the CS department at New Mexico State University, so became my boss as a student employee in the department. I don't recall his authorship of the chess program being mentioned by any of the profs or grad students (it was a very small department). When I moved to Los Alamos many years later I discovered that he lived just down the street--they'd kept their house in Los Alamos while in Las Cruces and had returned (for the skiing, he said).

The first article linked below has some details about the chess program not in the Wikipedia article.

Chess was his (serious) hobby; his research was in programming language design, specifically concerning type theory.

https://ladailypost.com/former-lanl-mathematician-computer-s...

https://ladailypost.com/obituary-mark-brimhall-wells-oct-7-2...

[chemeril]

I'd keep a house for the skiing at Pajarito too! For those reading along it's certainly worth the trip if you're in the area.

I have a vague recollection of coming across a physical 6x6 chessboard somewhere on lab property and found it a little odd, but never knew it had ties to MANIAC. Lots of history floating around in that place.

[QuantumSeed]

I was at NMSU during his time as head of the CS department. His interest in chess inspired to write my first chess playing program

[jhbadger]

I'm not sure if anyone actually plays this as an actual chess variant, but it is fun to write a chess player for this, mirroring the 1956 version. I remember reading about this in the 1980s and wrote a version in BASIC that could beat me every time (maybe not that impressive objectively, but it impressed me).

[huytersd]

I hate sudoku variant videos that rope me in with some impossible scenario and then explain a totally different set of rules.

[passwordoops]

For some reason your comment made me think of the classic Simpsons bit, "When I stamp your foot and say 'Hello Mister Thompson' you say hi back." Funny how the human mind works

[dvh]

Kruskal (author of the kruskal algorithm) won.

edit: pardon, wrong Kruskal

Joseph Kruskal had two notable brothers, Martin David Kruskal, co-inventor of solitons, and William Kruskal, who developed the Kruskal–Wallis one-way analysis of variance. One of Joseph Kruskal's nephews is notable computer scientist and professor Clyde Kruskal.

[pavel_lishin]

And apparently his mother popularized origami in America!

https://en.wikipedia.org/wiki/Lillian_Oppenheimer

[thriftwy]

This one seems to be the author of https://en.m.wikipedia.org/wiki/Kruskal–Szekeres_coordinates

Which are great at challenging commonly held beliefs about black holes.

[frankfrank13]

Found an example to play online with a bot! https://www.chessvariants.org/play/jocly/los-alamos-chess

[procgen]

I'd love to play against the original algorithm.

[Stevvo]

Doesn't follow the rules; enemy pawn promoted to queen upon reaching other side of board.

[davidw]

> The computer still needed about 20 minutes between moves.

Wow. Coding and debugging that is... a whole other level of difficulty compared to coding these days.

[pacaro]

My mother describes laying out the program flow on index cards (or recycled punch cards depending on what was available) on the living room floor and walking through the program/room with a notebook tracking the program state.

To her that was debugging

This was mid to late 60s

[tombert]

Every time I read about about how slow really old computers were, I say an internal "thank you" to Bell Labs for the transistor. I'll occasionally see a YouTube video about vacuum tube or relay based computers, and they'll show their operations on the order of 3-10 operations per second. And here I am writing this comment with something whose operations are on the order of billions per second.

[moffkalast]

> This program was written at Los Alamos Scientific Laboratory by Paul Stein and Mark Wells for the MANIAC I computer

I see the names of computers have gone significantly downhill since the 60s.

[kijin]

> Metropolis chose the name MANIAC in the hope of stopping the rash of silly acronyms for machine names [1]

Looks like even MANIAC was deliberately restrained to make it sound less maniac than the names of computers that came before it. Can't beat Manchester Baby, Colossus, The WITCH, SWAC and MADDIDA.

Being Los Alamos, they really should have named it ATOMIAC or something.

[1] https://en.wikipedia.org/wiki/MANIAC_I

[nxobject]

I miss adding "AC" to the end of computer names to emphasize that you indeed have an "automatic computer". ("MACbook Pro" doesn't count, I swear!)

[tibbetts]

I dunno, the new Microsoft / OpenAI computer is named “Stargate,” that’s a pretty solid name.

[CrociDB]

> The computer played three games. The first was played against itself. The second one was against a strong human player, who played without a queen. The human player won. In the third game, MANIAC I played against a laboratory assistant who had been taught the rules of chess in the preceding week specifically for the game. The computer won, marking the first time that a computer had beaten a human player in a chess-like game

I think this paragraph is kinda funny.

[tibbetts]

I was disappointed that the lab assistant is not named.

[ryukoposting]

This looks like fun. I like chess variants that use smaller boards like this. They offer much of the same high-level strategic thinking, but a much shorter play time and a lower barrier to entry.

My favorite one is Onitama, it's actually quite different from chess but the same intuitions apply.

[jamager]

Onitama is brilliant. For me the best along with Hive and Santorini.

[QuesnayJr]

If anyone wants to try playing this, it's one of the variants supported by Fairy Stockfish.

[JoeDaDude]

Not to be nitpicky (well, maybe a little)...

