I don't know if this is of interest to the developers, and I also don't know if anything can be done on the lichess end (vs. stockfish), but the opening indicated in the move list does not seem to account for transpositions.
For example, this game quickly transposed into a Benko gambit:
http://en.lichess.org/KxPgj6nW/black but the opening is indicated as some English opening
Certainly not a big deal at all. I was just pointing it out in case 1) the problem is of interest, and 2) it had not been mentioned yet
I have also found this annoying, and it also doesn't account for systems that can be reached through many move orders (King's Indian Attack and friends.) I can't comment on the technical aspect, but it would be nice to see the system recognize this stuff.
I think it would be awesome if the analysis wouldn't be so quick to label an opening. I would love if they somehow made the information given by the computer more helpful in improving one's opening play.
The problem with this is that the engine determines the opening. In order for transpositions to be recognized, it would have to have a greater set of opening books to then check against regularly throughout the opening to verify a transposition has not occurred. This is virtually impossible with the limited resources available to the servers used for analysis. This was already addressed maybe 2 weeks ago in another thread.
you talk about opening books, but as I see a 2 and a 3 MB chessbase opening KEY on my hd, I think that might be another way to go.
static shadow : so the engine could do it, but we don't have the logistics to implement it is what you're saying?
That's what I'm saying elan-the-bard.
Yersinia, it's the same thing. The engine would have to check every move against that key. It isn't like having it running on your computer. In order to get analysis done in a timely manner, each move is analyzed by a different computer across tons of machines that are donating crunching power. In order to check each move against the key, it would significantly slow down that process.
a simple FEN check against a Stringlist slows down a position evaluation significant? I can not believe this.
When it's being handled on a single machine? No. When it's being handled over multiple machines and then being recompiled? Yes.
That being said, I think there is something in the works on the devs' to-do list to examine if this is something that can be done. You'd have to ask Clarkey or Thibault or something though for a real reason. I'm just basing my response on the other threads that have addressed this in the past.
