This is what happens when you play random moves

I wanted to see what a "random" chess game would look like. Basically i give each possible move an equal probabilty, and see the game unfold. This is what came up. I didn't finish it, it was taking waay too long.

lol

According to my database, the position after 22...Kg8 has occurred nineteen times in mondern grandmaster praxis, most recently Karpov-Jussupow, Linares, 1985.

Play continued with the brilliant 23.c5! (preventing the devastating 23...Qxa3+!), and a draw was agreed.

There is an engine called "Brutus random" which plays purely reandom moves. It uses WinBoard protocol, so it should be able to play in most user interfaces for Chess (Shredder, WinBoard, Arena). The negine Pos 1.19 also has a random mode. (When started through the command "pos.exe --io-mode xboard --brain-mode random".) Having such engines play against each other is an easy way to generate random games.

I wrote a chess engine.  This is how it plays.

Interesting. I wrote such a chess program once (that's the first step in checking that it works, before you put the smarts into it) and that's how it played. I might note that there are (at least) a couple different ways to program such random play: (1) generate all possible moves and give each move equal probability; (2) choose randomly from the 16 units first, then have that unit move randomly. The results will be different.

This random play could actually be useful for analysis, since you could measure the average game length as you add different types of heuristics or lookahead depth, so in that way you might conceivably be able to rank the relative importance and effect of heuristics, or of lookahead depth.

Recently I wrote a Chess engine that does not think ahead, but just uses static criteria (like number of attackers and defenders of a square, and the value of the lowest defender) to decide upon the move. ( http://www.talkchess.com/forum/viewtopic.php?t=54757 ) This to investigate how efficiently it would beat a random mover.

Turns out it had some difficulty to win games. Without a preference for checks it tends to stalemate the opponent once it has reduced it to a bare King. (To detect stalemate you have to think 2 ply ahead, because you have to test if the opponent will have moves after yours.) And with a preference for checks, it tended to chase the King over the board with a Queen, until the 50-move rule caused a draw.

I could in the end solving it by only giving a bonus for checking with a piece that had not been checking on the move before. And making it refrain from capturing the opponents last minor reduced the number of stalemates to almost zero.

This is actually awesome guys! I've had my first half year course of programming on the university (of electrical engineering though) and we only made a board with pieces moving like in real life (controled by human - you could pick any piece you wanted and put it anywhere). Check it there: http://pastebin.com/hAs07BMa

However, ^this sound like an advanced programming so I should rather be quiet here. :)

Mmm? You're from today, too. Hmm.

Yep.

It would be really evil to give this to a non-chess player to study... don't do it... don't do it!

lol it's a start

I wrote an engine that 'randomly' moves all the pieces to spell the word 'BEER'.

I didn't answer your question before? Otherwise I'd quote your comment...

Mine does it the second way--pick a piece, then pick a move.  And yes, right now it is just to test if the rules and the moves work.

Is Rh2 a novelty?

i like 1.f6

no 2. h3

Instead of this hassle, you should have programmed an engine/bot to do this for you.