"Learning" Puzzles Seem Like Beta Testing

Sort:
MightyKingFork

I have recently been using the "Learning" mode which allows you to select the puzzle rating difficulty, puzzle type, and the puzzles are unrated without a timer. I am selecting 1-2 categories a day to focus on some basic tactics I need to work on. I am finding, however, that many of the puzzles are not even close to the category selected. Commonly, these puzzles have 8-10 seemingly generic categories selected - many without even the basic possible characteristics (ex: "bishop pair" when there aren't bishops on the board; stalemate when it is a clear 'mate in 1' - which ironically isn't an option). I am doing my best to up/down vote the themes to get them appropriately assigned. However, it is starting to take up a lot of my time. It is most frustrating to be working on 'stalemate' tactics only to have many be early game back rank mate in 1 tactics. It is almost like some of these puzzles are new to the database and haven't had any human review. Below is a recent example of MANY that I come across. It makes me want to find a different database with similar features (sorting by tactic/puzzle type). Anyone else notice this?

Jet_Unholy

I have had that happen too. I think "Learning Mode" is a great tool, but definitely some of them are miscategorized.

KingsBishop

yeah same thing.. there are several categories that are just not satisfactory. Like underpromotion the following is actually a remove the defender, and the fact that an underpromotion takes place is irrelevant in fact there was no reason to underpromote. The only reason the computer underpromotes is because it evaluated losing a queen to be worse than losing a bishop.

 

Martin_Stahl
Jet_Unholy wrote:

I have had that happen too. I think "Learning Mode" is a great tool, but definitely some of them are miscategorized.

 

The site didn't used to have categories. At one point they added the ability to suggest themes for puzzles and when you crowd source things like that, you get mixed results. If a theme has enough votes for a theme, it got flagged with it.

 

Some people suggested themes that were in the position but not really pertinent to the position/tactic. The learning system was built on that backbone.

 

People can still up-vote/down-vote themes, but I imagine not enough people do to fix it and there are still people that choose or up-vote incorrect themes.