I get my kicks above the waistline, sunshine
May 29, 2014 6:25 AM Subscribe
Randy Olson is conducting an analysis of chess since 1850. What's the advantage of playing white? Are games getting longer? What openings have fallen in and out of vogue? Are chess players becoming less focused on capturing pieces?
There is some interesting data here but it's hampered a little bit by the poster not really being a chess player, so there's been some misuse of terminology and some data is interpreted in an unidiomatic way. (I've been talking with him some on Hacker News.)
posted by dfan at 6:38 AM on May 29, 2014 [1 favorite]
posted by dfan at 6:38 AM on May 29, 2014 [1 favorite]
I just looked at the latest one on captures, and yeah, his hypotheses are kind of weird. (There was an influential book on rook strategies in 1850? Bobby Fischer got everyone to stop using knights?)
The trends of piece usage are interesting, but I'd want to see some other data before trying to interpret them. For example, I'd want to see what things look like if you cut games off at move 40, since you don't really want to count endgames as much.
posted by dfan at 6:46 AM on May 29, 2014
The trends of piece usage are interesting, but I'd want to see some other data before trying to interpret them. For example, I'd want to see what things look like if you cut games off at move 40, since you don't really want to count endgames as much.
posted by dfan at 6:46 AM on May 29, 2014
He's got some great posts aside from this; be sure to check out the rest of his blog.
posted by a snickering nuthatch at 7:04 AM on May 29, 2014
posted by a snickering nuthatch at 7:04 AM on May 29, 2014
Also charts like "Knight Move Rate" conflate "people decided to move knights more" and "people decided to trade knights less", since you can't move a piece you don't have.
But enough caviling! I will look at his other posts on topics I am less tetchy about. Certainly his presentation is nice.
posted by dfan at 7:18 AM on May 29, 2014
But enough caviling! I will look at his other posts on topics I am less tetchy about. Certainly his presentation is nice.
posted by dfan at 7:18 AM on May 29, 2014
the advantage of playing white was addressed definitively by grandmaster yefim bogoljubov in the 1920s. "when i'm white, i win because i'm white, and when i'm black, i win because i'm bogoljubov."
posted by bruce at 7:24 AM on May 29, 2014 [18 favorites]
posted by bruce at 7:24 AM on May 29, 2014 [18 favorites]
On a chess-related note, Dragoljub Velimirović (1942-2014), died last week. Taught chess by his mother--one of the top Yugoskavian players--Velimirović was a brilliant attacker. While the deep understanding and elegant play of the greats like Fischer, Kasparov, and Carlsen is certainly deserving of reverence, for me the most exhilarating and intense games come from the aggressive, beautiful fever dream sequences of risk-takers like Tal, Keres, Nezhmetdinov, Marshall, Morozevich, Chigorin, and Velimirović.
posted by whatgorilla at 7:55 AM on May 29, 2014 [6 favorites]
posted by whatgorilla at 7:55 AM on May 29, 2014 [6 favorites]
It seems that there were a lot of vicious attacking players in Yugoslavia - Bela Perenyi comes to mind. The "Yugoslav Attack" in the Dragon, too.
posted by thelonius at 8:18 AM on May 29, 2014
posted by thelonius at 8:18 AM on May 29, 2014
Games are getting longer in part because draws are more common (in part because we're taking fewer pieces?). ;o)
I know we're just seeing more people playing more accurately (for instance, I've read that Carlsen's moves correspond to the top-chess computer choices at a higher percentage than any other player in history--though I don't know if this is true); however, I much prefer the swashbuckling days of yesteryear to these 50-70 (and even >100) move games...and the increase in draws.
With the massive databases of millions of games in a simple .pgn format, this kind of analysis wasn't hard to do--but I don't see any real revelations or surprises here.
