Problem 21992

[oblonsky]

Hi

Just as I thought I was getting into the swing of tactics I come across this beauty.  In my naivete I thought I would play to win the queen by skewer, but no I missed the obvious mate: a straightforward mate in 11.  How I missed it I will never know.

regards
Bernard

[drahacikfm]

Soon you won't have a chance to be naive any more!  In the revised problem set coming soon, according to Richard, problems where you can win big material (such as a queen) instead of doing a mate, will be deleted.

[richard]

Hi

Just as I thought I was getting into the swing of tactics I come across this beauty.  In my naivete I thought I would play to win the queen by skewer, but no I missed the obvious mate: a straightforward mate in 11.  How I missed it I will never know.

regards
Bernard

:-)  I think you also missed the mate in 25 and 30 that was hiding behind the mate in 11 :-)

I just checked and the new problem generator did automatically remove this problem.  The new generator does remove a lot of these but it is still not perfect.  I've been reviewing the new set prior to upload and some material winning versus mate positions will still slip through. This is due to an oversight on my part. Namely when there is multiple mates at different lengths say a mate in 4 and a mate in 8, I allow that position as the mate in 8 is so much longer than the mate in 4, however mates are not checked for material winning lines so if the mate in 8 was also a material winning move it bypasses the material winning check. This doesn't impact many problems but is an area I'll look at improving in the future.   Also the material winning ambiguity check only kicks in when the mate in more than 2 moves away, so if there is a choice between a mate in 1 or 2 and a queen take, the user will still have to find the mate.  This is done by design, but if users feel this is still annoying, I can think about tightening this requirement to always reject these positions irrespective of mate length, but that will have to wait for a subsequent problem set run.

Regards,
Richard.

[slacker00]

That's a tough problem!  I couldn't weave my way through to find the mate  in 11. ;(

[drahacikfm]

Richard, I think this is a very important issue.  To make sure I understand what you wrote:  If a problem has a move that leads to mate-in-4, and another move that wins the queen outright, so that you are a whole queen up, and that move also leads to a mate-in-8, then this problem will still be in the problem set, and the queen-win will be marked wrong?  But if the queen-win does not lead to mate, the problem will be deleted?

I think that's a serious deficiency.  So we will still be in the same situation we are in now:  If we see a very simple win of a queen, we don't know if this move is correct, because maybe another move leads to mate.

The only difference now will be that if the queen-win DOES NOT lead to mate, then we know it is correct!  But if the queen-win DOES lead to mate, we know that maybe it is NOT correct!  In other words, non-mating queen-wins are more likely to be correct than mating queen-wins.  That's not a situation you want to be in when solving problems.  To make it worse, all queen wins lead to mate eventually, and it becomes a question of whether it's within the search depth of your program that generates the problems.

I think it is worth delaying the release of a new problem set until this issue is resolved.  Basically, if you play a completely winning move, it should not be marked wrong.  The only exception should be is if there is a mate-in-1 instead.

I don't mean to sound too harsh.  The site is great, and my son and I use it every day.  It will be almost perfect when winning moves are never marked wrong! 

[richard]

Hi Drahacik,

I agree and I'd have preferred to resolve this problem before problem set release, however overall the new problem set has a large number of the annoying mates removed.  It takes several  weeks to run a full problem set verification run so for now I've uploaded the current set and will look at resolving the issue discussed below in the next verification run. So releasing the new problem set should still improve overall quality and gives me a chance to get feedback on how people feel about the large reduction of mates from the set before I start another round of problem generator improvements.

Regards,
Richard.