El Ajedrecista [1], a chess machine designed by Spanish engineer Leonardo Torres Quevedo in in 1912 and later improved in 1915, defeated chess Grandmaster Savielly Tartakower in 1951 [2], a few years earlier than the unnamed lab assistant who lost to Los Alamos Chess. This would make Savielly the first human player defeated in a chess-like game. The lab assistant loss must have occurred sometime after 1956.

To be fair, El Ajedrecista played a very limited game, only capable of playing a white king and rook against a black king moved by a human opponent. Still, I think it qualifies as a "chess like game"

As a matter of casual trivia, El Ajedrecista inspired Claude Shannon, the founder of information theory, to build his own chess machine [3], which he named Caissac (after the goddess of chess Caissa [4]) in 1950.

[1]. https://www.chessprogramming.org/El_Ajedrecista

[2]. https://en.wikipedia.org/wiki/El_Ajedrecista

[3]. https://www.chess.com/article/view/the-man-who-built-the-che...

[4]. https://en.wikipedia.org/wiki/Ca%C3%AFssa

[keid]

Just to add some details, the Los Alamos Chess people acknowledged that theirs was not the first such computer program, that Alan Turing had previously done created one. Presumably Turing's program didn't ever beat a human, though.

https://discover.lanl.gov/publications/at-the-bradbury/2023-...

[redbell]

> The reduction of the board size and the number of pieces from standard chess was due to the very limited capacity of computers at the time.

I wonder what the authors might have achieved having access to modern computers that are billions times faster than their machines.

> The computer played three games. The first was played against itself. The second one was against a strong human player who played without a queen. The human player won.

Again, I wonder what they would think seeing AlphaGo beating Lee Sedol [1],[2]

1. https://en.m.wikipedia.org/wiki/AlphaGo_versus_Lee_Sedol

2. https://youtu.be/WXuK6gekU1Y

[polynomial]

Why did they lose the bishops rather than knights? Seems like if you were simplifying the game for early computers to learn, that's the move (remove the one piece that moves weird.)

[kqr]

Bishops have a worst branching factor of 9, while knights have only 8. And this difference is even larger on average; a bishop at b2 can move to 7 places, while the knight on the same location can move to just 4.

(This can be thought of dynamically as well. If a bishop moves one space closer to a corner, it loses three potential moves, but gains one move in the opposite corner. If a knight moves closer to a corner, it just loses moves, it never gains them.)

[OscarCunningham]

Has this been solved?

[bongodongobob]

No, we only have endgames solved up to 7 pieces. 24 pieces is waaaay too big to solve.

[once_inc]

How did I go through 4 decades of life without knowing about the en-passant capture? I've never heard of it before...

[vasco]

It's a common and fun way to make a noob angry that you're breaking the rules before you google it in front of them. I've had tough times "proving" it before Google though!

[QuesnayJr]

While a part of the standard chess rules for over a century, it's common for people who learn to play with their family as kids don't learn it. I didn't learn it, either, and only found out about it when I started playing chess programs online.

This is common enough that "Google en passant" is a meme in some circles.

[ChilledTonic]

I'm actually very interested in picking this up as a varietal to regularly play; the smaller chessboard size and reduced pieces would (theoretically) result in faster, pickup play.

[prometheus76]

5 X 5 is also really fun. It's the queen side basically, with a row of pawns in front. So there's only one row of open spaces between the two sides. It really stretches my brain to play it, but has helped me get better at calculation for sure.

[listenlight]

Has anyone here played losers chess? Some of the top players on ICC (chessclub.com) back in the day played losers…

[michaelcampbell]

Looks interesting as a total looker-on of chess. Does the smaller board and reduced rules make the game any easier?

[ryukoposting]

It makes the game go a lot faster, that's for sure. It lowers the barrier to entry by a lot, while still requiring the same modes of thinking.

There are a ton of chess-like games out there that have small boards and play times of 10-20 minutes. IMO as an everyday thing to do with friends, they're a lot more fun than chess.

[umanwizard]

Yes, that reduces the search space

[abound]

Not to be confused with Atomic Chess: https://en.m.wikipedia.org/wiki/Atomic_chess

[Y_Y]

1. Feynman to U-235

[AdmiralAsshat]

So judging from the omission of the bishop and the queen, the computer really hated calculating diagonals, huh? I would've assumed that calculating the knight's movements would've been the bigger headache.

[InitialLastName]

The queen is there. It was probably a more interesting game to keep the knights for their moveset rather than keep the bishops (given the need to reduce to 6x6).

Between the ~half-sized board, the intrinsic limitations in where they can go, and the reduced pawn movement, bishops would be very clumsy if left in.

[Sharlin]

I don’t think there’s anything difficult in calculating a knight’s moves. On the other hand rooks, bishops, and queens generate many moves at a time, making the game tree much wider, which might make a difference, or it might not. But anyway, as noted by the sibling, queens were there (except in the second game, in which Kruskal played without and the computer played with, to even the odds), and the omission of bishops in particular had nothing to do with evaluation difficulty per se.

[silisili]

They retained the queen, it was just the human played without one against the computer for some reason. Probably to give MANIAC a fighting chance.