THELONIUS: Check out the games of Nezhmetdinov if you want vicious. He's a monster of calculation and he's constantly leaving pieces hanging daring the opponent to take them and get attacked; I've seen games where he sacked his Queen on move 14 or 15, but the game went on for another 20 moves--and chess computers show that his sac was sound (today, I'd assume the players had gone over these novelties with computers, but he didn't have that resource available, which is why it's so astounding).
posted by whatgorilla at 8:22 AM on May 29, 2014 [10 favorites]
I know we're just seeing more people playing more accurately (for instance, I've read that Carlsen's moves correspond to the top-chess computer choices at a higher percentage than any other player in history--though I don't know if this is true); however, I much prefer the swashbuckling days of yesteryear to these 50-70 (and even >100) move games...and the increase in draws.
With the massive databases of millions of games in a simple .pgn format, this kind of analysis wasn't hard to do--but I don't see any real revelations or surprises here.
THELONIUS: Check out the games of Nezhmetdinov if you want vicious. He's a monster of calculation and he's constantly leaving pieces hanging daring the opponent to take them and get attacked; I've seen games where he sacked his Queen on move 14 or 15, but the game went on for another 20 moves--and chess computers show that his sac was sound (today, I'd assume the players had gone over these novelties with computers, but he didn't have that resource available, which is why it's so astounding).
posted by whatgorilla at 8:22 AM on May 29, 2014 [10 favorites]
Check out the games of Nezhmetdinov if you want vicious.
Wow, you aren't kidding.
posted by Blazecock Pileon at 9:31 AM on May 29, 2014 [1 favorite]
Wow, you aren't kidding.
posted by Blazecock Pileon at 9:31 AM on May 29, 2014 [1 favorite]
eponysterical
posted by Zerowensboring at 10:13 AM on May 29, 2014
posted by Zerowensboring at 10:13 AM on May 29, 2014
"What's the advantage of playing white?"
Your pawns never get caught in stop-and-frisk.
posted by klangklangston at 11:29 AM on May 29, 2014 [6 favorites]
Your pawns never get caught in stop-and-frisk.
posted by klangklangston at 11:29 AM on May 29, 2014 [6 favorites]
This is timely. I’m a chess n00B trying to teach my son chess. (Link to Ask MeFi)
posted by axoplasm at 11:59 AM on May 29, 2014
posted by axoplasm at 11:59 AM on May 29, 2014
Nezhmetdinov
Check this out! He throws away piece after piece for positional advantage, leading to the weirdest mate I've ever seen.
posted by justsomebodythatyouusedtoknow at 12:25 PM on May 29, 2014 [7 favorites]
Check this out! He throws away piece after piece for positional advantage, leading to the weirdest mate I've ever seen.
posted by justsomebodythatyouusedtoknow at 12:25 PM on May 29, 2014 [7 favorites]
White privilege slakes its thirst
White always goes first
Pawns lose their stride fast but
White Knights are the worst.
posted by aydeejones at 1:53 PM on May 29, 2014
White always goes first
Pawns lose their stride fast but
White Knights are the worst.
posted by aydeejones at 1:53 PM on May 29, 2014
As someone who doesn't play a great deal of chess, I was surprised to see how few different first moves there were. I always feel a bit boring playing e4 as white for example, because I would have assumed that there are lots of clever openings where you can juggle the first few moves about to mess with your opponent. Is it seen as strange if a top level player opens with something other than the top four there?
posted by lucidium at 4:01 PM on May 29, 2014
posted by lucidium at 4:01 PM on May 29, 2014
Is it seen as strange if a top level player opens with something other than the top four there?Yes. For example, Baadur Jobava (36th highest rated player in the world) played 1.b3 in a bunch of games this year and it caused a big stir.
Keep in mind that there's still a whole lot of interesting choices you can make after the first move, but it's quite rare for a top player to not start with 1.d4, 1.e4, 1.Nf3, or 1.c4.
posted by dfan at 6:23 PM on May 29, 2014
@Chrysostom nice title/reference!
posted by shoepal at 9:57 PM on May 29, 2014 [1 favorite]
posted by shoepal at 9:57 PM on May 29, 2014 [1 favorite]
I'm curious about the 'huge collection of 675,000+ chess tournament games'
I got a bargain-basement Chess Assistant 10 years ago, and it came with 2.24 million games from 1560-2003. I'm sure he could find a bigger data set to work with.
