Life In 19x19

sorin wrote:It occurred to me that Carlsen's decision to offer a draw in a position where he was ahead, and most chess masters (including Kasparov) thought it doesn't make sense ("why not just play and win the 12th game and be done with it?") - is somehow similar to some seemingly stupid decisions that strong Go AIs makes in yose, when they clearly throw away points when ahead, just to maximize winrate (and win anyway).

I guess Carlsen made his judgement using a "minimize risk at all costs" strategy, just like AlphaGo does, which to "mere mortals" it may look wrong.

Not to disagree with Carlsen, but it's not like mere mortals don't minimize risk when ahead. The thing is, go AI bots, when ahead, sometimes make moves that seem stupid in the endgame, even to humans who know how to play solidly and conservatively.

Now, human endgame calculations generally assume perfect play, which may be unrealistic. AI winrates assume imperfect play, with mistakes. The question is, which mistakes? The assumption of imperfect play may also be unrealistic.

To give an exaggerated example. Suppose that two shodans are playing an even game. In the endgame the winrate calculations assume that they make 10 kyu mistakes by both players. Playing to maximize such winrates may lead to 5 kyu mistakes. Which may then not only lose points, but may actually put the game in jeopardy against the actual shodan opponent.

To analogize to Carlsen's decision, he may have made his winrate calculations on the assumption that in rapid play he might make 5 kyu mistakes while Caruana would make 10 kyu mistakes.

Ken Regan is the foremost researcher of the objective evaluation of chess strength by looking at the moves played (as evaluated by computers), as opposed to results. In this post he estimates that the drop in quality from classical to rapid is 200-300 Elo.

I took this much time getting into this because checking that MAYBE my copy of Dine's "Chess Marches On" had survived the 2006 house fire << there are several thousand books here, so not that fast to check >> So I can;t quote, but to paraphrase .....

The reason is when posting a few rapid transit championship games (NY club) against Reshevsky he discussed the loss of strength when playing 10 second chess. But also that not all grandmasters were affected the same, so who strongest at normal time (40/2 hours then) not necessarily the same at 10 seconds.

This is of course famously also true with blindfold. Koltanowski not one of the top grandmasters a regular chess but at simultaneous blindfold he was.

I am rather sure that the same is true with go. It is not just that the blitz games aren't as good but that which players best at abnormally short time not going to be the same as if the games were played at normal time controls.

Indeed there is a difference in ability between playing "classical" chess vs rapid or blitz chess, and longer time limit go vs. fast games. Athletes know thta runners who have ability at long races (1 km +) are not so good at sprint races (e.g. 200 m). In the case of runners it it is known that different kinds of muscles favor the two kinds of race. Sprinters have more "fast twitch" muscle than distance runners "slow twitch". In blitz games there are different techniques used than in slow games. I, also, have heard from pro go players that there are more complicated fights in fast games than in slow games. Supposedly in slow games the player can read out the fight and see whether it is worth starting the fight. In a fast game, when there is not time enought to read out complicated fights, a player gambles that the opponent will make a mistake first (or last). In running races the sprinters won't compete against the distance runners in longer races and, vice versa, the distance runners won't compete against sprinters in sprint races. This dichotomy exists in bicycle races, most noticeable in races involving hilly courses versus short and straight courses. I know it wouldn't happen but I propose not mixing fast games and slow games in determining a champion. Speaking for myself alone I'll just mention that in chess my rating for "daily" games on chess.com is 400 points higher than my rating for 10 minute sudden death games.

As I've said before, slow vs fast is an oversimplification. Humans are not robots such that reading_quality = some_factor * time_spent_thinking.

No, humans get fatigued sometimes. Minds wander, sometime. Sometimes they over complicate things. And sure, sometimes they don't have enough time to come up with a good analysis.

So yeah, too much time can be a problem. Overthinking can be a problem. Too little time can be, too. Health can be a problem. Bad dinner last night can be a problem.

But to say any of these are a problem, we have to first identify what we are trying to achieve. Entertainment for viewers? Identification of a world champion? Quality of life for pros?

Without clarifying these specifics, this discussion is pretty meaningless.

By the way, may I ask the reason for the term "Mickey Mouse"? Is there any reason for this nomenclature other than to annoy me? Is there something fast about what Mickey Mouse does? Or is it simply to illustrate tournaments with fast time settings as childish?

While we are clarifying things, I'd like to know.

