H. G. Muller wrote on Sun, Dec 17, 2017 11:45 AM UTC:
This is a very good question, and is already on my to-do list for many years. (Alas, never high enough.) It is in fact why I started the 'Pair-o-Max' fork of Fairy-Max; this gives a configured bonus on the victim value for indicated piece types on every capture that leaves an odd number of them. (And it can also be made aware of drawishness due to lack of mating potential.) I never found the time to lauch a systematic investigation on this issue, though,
One thing I did determine, though (although not with great precision): pairs should not be counted combinatorially. If you have 4 Bishops, distributed 2+2 over the square shades, they are worth as much as two pairs. They do not count as 2x2 = 4 pairs. With 2+1 Bishops you should count only 1 pair.
I am pretty sure that in the case of multiple color-bound piece types there must be 'cross bonuses', e.g. that F+B on the same color should be worse than F+B on different color. In end-games of Team-Mate Chess, which features two color-bound types (FA and AG), in 3+2-men end-games of FA+AG vs FA or AG it mattered whether the stronger piece (FA) was on the same shade or opposite shade as the opponent piece for the end-game to be a general win or a general draw. That raises the interesting question how to score 2B+F. So a lot of testing would have to be done. A good start would be to determine the pair bonus if individual isolated color-bound piece types. By playing pairs of F, FA, FD, FAD, BD, BDD on different shade against those on the same shade, in the presence of only pieces that can access the entire board (and Pawns).
This is a very good question, and is already on my to-do list for many years. (Alas, never high enough.) It is in fact why I started the 'Pair-o-Max' fork of Fairy-Max; this gives a configured bonus on the victim value for indicated piece types on every capture that leaves an odd number of them. (And it can also be made aware of drawishness due to lack of mating potential.) I never found the time to lauch a systematic investigation on this issue, though,
One thing I did determine, though (although not with great precision): pairs should not be counted combinatorially. If you have 4 Bishops, distributed 2+2 over the square shades, they are worth as much as two pairs. They do not count as 2x2 = 4 pairs. With 2+1 Bishops you should count only 1 pair.
I am pretty sure that in the case of multiple color-bound piece types there must be 'cross bonuses', e.g. that F+B on the same color should be worse than F+B on different color. In end-games of Team-Mate Chess, which features two color-bound types (FA and AG), in 3+2-men end-games of FA+AG vs FA or AG it mattered whether the stronger piece (FA) was on the same shade or opposite shade as the opponent piece for the end-game to be a general win or a general draw. That raises the interesting question how to score 2B+F. So a lot of testing would have to be done. A good start would be to determine the pair bonus if individual isolated color-bound piece types. By playing pairs of F, FA, FD, FAD, BD, BDD on different shade against those on the same shade, in the presence of only pieces that can access the entire board (and Pawns).