lichess.org

Two possible solutions for a tactics quiz

it's going to take a while for me to find such a puzzle as i said it's rare
i'll try to be somewhat more clear on what i mean

as i said earlier objectively the engine is right 100% of the time or so the thing is on rare occasion a +4 engine evaluation is from a human point of view "not so good" as a +3 human solution.
getting a + 4 position remaining highly complex you can't convert is hardly interesting ( well it is but to win the game is another story) sometimes you get a +1 endgame you know for sure it's won. remain practical it's what it is all about.

I found an example with 2 good lines.
lichess.org/training/61918

I played Bxg5 and failed. Evaluation +1.3. This move is even considered as a blunder because of the much better move d4 (+5).

for me this isn't what i was thinking about
in this problem d4 is the move because you're up a piece at least or have a mating attack
while Bxg5 give you some attack but you're down a piece and it isn't that clear about where you're going
even if the computer tells you white is better i don't find this to be obvious.

#13, exactly.

1.Bg5 is no „solution“, e.g. a crystal-clear win. It‘s a number around 1-1.5, What’s that?

Better example than my made up case.

If you open the game, you can see that the evaluation scores were logged as only +0.6 for Bxg5 and +2.8 for d4 during the quick server analysis. Only if you let the engine think a bit longer these will go towards +1.5/+4.

0.6 vs. 2.8 would more obviously be "throwing away your win" (however, even from a human perspective, the position after Bxg5 is "unclear", not "winning").


There exist some rare cases where the engine actually dismissed an equally good solution because it couldn't find it quick enough, but these are really rare and it's still a much better bet that a human that complains about some puzzle is the one who is wrong.

"getting a + 4 position remaining highly complex you can't convert is hardly interesting ( well it is but to win the game is another story) sometimes you get a +1 endgame you know for sure it's won. remain practical it's what it is all about."

Is one of the points of puzzles not to learn to identify and use new tactics, or to learn to analyze complex positions?

Your argument amounts to: Yeah, I identified a much worse move than the best one here, but it's because I can find the best move on all subsequent turns. But this is a naive assumption, if you are at +4 evaluation and the best move you can come up with is one that leaves you at +1, then I think your chances of making mistakes in what you say you know "for sure" you can win are also pretty high.

If the point of them is to improve in some way, then simply reinforcing your ability to not make the best move is not very useful.

Yup, an endgame evaluation of +1 means that you won't queen a pawn or otherwise win more material during the next 20 or so moves, so chances are not bad that winning is not as trivial as you'd think.


Converting into a trivially won endgame by sacrifice (like giving your queen for a bishop to convert into in a technically won pawn endgame) will usually clock in at +5 upwards, too.

@Chuck_Fess
"Is one of the points of puzzles not to learn to identify and use new tactics, or to learn to analyze complex positions?"

It is both from my point of view.
Most puzzles can be solved by simply identifying the correct tactical motifs and then comig up with the correct move order.

But there are also a lot of puzzles where you must be able to analyze the position which you get in the end. These positions feel more like real games. I had a lot of puzzles where winning the exchange was not enough. You had to see that your opponent had a lot of compensation. There are also positions where winning the queen for a rook and a piece was not enough. In most cases it is, but not in all.

I suggest that all puzzles should play to checkmate. If you can defeat Stockfish, that's good enough.


as a puzzle it's defnitely interesting i agree. but adopting such patterns into your own games can be missleading because of what i said earlier.
i don't complain about the puzzle system implemented here just saying that from a practical point of view on rare occasions the top engine move isn't to be prefered in an otb game that's all.

Reconnecting