They are not childish, they just something which cannot be treated with the same respect as the Full Monty. John of course is a Full Monty man, particularly when it comes to Python.

In Chess, we have Armageddon Time Limits because:
1 - The scenario 'Retention of the title in the case of a draw' has become commercially distasteful.
2 - The sponsorship money FIDE has only goes so far.

Gone are the days when the world championship funds were in such excess that they could be spent on driving fleets of black taxis around London. These are the days of the tight budget.

What is the knock-on effect of such cartoon structures? The standard of Chess seems to be very high in this title match.
If the GMs only spend their time on rapid play, I can imagine that it would be normal for their standard of play to suffer. I think that they, like Go players, probably don't only spend their time in rapid play events. So I don't really buy that issue as being a super serious one, only a distasteful intrusion of reality into the game.

Kirby wrote:By the way, may I ask the reason for the term "Mickey Mouse"? Is there any reason for this nomenclature other than to annoy me? Is there something fast about what Mickey Mouse does? Or is it simply to illustrate tournaments with fast time settings as childish?

While we are clarifying things, I'd like to know.

Kirby wrote:Still, I think that you do have a point in that, rather than getting too upset or frustrated over what I perceive as biased, it would be more proactive of me to translate material on my own, and bring my own contributions to the table (though, I'm sure you'll still be able to bait me in the future by using "Mickey Mouse" in your posts).

I'm quite surprised this hasn't been settled in the last seven years actually, I'm curious about the origin of the term myself. It does seem to come from Chess, but only appears in informal discussions and analysis from what I've seen.

The term "Mickey Mouse [x]" arose shortly after the cartoon was produced by Disney. It was perhaps most applied within Business - A Mickey Mouse Operation.

Time settings are established with two goals in mind:

• Determine who is the strongest player
• Make the competition enjoyable for spectators.

Long enough time settings are necessary to fulfill the first goal. One certainly would not say that someone is the "best painter" after looking the productions he makes in 5 minutes.

Concerning the second goal, I said above that "intermediate" time settings would be ideal for me.

I suppose that short time settings are called "Mickey Mouse" because spectators watch blitz games like they would watch cartoons, considering them as easy entertainment, with a lot of action going on, and without taking much time to think.

There is nothing wrong with Mickey Mouse; cartoons or blitz games can also be enjoyable to watch, but that's a more superficial activity.

The term "Mickey Mouse [x]" arose shortly after the cartoon was produced by Disney. It was perhaps most applied within Business - A Mickey Mouse Operation.

The term certainly goes back a long way, at least to my childhood, and my understanding then of its origin was that it comes from Glasgow. If you can say "this disnae work, that disnae work" you have a Disney ("does not") operation, but MM was the public face of Disney (much more so then than now).

A shame Kirby egotistically assumes a useful term I and my compatriots have used for over 60 years - well before he was born - is used just to annoy him. Dinna fash yersel, bonnie lad.

John Fairbairn, my underline wrote: A shame Kirby egotistically assumes a useful term I and my compatriots have used for over 60 years - well before he was born - is used just to annoy him.

But it's a nice little bonus, eh?

jlt wrote:Time settings are established with two goals in mind:

• Determine who is the strongest player
• Make the competition enjoyable for spectators.

Long enough time settings are necessary to fulfill the first goal. One certainly would not say that someone is the "best painter" after looking the productions he makes in 5 minutes.

But what is "long enough"? Is it 2h/player? 4 hours? 8 hours? 16 hours?

John Fairbairn wrote:

The term "Mickey Mouse [x]" arose shortly after the cartoon was produced by Disney. It was perhaps most applied within Business - A Mickey Mouse Operation.

The term certainly goes back a long way, at least to my childhood, and my understanding then of its origin was that it comes from Glasgow. If you can say "this disnae work, that disnae work" you have a Disney ("does not") operation, but MM was the public face of Disney (much more so then than now).

A shame Kirby egotistically assumes a useful term I and my compatriots have used for over 60 years - well before he was born - is used just to annoy him. Dinna fash yersel, bonnie lad.

But those of us with long enough memories remember the real story!

From Chess one might argue long time settings don't help you find the strongest player (in terms of their intuition for the game, i.e. Carlsen) but it becomes more of a "who can remember more of stockfish's opening lines and pick appropriate ones based on what you think your opponent prepared too", and all the super GMs are more similar to each other in that respect than the more varied intuitive skill that faster time settings reveal.