There are many games played on lichess which do not appear in the database. Also some games may appear if one searches for the games of a specific player but not if one does a general search. This can seriously skew the statistics for the results of certain moves. Does anyone know if this issue been addressed or whether there is an intention to rectify it ?
@HighlyCruciferous said in #1:
> There are many games played on lichess which do not appear in the database. Also some games may appear if one searches for the games of a specific player but not if one does a general search. This can seriously skew the statistics for the results of certain moves. Does anyone know if this issue been addressed or whether there is an intention to rectify it ?
Usually Lichess site has a huge traffic of people . People play many games which is counted in billion's because it includes both Anonymous users as well as Register'd user's too . Maybe because of this you're finding error in Lichess database .
> There are many games played on lichess which do not appear in the database. Also some games may appear if one searches for the games of a specific player but not if one does a general search. This can seriously skew the statistics for the results of certain moves. Does anyone know if this issue been addressed or whether there is an intention to rectify it ?
Usually Lichess site has a huge traffic of people . People play many games which is counted in billion's because it includes both Anonymous users as well as Register'd user's too . Maybe because of this you're finding error in Lichess database .
From what I understand from github.com/lichess-org/lila-openingexplorer, opening explorer no longer samples games but rather includes every single game using the dumps from database.lichess.org/
The upside is that it is now as accurate as can be since all the games are indexed.
The downside is that it is only as accurate as the last available monthly dump (so you'll have to wait for April for March's games).
The personal explorer seems to still be indexing on the fly though.
The upside is that it is now as accurate as can be since all the games are indexed.
The downside is that it is only as accurate as the last available monthly dump (so you'll have to wait for April for March's games).
The personal explorer seems to still be indexing on the fly though.
No, it's not just a matter of waiting for the games to be indexed. Some games played well over a month ago are simply not included in the database/search engine, although you can still find them if you make a more specific search (players' handles, date, result etc)
i believe you'd have a much higher chance of a dev looking into it, if you gave an example they could actually look at.
Right you are. I'll try and find some of the games to which I'm referring.
I have noticed this too and got confused initially. Sometimes, you get to a position that was played X times in total. Then you make a move that was played Y times (obviously Y cannot be greater than X). Then you get to a new position where you find there are a total of Z games that were played. What's confusing is that sometimes you notice that Z is greater than Y. Sometimes Z is even greater than X!
This makes no sense at first glance and seems like a glitch. Because it doesn't matter whether Lichess indexes games from a sample or the whole database. Logically speaking, this should not be happening.
But I came to realize later why this was happening. You see, whenever you look up a specific game or position in the Lichess database, it considers the exact position and not the move order. What this means is that some positions can occur in multiple different ways via transpositions from other positions and they will still be regarded as the same. This also explains why sometimes if you follow a specific opening, it can get transposed to another after some moves. For example, in 1. f4 d5 2. e4 e5, the Bird opening, Williams Gambit gets transposed to King's Gambit, Falkbeer Countergambit. Up until you play the move e5, the position in Williams Gambit got played around 2 million times in Lichess. However, once you play the move e5, then you can see the position getting played around 5.5 million times.
I hope that clears it up for everyone who got confused because of this. It is definitely not a programming bug. Although not very intuitive, so I am wondering whether there could be a note or explanation or a visual cue somewhere that could help users understand how it works.
This makes no sense at first glance and seems like a glitch. Because it doesn't matter whether Lichess indexes games from a sample or the whole database. Logically speaking, this should not be happening.
But I came to realize later why this was happening. You see, whenever you look up a specific game or position in the Lichess database, it considers the exact position and not the move order. What this means is that some positions can occur in multiple different ways via transpositions from other positions and they will still be regarded as the same. This also explains why sometimes if you follow a specific opening, it can get transposed to another after some moves. For example, in 1. f4 d5 2. e4 e5, the Bird opening, Williams Gambit gets transposed to King's Gambit, Falkbeer Countergambit. Up until you play the move e5, the position in Williams Gambit got played around 2 million times in Lichess. However, once you play the move e5, then you can see the position getting played around 5.5 million times.
I hope that clears it up for everyone who got confused because of this. It is definitely not a programming bug. Although not very intuitive, so I am wondering whether there could be a note or explanation or a visual cue somewhere that could help users understand how it works.
I believe that is a seperate issue. Transpositions do not account for the fact that there are games which do not appear at all in the database. I'll try to come up with some examples soon
This topic has been archived and can no longer be replied to.