Problem #40956

[glennwilson]

http://chesstempo.com/chess-problems/40956


Per my computer analysis (it's not just me) move 2 should definitely accept 2...Qxh3 and 2...Rxh3 and also probably 2...Ne4 and 2....Ne4.

It only accepts 2...Qxh3.

[drahacikfm]

It's funny that you wrote 2...Ne4 and 2...Ne4, repeating the same move twice, because the computer analysis display for this problem shows the best move is 2...Qxh3 and the second best move is 2...Qxh3    Must be some kind of bug that repeated 2...Qxh3 instead of putting 2...Rxh3 for the second move.

Also, the top 4 moves are all over +3.0, so this problem should have been thrown out by the generator.

I'm sure Richard can explain all of this!

[richard]

Yes, this is an unfortunate bug in the generator, I've mentioned it in a couple of comment replies over the last few days.  Toga is occasionally outputting the same move twice (I believe one is probably a 'provisional' move that it is still thinking about - I think the full lines differ in both moves when this occurs but I haven't checked this problem yet), this completely confuses the generator as it wasn't a behaviour I was expecting and it causes alternative processing to falter.  Now that I have all the evals available on the client I should be able to fix this without a generator re-run simply by dynamically looking at the evals to determine alternatives rather than just using the alternatives that were tagged by the generator at generation time. I'll hopefully have a fix for this available in the next couple of days.

Regards,
Richard.