7-man EGTB

[gambit3]

as far as i knew, there was never any coffeehouse 50-move rule, that one was 21 moves. also, it came AFTER the 50 move rule was introduced to counter one of steinitz's strategies - if the game looks lost head for infinity if possible. steinitz won more than his fair share of [recorded] games by playing on in a dead drawn position, such as pawn walls facing off with no possibility of advancement for either side, until his opponent would simply give up in frustration and anger. clocks were introduced at the same time to counter his taking sometimes 4 hours per move, leading to games that literally would last for a week. not something that is manageable across more than a few games if a world champion is to be selected in a reasonable time. thirdly, coffeehouse rule limits were speculated initially introduced by none less than philidor, and that after his initial entry into an organised international affair (involving, among others, steinitz).

The issue today is that one would certainly not want to see an obvious win frustrated by the 50-move flag being waved by the defender. An example is in KBBKN, where there are now tractable wins deeper than 50 moves.

agreed, though i had already responded to it with:

in this regard, the 50 move rule is, in my opinion (note: this is only opinion) a good one: it likely will have solved more disputes than it has caused. it may be time for it to change to a longer number (100? 200? how many known mates are there that are longer than 50 moves total, anyway? kbnk is already 34!), though i believe the rule itself should not be eradicated.

your papers, i am still digesting them with gusto. very interesting information here. thanks for the ref. i'll have to take some time to go through them and [maybe] give something additional, if anything exists.
just a question regarding your statistical paper detailing heinz' work (the entry with ''statistical significance'' in the title on the page you gave in your ref): shouldn't it be redone given rybka? the effectiveness of the endings between the two engines are worlds apart, and rybka without tables can still find wins extremely quickly in endings, even with 12-15 men still on the board. also, it would be a further evidence for that theorem, precisely because the eval functions of the engines are so different. rybka makes a choice early about which move to play then goes about proving whether it is good. fritz allows the data to make the choice for it. very different approach. perhaps the results would also be differnet becuase of that. also, rybka convincingly beats fritz overall at this time. just thoughts and ideas leading to another question i have, namely, is that result with fritz still valid with rybka today?

[guyhaw]

Would be useful if you could quote original sources on the 50-move rule. I've never heard of a 21-move rule.


Heinz assumed a worst-case, 100% negative-correlation, at one point in his statistical analysis but didn't say so - and did not check the actual correlation. I assumed no correlation (but I didn't say so either).

I also got slightly sharper results because I recognised that the result could be win/draw/loss, not just a 'manufactured' win/loss.

Heinz chose Fritz as his experiment-engine and considered carefully what the effect of this choice was. I think he tried to minimise any Fritz-specific effects.

Renze Steenhuisen has just published some results about self-play in ICGA_Journal.

[gambit3]

philidor began using 21 moves as an arbitrary guideline when playing in coffeehouses. i guess this was primarily because he could see whether his opponent was actually making progress or not in the majority of cases. this information is contained in one of his diaries, published after his death, which is a piece in the anderson collection (donated by a very active chess archivist on his death, named anderson) in the royal melbourne museum. references should be available online if you don't feel like travelling 24000 miles just to read a (french) book for one day in a controlled environment. :> in this same document, he makes reference to tournaments in which the 50 move rule should be (in his opinion) reduced, previous to his first mention of it in his coffeehouse jousts.

i'll have to check myself to see if there is an actual refernece online about the document i'm referring to. if you wish to search, keywords would most likely include chess, philidor, history, anderson collection, diary, and possibly move rule. i'll get back to you on that if i can find it online, but don't hold your breath - melbourne museum isn't like the rijndael; it may not have published all of its displays online. also, being what it is, the anderson collection if of interest worldwide for chess historians and enthusiasts, so may be somewhere other than melbourne at this point in time.

interestingly, i came across this information about 10 years ago when searching out something related more to mathematics than chess, but that isn't really relevant to this, and at the time i thought it trivial...

still going through your published material, and some of the models are quite interesting. haven't gone enough in depth to understand the approach entirely yet of most of them, however it is still fascinating. once again, thanks for the ref.

[Renze]

Small correction on the topic I published on. Although it is related to self-play, it is on Deep Search behaviour.

Renze
http://www.st.ewi.tudelft.nl/~renze/
(See Research)