Thursday, 12 May 2011

A kind of Turing Test

A number of years ago (20 or more at a guess), I watched an interesting game at Belconnen Chess Club. It was between one of the club stalwarts and a newcomer to the club. The newcomer was quite a strong player (1800 ish), while the stalwart was a solid 1200 player. As it turned out, neither were aware of the others rating, and after much back and forth the game ended in a hard fought draw.
After the game the newcomer was chatting to me and asked about his opponent. "He's pretty good" he said, "he's rated about 1700/1800, isn't he?". "Ah, no" I replied, and informed him of his opponents real rating. And of course a couple of minutes later the stalwart also has a chat to me. "The guy I played, he isn't that good. He'd be about 1200, wouldn't he?"
I've always wondered if knowing a players rating has an influence on how we approach the game. For me I would say it is definitely the case, while for others I suspect they would play the same way regardless.
So here is an exercise. The following game was played recently, but I've left the names and ratings of the players out. How strong do you think both players are? (Fell free to give a numerical value, or some general description instead)

1.c4 e5 2.g3 Nf6 3.Bg2 Bc5 4.Nc3 c6 5.Qb3 0-0 6.e3 a5 7.Nge2 Na6 8.a3 Nc7 9.d4 exd4 10.exd4 Ba7 11.0-0 Re8 12.Be3 Ng4 13.Bf4 Ne6 14.h3 Nf6 15.Rfd1 d6 16.d5 Nc5 17.Qc2 g6 (D)
18.dxc6 Bf5 19.Qd2 Nb3 20.Qxd6 Qxd6 21.Rxd6 Ne4 22.Nxe4 Bxe4 23.Bxe4 Rxe4 24.Nc3 Rxc4 25.cxb7 Rb8 26.Rad1 Rxb7 27.Bh6 f5 28.Rd8+ Kf7 29.R8d7+ Rxd7 30.Rxd7+ Ke6 31.Rxa7 1-0


Paul said...

Knowing an opponent's rating is a doubled edged sword...I prefer not to know before the game...

Paul said...

1500's ??

PeterPullicino said...

both players play very strangely in the opening - eg. 5.Qb3?!. Black's c6!?... which leads me to believe both players are under 2000.

There does not seem to be much tactical awareness, or it was played at rapidplay speed.

I'm guessing White shows more understanding of the tactical potential of the pawn (cxb7 etc..) and back rank weakness, but maybe just got lucky?!

Anonymous said...

Black: 1200-1400?
White: 1500-1700

Anonymous said...

Black 1200
White 1500

Too chicken to put my name here, as I would hate to be completely embarrassed. :)

Mathew said...

Ratings have a huge influence on most people, had the 1800 newcomer known he was playing a bunny rated around 1200 he would of course played till the bitter end and made the weaker player prove at every part of the game he can make reasonable moves. The game you showed looks pretty sad for black but I would be tempted to say white is 1500 and black played a bad game and is rated close to 1950.

Anonymous said...

Shaun - will you post the answer?

I like Mathew's analysis - I think Black was playing for tricks against a weaker opponent, but got busted due to the power of the pawn c6xb7. - Peter

Shaun Press said...

Thanks for all the contributions, and most of you were in the ball park re ratings, especially for white. White was just a little over 1400, but Black was just a little under 1000 (and therefore may be pleased with his estimated rating). Nonetheless I thought Black played pretty well, and even the losing move (g6) made some sense, in that it looked like Bf5 was winning material. I was looking at the game on and off (while it was being played) and after the knight landed on b3 I wasn't sure what White had planned. It turned out White had seen what was coming after g6 and had calculated far enough to know it wasn't dangerous.
As for the opening, it is known to theory with Portisch even playing 5.Qb3 in that position.

Pete (Ireland) said...

You might be interested that somebody here (Ireland) was intending to test the significance of knowing one's opponent's rating in his (undergraduate)Psychology dissertation. The idea was to get volunteers to play each other on line, with some players knowing their opponents' ratings but others not, and then see if there was a difference in how people played. I'm not sure if he has completed this yet, but I can try to find out his findings if you are interested. I don't know if he included a computer to make it a real Turing test.