lichess.org
Donate

Training Unforgiving

Many of the tactics puzzles punish you for even slightly suboptimal, yet still trivially winning moves.
http://en.lichess.org/training/24964
It is clear to me that the final move of capturing with the queen forces an easy rook up endgame (counted as a fail), and not so clear why the knight capture is "better," or even that it is at all. I think most humans would favor trading down to a very clean, trivial win. This is only the most recent example among many and it is putting me off your tactics trainer.
While the computer is not perfect it still shows you the best combinations you need to play in order to secure the win. While moves like the one you mentioned may not be the "best" the practice will help you as long as you're fine with losing your puzzle points
Right, I don't really mind losing the points. Getting a high score isn't the point of tactics training, after all ;)
However, the effect on ratings of problems is more meaningful. I suspect that quite a few of the higher rated problems are rated highly due to strong human moves being fails and the computer's best move being somewhat inhuman, leading to lower average quality problems as one gets stronger. I believe the problem I mentioned to be a good example. http://en.lichess.org/training/21940 is another one.

I think of this type of training as a feedback system. You should get penalized for bad moves and rewarded for good moves. Isn't getting penalized for very strong moves counter-effective? When a move like Qxe6+ fails I feel as if I didn't understand the problem and need to check with an engine only to discover I had not failed in my analysis or missed anything. You might argue that I did in fact miss the stronger final move, but in actuality I made the practical decision to neglect analyzing it at all in favor of a clear, trivial win. You may argue that practical decisions don't apply to a tactics trainer where we have as much time as we want to come up with the best move, but isn't the point of a tactics trainer to prepare us for the real thing? Qxe6 is usually going to be the most practical move in a game with 100% winning chances, at least for a player like me. A move like this should at least give neutral feedback like "good move, try again," as I have seen some problems give.
#1 #3 ZugAddict, myself, and you all agree on this.

I would love for lichess Training to be even better than Chess Tempo, and in my spare time I'm experimenting with many possible ideas (such as better puzzle generation/grading) although it's much easier said than done.

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