I have been reading "The effects of time pressure on chess skill: an investigation into fast and slow processes underlying expert performance". The implications are very interesting for Go as well as chess...

"Abstract The ability to play chess is generally assumed to depend on two types of processes: slow processes such as search, and fast processes such as pattern recognition. It has been argued that an increase in time pressure during a game selectively hinders the ability to engage in slow processes. Here we study the effect of time pressure on expert chess performance in order to test the hypothesis that compared to weak players, strong players depend relatively heavily on fast processes. In the first study we examine the performance of players of various strengths at an online chess server, for games played under different time controls. In a second study we examine the effect of time controls on performance in world championship matches. Both studies consistently show that skill differences between players become less predictive of the game outcome as the time controls are tightened. This result indicates that slow processes are at least as important for strong players as they are for weak players. Our findings pose a challenge for current theorizing in the field of expertise and chess."

This is a very interesting finding. To rephrase the results, it found that strong players become weaker, relative to their own "slow" strength, than weak players do as time controls become shorter. Personally I think that if we assume as a very general idea that shorter time controls inhibit players from fully using all their skill, then the fact that those players with more skill show the larger declines should not be surprising.

Among other things, the paper contains the interesting observation that the frequency of blunders clearly increases in blitz games. However, since weaker players frequently blunder anyway, the relative impact of this increase is greater for the more skilled players! LOL

_________________
Dave Sigaty
"Short-lived are both the praiser and the praised, and rememberer and the remembered..."
- Marcus Aurelius; Meditations, VIII 21

Yes? Maybe you misinterpret the findings? Or maybe I cite the wrong part of it? The cited part suggests a greater variance in strong players' skills. Although by experience I would confirm your statement that strong players become weaker with shorter time limits, it does not follow from the citation. Maybe in the paper itself such an additional result is described? I will have to read it.

Like many research papers, the findings are stated in fairly academic terms. I may be interpreting them incorrectly. Read the paper and see what you think! I also found an interview of the chief author in this digizine where he describes the background of the study.

_________________
Dave Sigaty
"Short-lived are both the praiser and the praised, and rememberer and the remembered..."
- Marcus Aurelius; Meditations, VIII 21

To the question: Is lightning chess not a completely different game to traditional chess? the scientist replies: The rules are the same, but aspects like luck, bluff, coping with stress, become relatively more important.

Down with Mickey Mouse go!

Lightning means that certain things can't be read but have to be decided by other means (like best guess by experience, greatest winning probability estimate or database knowledge). Not luck though.

Interesting! I just skimmed it, but I'll have to read it fully when I get the chance. It makes sense that time controls effectively determine the weight "fast skills" vs. "slow skills" will play on the outcome, however I can't help but wonder if all of the assumptions they make/prior results they include (which may be perfectly valid for chess) still apply to go. For instance, they say that experts search deeper, but not wider. Chess has a relatively narrow tree width (compared to go), from my experience one of the things that will distinguish stronger go players from weaker ones is their ability to see, or at least check, blind spots (it is not uncommon for me to look at a stronger player's game and think "Wow, I hadn't even considered that point"). Nevertheless, I'm sure much of the core result would be similar (lightning hurts most those who would use their time in ways that increase their ability the greatest).



I would contend that any time you are playing into a position where you have not calculated the final result there is an element of luck.

It is not an element of luck but of basing one's decision on even less complete information than usual. Luck does not exist in the decision making (unless you are able to roll a mental die) but one might say in the perfect play game result, which follows one's decision, being unknown. It is unknown for all but the late endgame decisions though.

***

The study finds that fast games hurt stronger players relatively more than weaker players, mentions that other studies have said the opposite and makes various speculations about possible causes but does not prove any finding nor cause. I would share the paper's major guess of cause (the relative blunder ratio) but, of course, also cannot prove it (regardless of having made similar observations with my KGS fast games vs. playing the same opponents in slow real word games).

So the paper says almost nothing but was fun to read anyway;)

We are not measuring decision making, we are measuring game outcomes and using that as a proxy for measuring decision quality. If the outcome of the game is not calculated before each decision, there is an element of luck in the result for the players. In this case we are not looking at whether the decisions of players lead to maximum probability of winning given known information, we are looking at how the games actually resulted.

Suppose that a player A consistently made decisions that should lead to an inferior position, however exploiting them required a 9-ply depth search. Given the time constraints player B is only able to do a 6-ply depth search, and thus these moves instead consistently lead to an advantageous position for player A. Neither player is aware they are mistakes until the post-mortem, however they are ultimately realized to be mistakes. I claim it would be reasonable to say that player A has been lucky, and that while player A might be able to consistently leverage a net positive result (even over time) with these types of moves at short time controls, they are not necessarily good decisions.

You use this term a lot. Did you have a bad experience at EuroDisney?

In go, one slow process is estimating the score and the more your game depends on an accurate estimate, the more likely it is to suffer under tight time controls. I don't know a lot about chess, but I've heard people say that positional assessments can be made more quickly in that game.

Even score counting is much quicker and more accurate for stronger players though.

_________________
We don't know who we are; we don't know where we are.
Each of us woke up one moment and here we were in the darkness.
We're nameless things with no memory; no knowledge of what went before,
No understanding of what is now, no knowledge of what will be.

Not really; he once saw when Mickey's hand was pointing at 3 and Goofey's tail at 7, that the watch he had been sold was not really a Rolex.

Best wishes.

_________________
No aji, keshi, kifu or kikashi has been harmed in the compiling of this post.
http://www.gogod.co.uk

Agree with the first conclusion, not so sure about the second. My reading is that the paper was interested in examining differences between slow processes (reading depth) and fast processes (pattern recognition), but found that the experimental chess results shed no light on this question, as they were dominated instead by "blunder" noise. Note that only strong (high dan level) chess players were involved, and in chess a fairly small blunder can easily decide the outcome of the game, even against a considerably weaker opponent.

All fine and well, but I was talking about the (strong) players' own perspective. In the paper, also both perspectives are used. The studies study the results but the speculation / general discussion is about how the players think.

This could be one of those processes that goes by the wayside when the time controls are tightened. The stronger player may count better and faster, but perhaps they are also more accustomed to tailoring their strategy to what they count. Play fast enough that there isn't time for either player to get a good count and it will hurt the player who is used to counting more than the one who is not. Perhaps it would be more accurate to say it removes an advantage that the player who regularly counts would typically have. Still though, it is bringing them closer to equality.

Fair enough, even from the player's perspective, I would say that if you make a poor move that you do not realize is a poor move at the time, and your opponent fails to punish it, you should think of yourself as lucky.

Fifteen or twenty years ago at a US Go Congress I was playing blitz (5 minute absolute time) games with a Japanese pro. He said that pros generally feel that in blitz go between pros and amateurs the amateurs need at least two handicap stones more than in slow time limits. Doesn't that seem to contradict the results of the paper since pros are stronger but amateur strength drops more?

Taking dynamic handicap against Saijo, then 8p, in 1999 he beat me down from H3 to H9 in 10' lightning, when I finally won. In 2002, we played a similar number of lightning games of which he won all H3 while I won all H4 games. Slow games I won most H3 games. - So I'd say it depends. If Black does not know how to win, then any handicap is necessary, quite like in amateur-amateur lightning games. If Black knows how to win, then one stone more than in slow games can suffice.