Why do humans play chess in such a risk-averse manner?

Today I ask why computers playing among themselves have produced livelier games than recent matches of humans equipped with computer preparation.

That is from Kenneth Regan, much more at the link.  And here is part of his answer:

The reason may literally be that the computers have greater contempt for each other. The contempt factor is a term in a program’s evaluation function that makes it pretend to be a couple tenths of a pawn better off than it is, in situations where a drawing or drawish continuation is available.

The computers also have no awareness of high stakes that puts “staying in the game” ahead of maximizing one’s chance of winning.

The context of course, is the recent Anand-Gelfand world chess championship match, which featured unprecedented levels of computer preparation, and, arguably, a lot of very boring games of “theoretical interest” only.

So will iPhones make us all more boring?

Comments

completely off topic but...

Markets in everything, caskets edition: http://www.washingtonpost.com/politics/louisiana-monks-go-to-court-to-sell-their-caskets/2012/05/29/gJQA7VMK0U_story.html

What was the difference between winning the Anand-Gelfand match, and finishing second?

Not being the "Lightning Kid". It was champion's prerogative all over again - Anand knew he just had to make it to the rapids.

It was about 500K. 1.5 MIlion to the winner, 1 Million to the loser, so you win either way. They played a lot of short junk draws as well. Carlsen and Nakamura are the only players left with a pair onions on them.

What about Aronian, Topalov, Ivanchuk and Radjabov?

Is there a objective quantitative index to evaluate the boring-ness of a game?

Yes. When even Kramnik thinks they should have played on: http://chessbase.com/newsdetail.asp?newsid=8194

I think that the apt Tyleresque response to 'will iPhones make us boring?' is that we are already more boring than we think we are.

Tyler as usual is pretty late to the trend.
Boring though in this case is a good thing, its just means that the market for human concentration and mind-space is becoming more competitive.

very nice. +100.

The games were not boring because the other available entertainment options were better but because lines are known very deeply and to little advantage for either side.

The thing that make chess games exciting to observe is attacking speculative moves like those of M. Tal. Preparation cuts down on the chances that those moves will be played in a game because GMs will have looked at them with computers and know that the attack fails.

Thanks. We are all Cowensians now.

At a dinner party recently the spouse of a physician continually used her iPhone to come up with clever facts and ripostes. She was the most boring person in the room.

I think big data has already made politics and business more boring than it was.

Yes, but computers are also known for taking ages to figure out that they are in a completely positionally drawn position. Though Rybka's not so bad these days.
I think risk-aversion in chess is the same thing psychologically as risk aversion in markets.

How could it not be?

The punishment - now Anand will have to live with his passivity forever: http://chessbase.com/news/2012/moscow/closing11.jpg

I can't read this article without still thinking about (and laughing at) #tylertweets.

This psychological phenomenon is old hat in competitive poker. The difference between "Oh no, what if he has the cards" and "It might be time to teach this fish he can't use position to push me around" is a big game changer. This is why pretty much any poker book you read tells you not to play higher stakes than your bankroll justifies. You get too cautious because this specific instance matters too much. Similarly, getting aggressive near the bubble is almost always big +EV but it's completely contrary to what instinct tells you to do (until you condition yourself otherwise).
Optimal strategies for an iterated prisoner's dilemma are very different than for a single game. We seem to naturally want to "stay in the game" and maximize our chance of doing OK right now vice a higher volatility but higher expected payoff strategy. The best players in many competitive activities either don't have or have found a way to neutralize this instinct.

Also from the link:
"According to my statistical model of player move choice mentioned here, this match had the highest standard in chess history. ---------------------------------------------------------------------------------------------------------------------------------------------- Still, the players had those moves in their head as they sat down computer-free at the board, and if what matters is the quality of the moves made by their hands regardless of where they came from, then this was history’s human chess pinnacle." Sounds quite different from the spirit of the Post ??

Because humans are risk averse and computers aren't.

As TC describes, computers can even be tweaked to be risk-loving - presumably this is what is meant by 'contempt'.

On a different matter, there's something very wrong that the world championship was not between Aronian and Carlsen.

Also, like in poker and backgammon, strong computers can play in seemingly reckless ways because they will play precisely on later moves; by contrast, a move that is optimal in a game-theoretic sense might be a disaster for a weaker player because it leads to demanding positions ahead.

