solution of problem 41807

[uri blass]

problem 41807

I think that Qf2+ kh3 Ne3 is alternative solution.

Qxf4 is stronger but Ne3 is also winning(Rxe3 Qxe3+ Nf3 Qb3 Qxb3 Rxb3)

[uri blass]

and now another problem 36940
 
After Qh1+ Ke3 the fastest way to mate is Qe1+ but unfortunately I saw
that Qe4+ Kd2 Qe2+ Kc3 Qc2+ Kb4 Qb2+ and Qxb7 is a simple win so I did not calculate other options(note that Qb2+ is not the best but is good enough to win).

Uri

[uri blass]

problem 49840

unfortunately I found a simple win in the endgame so I did not search for mate.
I could find mate in case that I know that it is the target.

After Ra4(my move) black is completely passive(no possible plan to free the pieces) and it is clear that white is winning for example by moving with the king to capture the pawns in the queen side.

Maybe I could understand that it is not tactics but I think that winning the game is clearly enough for practical purposes.

Uri

[kennyc]

I'm not in front of a board so I might be missing something...

On the first problem, 41807, why would white respond to 2. ... Ne3 with 3. Rxe3?  How about 3. Qd3 for a supported attack on the knight, instead of just sacrificing the rook?

As for the second problem, that one is odd because your solution is mate in 6, which by Richards programming does not count (I think mate in 3 or maybe 4 would count as alts).  Here's where it gets strange, the lesser non-mating move 2. ... Rh8 counts as an alternate???

[uri blass]

for 41807 the reason is Qxg2+ and Qg4 mate(did not check it with a  computer
and usually I check problems with a computer not during solving them but before posting here to be sure that I have no error)

I did a winning mistake in 42238 and it seems that I lose significant number of rating points for missing mate in 1 even if my moves are winning

I found
1.Rxg7 Qxg7 2.Qxe8+ Qg8 but blundered by the winning move 3.Qxe5+ that is mate in 3 instead of mate in 1(did not think about mate in 3 but simply about winning endgame as usual)

I feel that the rating is not fair and I should not lose many rating points for missing mate if I do not miss the win.

It seems to me that I do not get partial points for finding the first moves but maybe I am wrong about it.

Uri

[uri blass]

another annoying problem when I missed mate in 1 only because I was happy
with winning

 problem 45536

I noticed simplifying to winning endgame by Qf3+ so I even did not analyze Qg1#

[andreacoda]

Hi there. My 2 cents are that what we have to do is to find the "best" move. If there is a mate in 3 and a mate in 1, no surprise the system expects you to find the mate in 1...

I also don't think there is any partial point system implemented: either we get the problem right, or we get it wrong - no grey area in the middle!

[richard]

Hi Uri and Andrea,

There are a couple of reasons why mate in 1 doesn't allow alternatives, the main reason is a pragmatic one from the generator's point of view. Allowing alternatives for mate in 1 often opens up a large number of alternatives, for example when there is a mate in 1 it is quite common for there to be either longer mates or material winning chances, if there are too many alternatives (more than 2) I can't keep the problem. Given that all mating sequences end in a mate in 1 this would lead to a fairly large loss of mating problems from the system.

The second reason (which is really an attempt to justify reason 1 :-) ), is that mate in 1 should be fairly obvious and that missing it probably indicates there was something about the position that you didn't understand and/or notice (in fact I would suggest that is always the case).  The counter argument (especially for blitz where time is obviously important) is that if you see a win then you shouldn't be bothered looking at the position any longer, if you are satisfied that the line you have chosen is winning then you should feel safe to play it, unfortunately the current system has a condition on that, "as long as you've checked for a mate in 1", which is a bit inconsistent with the way everything else works.   I don't think Uri (correct me if I'm wrong) is saying that he thinks his move should have been the "best move" here, but that he should have at least been given an "alternative try again" as would be the case in all other situations except mate in 1. I tend to agree that the inconsistency is unfortunate, when I have a few more problems generated I may get rid of the "no alternatives for mate in 1" rule.

Regards,
Richard.

[richard]

Kennyc,

That is a pretty accurate description of how the generator sees things at the moment.  The reason you see the non-mate as an alternative over the mate is that there are different rules for alternatives for mates and non-mates which is why the +7 is allowed but the mate in 6 is not (currently I don't check how far in the +7 becomes realised, for example it might have been further out than the mate in 6  and if so should probably have been rejected).  How long the mate has to be to be alternative depends on how long the shortest mate is.  There is a different "alternative window" for mate in 2 and above. For mate in N, longer mates must be less than or equal to mate in n+3 to be an alternative, so for mate in 2 the alts must be mate in 3,4 or 5 and for mate in 3 the mates must be mate in 4,5 or 6 (this is off the top of my head, I might be out by +1 or -1 here on the length here).

Regards,
Richard.

[uri blass]

Kennyc,

That is a pretty accurate description of how the generator sees things at the moment.  The reason you see the non-mate as an alternative over the mate is that there are different rules for alternatives for mates and non-mates which is why the +7 is allowed but the mate in 6 is not (currently I don't check how far in the +7 becomes realised, for example it might have been further out than the mate in 6  and if so should probably have been rejected).  How long the mate has to be to be alternative depends on how long the shortest mate is.  There is a different "alternative window" for mate in 2 and above. For mate in N, longer mates must be less than or equal to mate in n+3 to be an alternative, so for mate in 2 the alts must be mate in 3,4 or 5 and for mate in 3 the mates must be mate in 4,5 or 6 (this is off the top of my head, I might be out by +1 or -1 here on the length here).

Regards,
Richard.

It is not logical to reject longer mates as wrong solutions if you accept winning
moves when the computer does not see the mate.

Practically I played the move because it wins material and not because of seeing a long mate and winning a lot of material can also be translated to long mate if you search deep enough.

I also think that there should be different rating for different failures.
People who blunder by move that lose the game should not get  the same rating as people who blunder by a drawing move.

Uri