I am not blowing my own trumpet here, my rating is rubbish and I am not claiming it was pure skill, but is this the quickest possible checkmate?? http://en.lichess.org/19T7rcHC
The quickest possible for white, black can mate half a move quicker.
The fastest mate is the same kind of idea, but the other way around.
2...g5? should be called a blunder not an error.
I agree.... what is the technical difference between a mistake and a blunder anyway?? If this is explained in another thread, what is the link for it??
Those classifications are based on the evaluation difference compared to the best move.
If your move worsens the evaluation by between 0.5 and 1.0, it is classified as an inaccuracy.
If your move worsens the evaluation by between 1.0 and 3.0, it is classified as a mistake.
If your move worsens the evaluation by more than 3.0, it is classified as a blunder.
That only works if both before and after evaluations are numeric. The logic for allowing or missing mate is different.
If you allow a mate, and you had an advantage before allowing the mate, then it is classified as a blunder.
If you allow a mate, and you were worse, but by less than 5.0, then it is classified as a mistake.
If you allow a mate, and you were already worse by more than 5.0, then it is classified as an inaccuracy.
If you miss a mate, and after missing the mate, you are less than +5.0, it is considered a blunder.
If you miss a mate, and after missing the mate, you are between +5.0 and +10.0, then it is considered a mistake.
If you miss a mate, and after missing the mate, you are still more than +10 ahead, then it is just considered an inaccuracy.
It's hard to come up with a classification that's completely satisfactory for missing or allowing mates. Personally I'd just always count those as blunders; even if you're +12 after missing mate, you still might run out of time or mouseslip or make a further mistake. Mate would just end the game :)
At any rate, that's how moves are classified at the moment.
If your move worsens the evaluation by between 0.5 and 1.0, it is classified as an inaccuracy.
If your move worsens the evaluation by between 1.0 and 3.0, it is classified as a mistake.
If your move worsens the evaluation by more than 3.0, it is classified as a blunder.
That only works if both before and after evaluations are numeric. The logic for allowing or missing mate is different.
If you allow a mate, and you had an advantage before allowing the mate, then it is classified as a blunder.
If you allow a mate, and you were worse, but by less than 5.0, then it is classified as a mistake.
If you allow a mate, and you were already worse by more than 5.0, then it is classified as an inaccuracy.
If you miss a mate, and after missing the mate, you are less than +5.0, it is considered a blunder.
If you miss a mate, and after missing the mate, you are between +5.0 and +10.0, then it is considered a mistake.
If you miss a mate, and after missing the mate, you are still more than +10 ahead, then it is just considered an inaccuracy.
It's hard to come up with a classification that's completely satisfactory for missing or allowing mates. Personally I'd just always count those as blunders; even if you're +12 after missing mate, you still might run out of time or mouseslip or make a further mistake. Mate would just end the game :)
At any rate, that's how moves are classified at the moment.
I agree that such a classification scheme is difficult!
Proof-Number Search (not implemented by Stockfish) is a purely numeric approach, although the computational power required for it greatly varies by position!
http://en.wikipedia.org/wiki/Proof-number_search
The basic concept is that each search tree node has a proof value and a disproof value:
"The proof and disproof numbers represent lower bounds on the number of nodes to be evaluated to prove (or disprove) certain nodes. By always selecting the most proving (disproving) node to expand, an efficient search is generated."
I'm of the opinion that moves which result in a simple win (are easily proven to win and easily proven impossible to lose) are good moves, even if faster wins exist.
I don't expect lichess to implement Proof-Number Search-based evaluation anytime soon, but in theory it simplifies the task of assigning (mistake) annotations to moves which increase the proof value or decrease the disproof value!
Proof-Number Search (not implemented by Stockfish) is a purely numeric approach, although the computational power required for it greatly varies by position!
http://en.wikipedia.org/wiki/Proof-number_search
The basic concept is that each search tree node has a proof value and a disproof value:
"The proof and disproof numbers represent lower bounds on the number of nodes to be evaluated to prove (or disprove) certain nodes. By always selecting the most proving (disproving) node to expand, an efficient search is generated."
I'm of the opinion that moves which result in a simple win (are easily proven to win and easily proven impossible to lose) are good moves, even if faster wins exist.
I don't expect lichess to implement Proof-Number Search-based evaluation anytime soon, but in theory it simplifies the task of assigning (mistake) annotations to moves which increase the proof value or decrease the disproof value!
<<Personally I'd just always count those as blunders; even if you're +12 after missing mate, you still might run out of time or mouseslip or make a further mistake. Mate would just end the game :)>>
@ OneOfTheQ I think your approach is right in most internet games but sometimes it is better to keep steady +10 advantage rather than trying to sac your Q to force a mate in 10 sequence. Engines are very good at evaluating this kind of tactical forcing mates with long sequences but it is not alway true for humans. Especially in faster time controls like 1 0 or 3 0 or 5 5 I would prefer to keep +10 advantage than to find such a sequence and lose on time or by miscalculating. In longer games though your approach is right and should work well in almost all OTB games.
@ OneOfTheQ I think your approach is right in most internet games but sometimes it is better to keep steady +10 advantage rather than trying to sac your Q to force a mate in 10 sequence. Engines are very good at evaluating this kind of tactical forcing mates with long sequences but it is not alway true for humans. Especially in faster time controls like 1 0 or 3 0 or 5 5 I would prefer to keep +10 advantage than to find such a sequence and lose on time or by miscalculating. In longer games though your approach is right and should work well in almost all OTB games.
Yes, I agree that it often makes more practical sense to make an inferior move than look for the optimal move.
However, the point of the analysis is not to show us best practical play for humans; engines aren't very good at that :)
The point of the analysis is to show us our objective mistakes, and staying +10 when there's a mate in 5 is certainly some sort of mistake.
If the player were strong enough to see the mate in 5 quickly, and found it, he would never choose to ignore that in favor of the +10, unless he's sadistic and wants to toy with his opponent :)
So, if a player passes up a mate in 5 for a +10 line, that means they missed the mate, which is certainly a mistake. SF is just pointing out those mistakes. It's not very well qualified to comment on the practical and psychological sides of the game :)
However, the point of the analysis is not to show us best practical play for humans; engines aren't very good at that :)
The point of the analysis is to show us our objective mistakes, and staying +10 when there's a mate in 5 is certainly some sort of mistake.
If the player were strong enough to see the mate in 5 quickly, and found it, he would never choose to ignore that in favor of the +10, unless he's sadistic and wants to toy with his opponent :)
So, if a player passes up a mate in 5 for a +10 line, that means they missed the mate, which is certainly a mistake. SF is just pointing out those mistakes. It's not very well qualified to comment on the practical and psychological sides of the game :)
This topic has been archived and can no longer be replied to.