lichess.org
Donate

LiChess Training improvement ideas

31,

No it is not. It is very small.

Take 1500 puzzles at a time, that would be 1500/150 = 10 links. 10 links is a piece of cake.

60000/1500 = 40

It is only 40 times that you repeat.
I think we'd need to trim the tactical motifs a bit. You don't want to give too much away with the motif. For example, mate in x would satisfy mate in 1 - mate in x, without giving too much away (for example, someone specifically looking for a mate in 3).
The reason for specific motif is that, a player can later be profiled on number of mate in 2 solved, number of mate in 5 solved and so on.

But of course, the user is only given a slight hint. "White to play and mate".

I am also thinking of puzzles for kids, for starters, "White to play mate in 1" would drive them to solve more.

I also thought to run an engine on all puzzles searching for those positions where there is mate say mate in 10 or less. Then those puzzles will not be tagged manually which will help a lot in saving time for manual tagging.
I wish lichess added the Chess Mentor feature. That would be Amazing.
I stumbled upon this thread and read some of the comments, and it seems that some of you are planning to commit themselves to the monumental task of tagging tens of thousands of puzzles. To add in on this.

As I understand it the puzzles come from game analysis that people request of their played games (http://en.lichess.org/blog/U4sjakQAAEAAhH9d/how-training-puzzles-are-generated note that I didn't read this). This means there is a database of analysis available.

I do agree that it might be useful for some people to have tagged puzzles, but the easiest way to obtain them seems to me to let the puzzle-generator add them automatically. For example:
-if the best moves lead to a mate by knight where the king's neighbouring squares are all occupied: smothered mate
-if they lead to a mate by queen or rook of the king on the last row, with pawns on the previous last one: back-rank checkmate
-if they lead to material gain after a check: fork

Obviously the different tags you have in mind need more tightly-defined conditions than the ones above; these are just to give an general idea. I haven't given it much thought yet, but I believe you can generate a useful set of tags in very little time—except for the code writing part that is :-)

This topic has been archived and can no longer be replied to.