I like the changes over time idea. This is very interesting stuff, and cool diagrams, but it would be even more interesting with a top-level player contributing to the analysis.
He is surprised that so few games end in checkmate. I'm surprised it happens at all in master play.
In the ELO rating/wins picture, it would be nice to see a line to show the theoretical prediction. Or is the whole idea stupid? Do ELO ratings strongly predict winning percentages because the ratings themselves are generated from wins and losses?
posted by MtDewd at 6:22 AM on May 30, 2014 [1 favorite]
I got a bargain-basement Chess Assistant 10 years ago, and it came with 2.24 million games from 1560-2003. I'm sure he could find a bigger data set to work with.
I like the changes over time idea. This is very interesting stuff, and cool diagrams, but it would be even more interesting with a top-level player contributing to the analysis.
He is surprised that so few games end in checkmate. I'm surprised it happens at all in master play.
In the ELO rating/wins picture, it would be nice to see a line to show the theoretical prediction. Or is the whole idea stupid? Do ELO ratings strongly predict winning percentages because the ratings themselves are generated from wins and losses?
posted by MtDewd at 6:22 AM on May 30, 2014 [1 favorite]
He mentions a few times that he only has reliable data from 1850. Not sure what makes it reliable or not.
posted by Chrysostom at 6:29 AM on May 30, 2014
posted by Chrysostom at 6:29 AM on May 30, 2014
Do ELO ratings strongly predict winning percentages because the ratings themselves are generated from wins and losses?The ratings are indeed generated from wins and losses, but you also need to take a lot of care to make them predict future results well. For example, you could imagine just giving players 1 rating point for a win and -1 rating point for a loss regardless of who they played. Arpad Elo (after whom the rating system is named) came up with a nice statistical model for player strength and performance, and designed the rating system to reflect it. It's not perfect (people have made recent tweaks that increase its predictive power, for example adding a variability term so people with more stable ratings are taken more seriously), but it's pretty good.
posted by dfan at 7:43 AM on May 30, 2014
Yes. For example, Baadur Jobava (36th highest rated player in the world) played 1.b3 in a bunch of games this year and it caused a big stir.
I think Pavel Blatny plays this move a lot
posted by thelonius at 7:58 AM on May 30, 2014
I think Pavel Blatny plays this move a lot
posted by thelonius at 7:58 AM on May 30, 2014
I think Pavel Blatny plays this move [1.b3] a lotYep, although he's just an "ordinary" GM (rated #1772 in the world among active players).
posted by dfan at 8:09 AM on May 30, 2014
Wouldn't you expect fewer checkmates from strategic games??
posted by nzero at 10:31 AM on May 30, 2014
posted by nzero at 10:31 AM on May 30, 2014
At the top levels, I wouldn't expect any, but sometimes even Grandmasters get surprised.
There was a scene in a Mission Impossible episode where the IMF guy is posing as a grandmaster (?), but is really using a computer, and the grandmaster he beats is totally shocked. I was a teenager at the time I saw it, and not a good player, but that looked pretty hokey to me even then.
I'm not sure what kind of expectations you can have about checkmate. If you required games to be played out, then all games would end in checkmate or a draw. Almost all decisive games end with resignation, at a point (and often past it) when a player sees he has no reason to continue. A 'strategic' game might just take longer to get to the end.
posted by MtDewd at 5:56 PM on May 31, 2014
There was a scene in a Mission Impossible episode where the IMF guy is posing as a grandmaster (?), but is really using a computer, and the grandmaster he beats is totally shocked. I was a teenager at the time I saw it, and not a good player, but that looked pretty hokey to me even then.
I'm not sure what kind of expectations you can have about checkmate. If you required games to be played out, then all games would end in checkmate or a draw. Almost all decisive games end with resignation, at a point (and often past it) when a player sees he has no reason to continue. A 'strategic' game might just take longer to get to the end.
posted by MtDewd at 5:56 PM on May 31, 2014
« Older It's called Vindaloo. | A train across Ukraine Newer »
This thread has been archived and is closed to new comments
posted by empath at 6:37 AM on May 29, 2014