an ordinary chessplayer wrote on 03/02/21 at 20:19:08:

I never heard about anybody choosing their first move based on what the engine shows. In fact the first move says nothing about the usage of an engine. I answer 1.d4 always with f5 despite engines don't like it at all but it would be a very big mistake of my opponents to think that I didn't use the strongest engines very extensively.

The choice of an opening is often something very complex at masterlevel.
- What is the evaluation of the engine?
- Did my opponent recently play this opening or not?
- Does my opponent have a lot of experience with this opening or not?
- Did I recently check this opening or not?
- Which games could my opponent know from me about this opening?
- Do I still remember the important details or not?
- Should I play for a win or is a draw sufficient?
- How do I feel/ what is my mood?

Most masters go through such checklist already at home when preparing for the game and when surprised at the board will take a lot of time for their first moves. They make a selection out of their repertoire based on a list of criteria but the evaluation of the engine is definitely part of it. A really losing continuation most professionals won't play but they do variate between moves which are rather close of evaluation.

an ordinary chessplayer wrote on 03/02/21 at 16:29:05:

I think you are largely underestimating how much those guys know of openings. I see a lot of those guys practicing whole day all different openings so they can switch any time in a tournament and try to poke a hole in the repertoire of the opponent. So I don't bite it that Gabuzyan never checked with an engine any lines after 1.e4, 1.c4, 1.b3 or 1.f4.

Now you claim that his performance is 2 classes above your current level so then you must be a grandmaster and then I need to shut up. Gabuzyan had a performance of 2790 see https://en.chessbase.com/post/hovhannes-gabuzyan-wins-armenian-championship-2021 and as 1 class = 100 points (difference between e.g. FM and IM) then you must be 2590 fide.

brabo wrote on 03/02/21 at 08:51:44:

Yes, that's what I want it for. First I analyze using my brain, then I check it with an engine. My interest is not so much finding the best move, but in getting better at finding good moves when I play OTB.

Compared to the best engines, even Carlsen's games are mediocre, so I don't really understand why you added that word.

MartinC wrote on 03/02/21 at 09:13:47:

Yes, I don't really care what engine I use, they are all good enough and have been for a long time.

I view claims of a revolution in chess openings to be made either by marketing types or by people who spend so much time looking at engine evaluations that they have lost sight of the pragmatic requirement to just play decent moves. In the recently completed Armenian Championship, the winner played 1.c4 twice, 1.b3 twice, 1.e4 once, and 1.f4 once. I guess he did not ask the engine to approve of his openings beforehand. As black he played the Najdorf, but also the Benko Gambit. Since this performance is two classes above my current level, my takeaway is I don't have to worry about whether some 3400-Elo engine thinks those openings are junk. In other words, objectively speaking, subjectivity works fine.

By the way, H.Gabuzyan - L.Babujian, 81st ch-ARM 2021 (Gabuzyan's sole loss) is a fascinating game and a good illustration of how the opening *influences* the game result, but does not directly *cause* it.

brabo wrote on 03/02/21 at 08:51:44:


I wonder slightly about that last. Computers have been more than strong enough to catch any/all realistically soluble mistakes that we make for a very long time now.

I like playing about with LC0 because you get an unprejudiced viewpoint on opening theory, which is interesting in its own right.

I have high hopes that both Stockfish and Leela would gain from cooperation. I belive that highly training neural networks are the key to all engines improvments.

By the way I looked again at

https://ccrl.chessdom.com/ccrl/404/

And it apears to change a lot every day or week.
Right now Fat Fritz 2 is rated 1 elo higher than Stockfish 13.
1 elo is within the margin of error.

@Bibs - Good link, thanks for that. The history is more extensive than I knew. It's certainly not a case of oops, sorry, didn't realize.

MartinC wrote on 02/19/21 at 09:36:18:


Yes, agreed, doing it without the deception would be reasonable. I found a similar conclusion in the lichess comments, I won't give the author's name in case they don't want to be mentioned here, but it is #49 there.

Quote:

So there are two issues here: (1) Infraction against the Stockfish developers (partially mitigated afterwards); (2) deception against the customers. Either one by itself would be serious, taken together it's alarming. And as mentioned, totally unnecessary for ChessBase to have done it this way.

I really think this is a very bad look for Chessbase. Generating unnecessary bad will toward them. Really bad.

No idea about the guy Albert Silver at all. Never heard of him before this nonsense. 

My impression is that basically there has been plenty of good will toward the company till now, as the main database is good and used by nearly all. I use CB 15 and have lots of their opening downloads. But this all looks very dodgy.