Play against SF from all your repertoire openings multiple times and
analyze them to where you went wrong and what is the best move there.
Seems useful.What do you think ?
If you want to play the openings like a computer. However, this isn't good training after the opening, because you need to know how to punish you opponent's errors.
I couldn't punish SF8 ' s blunder on move 6.
Yes that is useful.
Really ? (Wow @tpr finally liked my training methods )
Of course playing and losing against a stronger opponent human or engine and then analysing is beneficial.
Not sure if practicing openings against a computer is such a good idea. Computers have difficulty finding plans and developing schemes in the opening. So you'll mostly learn concrete moves and get zero understanding for the position. You can better play games against random opponents and then check the opening with a database. The engine is of course a useful tool, but I think a database is more important in the opening.
#2 For the last 10+ years, TCEC and other engine-engine contests have used predefined start positions; engines don't play openings well.
@Toadofsky
That is not why TCEC uses its own opening book for the tournament (and they have had stages without books at all in the past).
They do it for 3 primary reasons:
1) Variety. It gets pretty stale pretty quickly watching engines play more or less the same thing every time.
2) Reducing draws in the later stages.
The top engines on TCEC hardware at TCEC time controls are incredibly strong, so from very balanced positions will draw virtually every game. Cato and others involved in this have tried to pick openings that are unbalanced enough that it's near the edge between win/draw, so that because of the small drawing margin engines have the opportunity to push the advantage and win with the superior side and defend precisely and draw with the inferior side (past the first stages it's always at least a double round-robin, so each matchup plays each side of the same opening for something approaching fairness).
3) Theoretical interest.
This one is a newer consideration, but some openings are chosen precisely because they are topical and people are interested in seeing how top engines on such strong hardware with so much time play them out.
On that point of #3 being a newer consideration, how and why the openings have been picked change constantly. The audience constantly complains about something. "The openings led to too many draws", "The openings are too unbalanced now", "No one plays this opening at top levels", etc.
How/why the openings are picked changes based on that quite a bit.
I certainly would not say that SF, K, and H on any decent hardware at tournament time controls play the openings poorly or "not well" any more. Even at bullet/blitz time controls on reasonable hardware I wouldn't say they generally play the opening poorly any more; they'll definitely play suboptimally more often under those conditions, but I'm not at all sure that even then you could generally say they don't play it well.
You might find some positions where they don't play the same thing that humans have gleaned from >100 years of analysis (with much of the recent work obviously computer assisted, even), and even in those cases they're not even usually wrong; they just prefer a different move.
Saying that since engines in some small number of cases can't find in a couple minutes what humans found over the course of several decades they don't do it well is a smidge unfair :)
Now, running SF on your phone for a couple seconds is likely to give some very suboptimal advice in many positions, but that's not unique to the opening :)
Even SF on my very "meh" laptop plays the opening rather well even in blitz, and much more so if we're talking tournament time controls.
Obviously they're not perfect, and as already conceded there will be positions they handle suboptimally even objectively (and again, this isn't unique to the opening; there are positions from all phases of the game they misplay), but the number of positions they play well in the opening under tournament time controls on a modern processor vastly exceeds the number they play poorly.
To make some pretense of this being related to the OP's question, my $0.02 is that that is a useful training method.
The flaws others have pointed out might be flaws if that was literally the only thing you ever did, but I seriously doubt that was what you had in mind, @savagechess2k :)
I also play from position ( various variations of the line I want to practice )
This topic has been archived and can no longer be replied to.