Slashdot Mirror
A Data-Driven Exploration of the Evolution of Chess
HughPickens.com writes Randy Olsen has a interesting article where he explores a data set of over 650,000 chess tournament games ranging back to the 15th century and looks at how chess has changed over time. His findings include:
Chess games are getting longer. Chess games have been getting steadily longer since 1970, increasing from 75 ply (37 moves) per game in 1970 to a whopping 85 ply (42 moves) per game in 2014. "This trend could possibly be telling us that defensive play is becoming more common in chess nowaday," writes Olsen. "Even the world's current best chess player, Magnus Carlsen, was forced to adopt a more defensive play style (instead of his traditional aggressive style) to compete with the world's elite."
The first-move advantage has always existed. White consistently wins 56% and Black only 44% of the games every year between 1850 and 2014 and the first-move advantage becomes more pronounced the more skilled the chess players are. "Despite 150+ years of revolutions and refinement of chess, the first-move advantage has effectively remained untouched. The only way around it is to make sure that competitors play an even number of games as White and Black."
Draws are much more common nowadays. Only 1 in 10 games ended in a draw in 1850, whereas 1 in 3 games ended in a draw in 2013. "Since the early 20th century, chess experts have feared that the over-analysis of chess will lead "draw death," where experts will become so skilled at chess that it will be impossible to decisively win a game any more." Interestingly chess prodigy and world champion Jose Raul Capablanca said in the 1920's that he believed chess would be exhausted in the near future and that games between masters would always end in draws. Capablanca proposed a more complex variant of chess to help prevent "draw death," but it never really seemed to catch on.
Chess games are getting longer. Chess games have been getting steadily longer since 1970, increasing from 75 ply (37 moves) per game in 1970 to a whopping 85 ply (42 moves) per game in 2014. "This trend could possibly be telling us that defensive play is becoming more common in chess nowaday," writes Olsen. "Even the world's current best chess player, Magnus Carlsen, was forced to adopt a more defensive play style (instead of his traditional aggressive style) to compete with the world's elite."
The first-move advantage has always existed. White consistently wins 56% and Black only 44% of the games every year between 1850 and 2014 and the first-move advantage becomes more pronounced the more skilled the chess players are. "Despite 150+ years of revolutions and refinement of chess, the first-move advantage has effectively remained untouched. The only way around it is to make sure that competitors play an even number of games as White and Black."
Draws are much more common nowadays. Only 1 in 10 games ended in a draw in 1850, whereas 1 in 3 games ended in a draw in 2013. "Since the early 20th century, chess experts have feared that the over-analysis of chess will lead "draw death," where experts will become so skilled at chess that it will be impossible to decisively win a game any more." Interestingly chess prodigy and world champion Jose Raul Capablanca said in the 1920's that he believed chess would be exhausted in the near future and that games between masters would always end in draws. Capablanca proposed a more complex variant of chess to help prevent "draw death," but it never really seemed to catch on.
first post.
We evolve as chess players from enthusiastic amateurs who leverage our native skills to hard core analysts with a library of books on chess strategy. At what point does the game cease to be fun and become an obsession?
...omphaloskepsis often...
I actually had to read the article to figure this out. The statement that White wins 56% and Black 44% is for games in which a non-draw decision is reached (per the actual article). But with 10% to 33% draws, the actual difference in score is definitely lower. Conventional scoring is 1 for a win and 0.5 each for a draw.
So White does have a persistent advantage, but the spread is lower than 8% going by score. And I think you have to go by score, that's what counts in tournament play.
Let's say over the time period in question there are 20% draws (just for the sake of calculation). Out of 1000 games there are 200 draws. White wins 56% of 800 or 448, so Black wins 352. White scores 548, Black 452, or 54.8% to 45.2%. Still a clear White advantage, but somewhat less, and lesser still as we approach the modern 33% drawn.
First, for someone who has been playing chess competitively for the last twenty years, none of the results of the analysis is a revelation. Like so many "data" posts that seem to be in vogue, this one states quite the obvious viz the game of chess has evolved and has improved in quality. Hence opening colour matters, games are longer and many end in draws. DUH! As a secondary point, the OP makes a big show of the "steady increase" increase in length of game from the 1970s. On closer inspection, what is implied is that the average game has gone from 37 moves to 42 moves. For a chess player, that increase is hardly significant and can be attributed more as a result of prevailing opening theory and chess playing style than reflective of anything else. A clear case of data-blindness.
When did 'data driven' become a buzzword? And why is it a thing? Shouldn't every news article be data driven to some degree?
"First they came for the slanderers and i said nothing."
Please someone post a link to the first "first post" message on Slashdot. For the sake of documenting history.
My karma ran over your dogma
I used to help direct tournaments, including the US Chess Championship. The number of draws is pretty amazing, and part of it is the level of competition. They get material reduced to a certain point and position, realize that neither player is likely to make a catastrophic mistake, and offer a draw. I watched Josh Waitzkin and Boris Gulko battle it out to a closed position where they had lots of movement available, but neither could get a decisive advantage without a blunder, and that just wasn't going to happen. They drew, then went back to the break room to replay the game and see if there was a way around it.
And then there's a certain player who would offer a quick draw then go to the nearest casino to play poker, which he was quite good at and normally won more than the tournament would have produced. And still get his appearance fee for the tournament.
When you sympathize with stupidity, you start thinking like an idiot.
If you have ever written a chess playing program, or only a chess move validator, you would have seen how hard this very issue is. It is only slightly easier than checkmate & stalemate detection. The reason being, that you have to analyze the opponents' moves. So before you know, even when only validating, you are already recursing.
Then there are plenty ambiguous situations, like, a piece that could attack the back row, but cannot do so because it would place its king in check. Your next iteration will miss this as a valid move, however for the castling opportunity it matters. So you will have to write code just for such exception. Or you might not want to iterate at all. Other details are pawns that can only move forward but capture diagonal (admittingly more related to check/mate/stalemate). Promoting. En-passent. Castling. It's all stuff that easily eats up kilobytes of code. The rules of chess look clear, straightforward and relative simple, but actually implementing it reveals a truckload of special cases.
For a computer with only 16kB of memory, it is a forgivable error, since you need some logic for gameplay too, not only move validation. For PC software it's indeed very clumsy and no-one apparently ever bothered to validate it. However, the similarity of the error is most likely pure coincidental.
Although the opening may seem like that, the complexity of chess is such that it's unlikely that every board position has been played.
However, this is incredibly counter-intuitive because of the numbers involved.
Do you know how many combinations there are of a standard 52-card pack of cards? 52! (factorial) = 80658175170943878571660636856403766975289505440883277824000000000000.
It would take - on average - billions of years of billions of people each shuffling billions of deck a second to end up with the exact same deck twice.
Chess's complexity provides slightly less more possible states than that but potentially much larger (over twice as complex, so billions of billions of billions of billions of....) actual viable game trees. However, Go's complexity is greater even on a 9x9 board. On the standard 19x19 board it's UNBELIEVABLY more complex.
Feudal's complexity doesn't even come close.
There are about 300k games played per week just on FICS. There are a few hundred USCF games played each week just in Louisville KY (where I play). I would imagine if you managed to pull from all of the sources, 600k wouldn't even amount to a day's worth of games.
The set the author used suffers greatly from selection bias. Games are usually only included in commercial databases because they're interesting, or were played by interesting people. So I'm not sure anything interesting can be drawn from his results.
Also, there needs to be some control put in place to account for rating differences. The Eli system isn't that old, and in the past players with drastically different levels of skill were more likely to play each other.