2

On lichess.org, how are the ratings for the tactical puzzles computed? I am trying to understand how some puzzles I find challenging are as low as the 1400s, while others that seem obvious are at 2100+. As an example, consider this puzzle where the mate seems obvious (the white king has nowhere to go), so I would expect a rating in the 1200s, but the actual rating on the site (at the time of writing) is 2102.

4r3/8/p2p4/6k1/P7/1p2N2R/1P3r1R/K7 b - - 0 1

(Possibly a duplicate of "How are online tactic exercises ratings calculated?"? But I ask about the lichess.org puzzles specifically because the ratings there are so often surprising, especially for puzzles like the one above that have been played more than 30000 times.)

Rewan Demontay
  • 16,942
  • 4
  • 65
  • 109
user23739
  • 21
  • 2
  • @NoseKnowsAll If someone can confirm that that is how lichess.org computes the ratings. (See the paragraph at the end of my question.) – user23739 Jun 02 '20 at 17:32
  • That other post directly answers this question. In particular here, move order is sometimes overlooked; also some people might not be calculating that black is already down a piece at the beginning of the puzzle, so you need more than just winning the knight with Rxe3. – NoseKnowsAll Jun 02 '20 at 17:34
  • You can confirm then that lichess.org uses that system? My question here is roughly "I know how puzzle ratings are typically computed, but the evidence available to me suggests that lichess.org might be doing something different; are they in fact doing something different, or is my assessment of problem difficulty really that far off from the data?" – user23739 Jun 02 '20 at 17:41

0 Answers0