lichess.org
Donate

More material puzzles?

It would be good to make more tactics for material advantage, rather than nearly every single one being a mate in X. Tactics are move practical in games, not every single game will contain a checkmate pattern at the opening / middlegame.
The issue with those is that you the "you can do better" messages are hard to get right. Often people will fail puzzles because instead of being +20.00 they will be +19.50 - which is probably why a lot of puzzles hover around the 0 like/dislike rating.
It would be very cool to be able to search puzzle by terms like "gain material, remove the defender, pinned piece"

or to chose how much material do you want

"Win 1 pawn in 4 moves"
"Win 1 knight in 5 moves" .....
I disagree. Puzzles are also about seeing what's going on. It's for us to tell if there's a mate available, or a decisive material gain. If we already know in which direction to go, it kills it, in my opinion.
but today we already know that it's a mate in X, because every puzzle we get is a mate in X.
Replies:

habatur

...within 5% = correct... within 10% = good move... I suggest evaluating the material puzzles at advantage more than 1 pawn (unless obvious) a good move but try again IF it also gains at least 50% of the material gained by the best line as 'good move but can do better...'

mindfreakkk

Yes, that would be nice.

smndvd

Exactly. But in most situations, mate is not available and many other tactics are. A mixed mate and material puzzles is preferable
What about positional puzzles?

"White to play and gain space in 4 moves"

Sure a mix is good ! But just without naming it, that was my point.
Oh ok. Of course without naming it, that would make it useless. But now I could just assume that the one I'm doing is mates! Positional puzzles....good...but need to be very careful. And how can those be automated? For example, what would classify as 'good move but there is better'? So far I have gotten only 1 material puzzle.
Problem with positional puzzles is that A) They can be very ambiguous, and most importantly B) How do you define positional advantage from a computational perspective? How do you say: "This is the start of a positional advantage problem, and this is the end of it"?

As far as I'm aware, even engines now-a-days have a lot of difficulty with positional analysis. At the end of the day, I'm going to have to write an algorithm that determines these characteristics and without a way to quantify positional advantage it gets very tricky.

For a bit of perspective (this isn't exactly efficient code), but it took me 700 lines to determine if a position was check or not, and it took equally as long to determine if a position had tension or not: github.com/clarkerubber/problem-creator/blob/master/functions/captureAndPromotion.php#L1077-L1734

This topic has been archived and can no longer be replied to.