Puzzles Update

James 'Clarkey' ClarkeAnnouncements

You've waited patiently and now it's here

Updating the tactics trainer has been under development for some time now and we're finally ready to reveal what's new. It was teased in the last developer update that change was in air starting with 40k new puzzles, but under the hood there's been some fundamental changes aswell.

So what's new? Let's find out.

Spaced Repetition

With the previous system pass-or-fail you would never play the same puzzle again. That makes it a little difficult to learn from your mistakes don't you think? So what's the solution? Thankfully there's a rather simple learning tool referred to as spaced repetition to help out. With this new system you will re-attempt puzzles that you've once failed in increasing intervals until you no longer fail the tactic. Specifically the system that we are using is a slight modification to the Leitner System, using 3 boxes with Box 1 being filled with new puzzles.

The Puzzles

So what's new with the actual puzzles? The short answer: there's now guaranteed to be only one good and acceptable solution to each puzzle. Gone are the days of 'good move but you could do better' or being failed for playing a mate-in-3 when there was mate-in-2.

The games the puzzle positions are sourced from are now strictly from players above 2000 rating on average. Previously the positions were sourced from any game played by any players, the issue with this is that frequently the puzzles would start from positions that originate from highly unsound play from both sides. These puzzles are less interesting because they are unlikely to appear in normal play, and usually the solutions aren't very sharp. But sourcing the positions from high level games means that the solutions are generally much sharper and instructive.

Puzzles also start from more balanced positions in terms of evaluation and material on the board, whereas previously the starting position could already be highly advantageous to the player.

Puzzle Selection

Beyond spaced repetition we'e also made some slight tweaks to how puzzles are selected.

  • The 10k puzzle limit has been abolished. There is now absolutely no limit to the amount of puzzles you can attempt.
  • Selection bias due to user votes is no longer a factor.

Looking Forwards

Previously we ceased puzzle generation at 60k puzzles as we regarded this as enough puzzles for anyone. But this time around we're going to continue to generate puzzles indefinitely, keeping the database supplied with fresh puzzles, and we'll explain why in a moment.

You might notice that the difficulty selector has been removed, this is for a couple of reasons. Firstly, it is not compatible with spaced repetition, but primarily this is because we plan to implement custom puzzle sets and filters. But for this to be effective we need a lot more puzzles to be able to satisfy your various training regimes. So stay tuned for that.

That's all folks! We hope you appreciate the improvements and happy puzzling!