Bug in calculation of Puzzle Popularity? #12408
Description
I'm looking through the puzzle database (downloaded from https://database.lichess.org/#puzzles on 15 Feb. 2023) and I've come across 8 puzzles that have a "popularity" greater than 100. According to the documentation, this shouldn't happen.
Here is a list of the puzzles:
| PuzzleId | FEN | Moves | Rating | RatingDeviation | Popularity | NbPlays | Themes | GameUrl | OpeningFamily | OpeningVariation |
|---|---|---|---|---|---|---|---|---|---|---|
| P9tbW | r1bq1r2/pppp1pkp/2n3pN/2b1P3/P7/6Q1/1PP2PPP/RNB2RK1 b - - 3 12 | d7d6 c1g5 f7f6 e5f6 | 1934 | 89 | 104 | 42 | crushing opening short | https://lichess.org/Wot0bma8/black#24 | Ruy_Lopez | Ruy_Lopez_Open_Berlin_Defense |
| ZTa65 | 7r/pp1kn1Rp/3p4/4p3/1n2P3/2NPP1P1/PPP4P/6K1 b - - 0 17 | b4c2 c3d5 h8e8 d5f6 d7d8 f6e8 | 1446 | 75 | 103 | 1203 | crushing endgame fork long | https://lichess.org/4R1f08Mm/black#34 | Italian_Game | Italian_Game_Giuoco_Pianissimo |
| nlayx | r2q1rk1/p3npbp/2p3p1/1p2Nb2/3Pp3/4P2P/PP2BPPB/R2Q1RK1 b - - 1 15 | f7f6 d1b3 g8h8 e5f7 f8f7 b3f7 | 1429 | 86 | 105 | 46 | crushing fork long middlegame | https://lichess.org/jj2XO4xp/black#30 | Modern_Defense | Modern_Defense_Other_variations |
| ptOrF | 3r4/1p6/p1n1b3/2P1k1N1/1P1pp1P1/P3P1K1/B7/5R2 b - - 0 46 | e6a2 f1f5 | 1445 | 160 | 106 | 57 | endgame mate mateIn1 oneMove | https://lichess.org/EpN14kb3/black#92 | nan | nan |
| rTsjo | 2rqk2r/3bbpp1/p2ppn1p/8/1n1NPPPP/2N1B3/1PPQB3/2KR3R w k - 0 15 | g4g5 c8c3 g5f6 d8a5 d2c3 b4a2 c1d2 a2c3 | 2355 | 78 | 103 | 83 | middlegame | https://lichess.org/yedDQT52#29 | Sicilian_Defense | Sicilian_Defense_Scheveningen_Variation |
| uo1zm | r3kb1r/ppqn1pp1/2p1p3/7p/3P2nB/P1N2B1P/1PP2PP1/R2Q1RK1 w kq - 1 14 | c3e4 c7h2 | 808 | 83 | 129 | 101 | mate mateIn1 oneMove opening | https://lichess.org/phxVvBqe#27 | Scandinavian_Defense | Scandinavian_Defense_Classical_Variation |
| xB5aH | 2k5/ppp2ppp/8/8/1PKB4/3R2P1/q1P3QP/3r4 w - - 0 24 | d3b3 a2a6 c4c3 a6a1 | 1645 | 202 | 200 | 7 | crushing endgame short | https://lichess.org/68NIJbqI#47 | nan | nan |
| ynV70 | 8/p4k2/2p5/5q2/4R3/2Q2Prr/PP3K2/3R4 w - - 2 31 | d1d7 f5d7 e4f4 f7g8 c3c4 g8g7 | 2365 | 130 | 106 | 23 | crushing defensiveMove endgame hangingPiece long | https://lichess.org/snTg0Omv#61 | nan | nan |
Also, the documentation on the website seems to be a little off. It says that
"Popularity is a number between 100 (best) and -100 (worst).
It is calculated as (upvotes - downvotes)/(upvotes + downvotes), although ..."
However, I think there is a missing factor of 100 in that formula (otherwise popularity would be between 1 and -1)
It is unclear how this affects operation, if at all, because I do not know what "popularity" is used for.