Problem Set Updated

[richard]

I've just updated the problem set based on changes in the automatic problem generator.  Most of the changes have been the removal of many problems due to much stricter ambiguity checking. This will unfortunately have killed off some interesting problems , but overall the problem set should now have much fewer, "Hey my move was still pretty good but got marked wrong." positions.  I've also removed positions that started with overwhelming material advantage to the tactician.  This should reduce the feeling created by some positions of, "Well any non-blundering move here should win". I've deliberately left the definition of "overwhelming material advantage" at a reasonably large 5.5 pawn advantage so as not to throw out too many interesting positions, hence only massively crushing positions have been discarded. 

These changes don't apply to the positions which end in mate.  There may still be very long mates that are considered "best move" even if there was a shorter non-mating sequence that led to a clear advantage. In other words, mates are always preferred over non-mates and shortest mates are still always preferred over longer mates, hence longer mates will be marked wrong if a shorter one was available.

At least one area that requires further thought, is the move pruning algorithm employed after a tactical position is found.  Sometimes the sequence of moves the user is required to get right is too short to show the main point of the tactic.  The pruning is deliberately conservative to avoid tactics dragging out beyond the main tactical point and therefore requiring to the user to choose moves after the tactic is "over".  Over zealous pruning can leave the user wondering, "Well I don't really see where this tactic was going" even if they did select the correct first move(s).  In order to reduce the annoyance of this situation I will be adding the ability to see the full unpruned  line the computer picked for the tactic after the problem has been completed (or by viewing it in the problem viewer). It may be a week or two before I get time to add this feature however.

Anyway, I'm sure there will still be further rounds of refining to do, on the problem generator. Please continue to post problems you notice issues with to the forum to help me further improve the generator.  Once the generator has settled down I'll work on adding new problems to the system as quickly as possible.

Regards,
Richard.