syzygy wrote:
Crafty, or any other engine, is pretty useless for analysing a long TB endgame.
I guess that depends on what information you're trying to obtain from the analysis.
Example, if you want to know how many of the moves by the losing side result in short mates it's irrelevant if the engine sees to the end of the mate. The only thing that matters is that it be able to determine the moves that result in relatively quick mates.
i.e. I want to know the number of moves in each position in which the engine can actually find a winning line. These occur principally because the preceding move by the opponent was a mistake. There are many uses for this information.
One of them is that it determines how difficult a position it is. If a large fraction of the opponents moves result in mate or other bad things happening within the search horizon then the problem isn't that hard. If however most of them lead to long lines with no obvious loss of material or positional degradation then the line is "hard".
If you want to know something about proof tree sizes its nice to have long mates from which you can extract information. Long mates are not commonly found in over the board games so using data from the table bases is an expedient method to find some. If all but one or two moves result in an immediate mate then the tree size is likely to be very small compared to what it's length an branching factor would predict.
Either of these are sufficient reasons to study these types of positions. Enough said.
I'll look at the link when I get a spare minute.
syzygy wrote:Of course such endgames have little to do with FIDE chess.
Maybe, ... maybe not. Regardless of either case they are worthy of study.
Regards,
Zen
Only 2 defining forces have ever offered to die for you.....Jesus Christ and the American Soldier. One died for your soul, the other for your freedom.