What do you know about "Mathematics of Poker?" :).

If iPhones make people more boring, it will be because the iPhones become more interesting. We must get ahead of the curve and define marriage in law as a sacred bond between two people who are both biologically human.

I suspect there are more factors that go into play as to why the match was relatively boring. I say this taking into account that the recent short matches between Anand and Kramnik, Topalov and Anand, and Kramnik Topalov, were also contests with an incredible degree of computer assisted preparation. A couple of decades earlier there was a match between Anand and Kasparov where the computers were very very important. Take for example game 10, where Kasparov played a very well known variation and introduced a novelty (Bc2) which sacrificed a whole rook to enter a winning endgame. I also believe Anand is simply not very excited about playing chess. He has been a Botvinnik style world champion, who wins his matches but can't win tournaments, and in this match it was Gelfand who seemed to have the upper hand in general terms.

Now, going back to the computer and preparation, it is no secret that memory has become crucial for top level chess. You literally have to memorize hundreds of moves/lines/variations if you are going to play the popular openings. That is one of the reasons why Alexander Morozevich plays less known lines, but it seems in the ling run this strategy might not be very strong to make it all the way to the top. At the end it seems Fischer might have been right. We need to shuffle the pieces at the beginning of the game so games really start earlier, not at move 30 after all the theory has been repeated...

Would it be an interesting variant if Chess games had to start from random positions? Or more likely random starting positions that are computationally evaluated to be not overly asymmetrically advantaged to one side.

Well, as of now the idea that has been implemented is for both sides to have exactly the same computer generated random position. Tournaments have already been run using this scheme and a common thing is that "normal" chess position do come out. The general principles of chess at the end impose themselves most of the time irrespective of what is the starting arrangement. The important thing is that, again, its not necessarily about who knows more variations or has the better chess engine to confirm analysis.

On a related note, Kasparov once said that he checked the opening analysis which he had developed during the last 30-35 years of active play, and the engines did find lots of mistakes, but still, relatively speaking, most of his analysis were correct. In the earlier days he used the computer mainly as a database.

Rahul: This is what you want http://chess960.net/

To be sure, my web article recommends a non-random variant, Bronstein's baseline chess with Fischer's castling rule, which can lead to even more variety.

"He has been a Botvinnik style world champion, who wins his matches but can’t win tournaments" except this one, of course: http://en.wikipedia.org/wiki/World_Chess_Championship_2007
Of course the same goes for Botvinnik: http://en.wikipedia.org/wiki/World_Chess_Championship_1948

yes of course, but you know very well what I mean. They simply couldn't dominate tournaments like Kasparov or Karpov

I think the match reflected the stakes on each side. Anand did not have to prove he is a better player than Gelfand -- that's demonstrated by ratings differences and years of results. All Anand had to do was get through it with the title intact, which in practical terms meant making it to the rapid playoffs. For Gelfand it was a big success to show that he could play with Anand, that he belonged at the same table. Draws work fine for this. In addition, Anand is a much better tactician than Gelfand, so Boris had no incentive to mix it up. The one time he did he got royally burned, smashed right out of the opening. But Anand had little incentive to mix it up either.

The Aronian-Kramnik match was superb, however, a pleasure to follow. Aronian has a crowd-pleasing style and has come a long way in terms of soundness. I think he still has a tendency to underestimate defensive resources; he thinks that the initiative is worth a shade more than it actually is. But better this than the other way around.

The latest article at ChessVibes.com (http://www.chessvibes.com/reports/anand-gelfand-the-aftermath?page=1) has more to that effect, including this quotation from Judit Polgar on a blog post:

"Some say the 12 classical games were not so exciting. I have admit, I think that some of them could have been played out longer really. On the other hand, the rules do not forbid to offer a draw any time of the game at all. So when you have big, no sorry, ENORMOUS pressure and there is nothing more you want but to win the overall match and you believe that half a point makes you to get closer then, of course, you do not care about the spectators, if they are happy with what they see or not."

She also says Anand "was not in great form" while Gelfand was "fighting [for] his life"---opinions shared by many, but again the cold numbers come out Anand 3002, Gelfand 2920.

Comments for this post are closed