H. G. Muller wrote on Wed, Oct 31, 2018 07:49 AM UTC:
I would not consider that sufficiently different from CRC to justify a new name. In general I do not consider something a new chess variant just because it starts from a different position. Chess960 itself is already a boundary case, but there it could be argued it is different from FIDE because of the extended castling rule. If this argument is accepted, I would consider Chess960 a single chess variant, differing from FIDE by its castling rule, and not 960 different Chess variants differing from each other only by their opening position. So IMO CRC is just Capablanca Chess with Fischer castling rules. That you can play it as a number of sub-variants by picking from different sets of initial positions does not make it different variants.
This is an issue that is completely independent of how to map shuffled opening positions on numbers, though. There is no logical need for the positions to be contiguously numbered. It is much more important that there exist a direct method for converting a given number N to a position, and a given position to a number N (i.e. without first having to do that conversion for all numbers < N to determine how many of those violated the setup restrictions and thus should be skipped).
If the fraction of valid positions in the numbered range is not a small minority, simply retrying generation of a start position from a new random number if the original try produced an invalid one is a perfectly acceptable solution. It seems in any case much better than that different restrictions on the initial position would lead to different numbering schemes.
I would not consider that sufficiently different from CRC to justify a new name. In general I do not consider something a new chess variant just because it starts from a different position. Chess960 itself is already a boundary case, but there it could be argued it is different from FIDE because of the extended castling rule. If this argument is accepted, I would consider Chess960 a single chess variant, differing from FIDE by its castling rule, and not 960 different Chess variants differing from each other only by their opening position. So IMO CRC is just Capablanca Chess with Fischer castling rules. That you can play it as a number of sub-variants by picking from different sets of initial positions does not make it different variants.
This is an issue that is completely independent of how to map shuffled opening positions on numbers, though. There is no logical need for the positions to be contiguously numbered. It is much more important that there exist a direct method for converting a given number N to a position, and a given position to a number N (i.e. without first having to do that conversion for all numbers < N to determine how many of those violated the setup restrictions and thus should be skipped).
If the fraction of valid positions in the numbered range is not a small minority, simply retrying generation of a start position from a new random number if the original try produced an invalid one is a perfectly acceptable solution. It seems in any case much better than that different restrictions on the initial position would lead to different numbering schemes.