Korchnoi and the French Defence

Btw, if you go to Stockfish 12 at depth 45, it will say e6 as number 1 move to e4 

@mpaetz - I’m confused. If a computer can get you to an advantageous position, can it not also give you clues as to why and suggest ways to exploit it?

I should think that your assertions are belied by the facts, right? If they didn’t help, they wouldn’t be used. A preparation team would be accused of malpractice if they didn’t use computers to analyze lines. And if more beastly computer analysis comes along, then the lines that this suggest will be exploited.

Because what the computer sees as a slight advantage can be cashed in by playing the computer's further moves. You and I will not be able to play find all these moves, and our opponents will certainly never play the lines the computer thinks its opponent will choose, so very quickly the position will become very different. A human is better off playing moves that we can understand the reason for choosing and plan ahead appropriately. Analyzing, and discovering blunders, are the realm in which computers are most helpful. Finding an opening line in which, after 20 moves, one side has an advantage, but the other side has six different choices ranging from -.31 to -.67 is useless in practical terms.

Could be @winawer. I don’t know. I do know that it doesn’t do great in tournaments. Stockfish won last year’s TCEC (not CCC—different tournaments) on the strength of wins in the English and French.

Stockfish–Leela Chess Zero, game 92 (French Winawer) 1. e4 e6 2. d4 d5 3. Nc3 Bb4 4. e5 Qd7 5. Qg4 Bf8 6. Nf3 b6 7. Bb5 c6 8. Be2 Ba6 9. O-O Ne7 10. Bxa6 Nxa6 11. Ne2 c5 12. Ng3 Nb4 13. c3 Nd3 14. Rd1 c4 15. b3 Nxc1 16. Rdxc1 cxb3 17. axb3 g6 18. Qg5 Ng8 19. Qf4 Bh6 20. Ng5 Bg7 21. Ra6 f6 22. N5e4 fxe5 23. dxe5 dxe4 24. Nxe4 Nh6 25. Nd6+ Ke7 26. Rca1 Rhd8 27. g3 g5 28. Qxg5+ Kf8 29. Rxa7 Nf7 30. Nxf7 Rxa7 31. Rxa7 Qxa7 32. Nxd8 Qd7 33. c4 h6 34. Nxe6+ Qxe6 35. Qd8+ Kf7 36. f4 Qf5 37. Qxb6 Qc2 38. Qd6 Qe2 39. Qd7+ Kg6 40. Qe6+ Kh7 41. Qf5+ Kh8 42. c5 Qe3+ 43. Kg2 Qxc5 44. Qe6 Qb5 45. Kh3 Qf1+ 46. Kh4 Qg2 47. h3 Qf3 48. Qc8+ Kh7 49. Qf5+ Kh8 50. e6 Qxb3 51. e7 Qg8 52. Qg6 Qc8 53. e8=Q+ Qxe8 54. Qxe8+ Kh7 1-0

Sorry, I’m on mobile so I can’t show it on the board.

Leela played a sub optimal French. The main lines of the Winawer are the strongest. 

It’s funny. I watched a lot of it (Covid free time) and I remember the commentary of GM Oleksiyenko saying several times that this or that move was anti-positional or this or that move strayed far from theory and was probably a mistake. I remember him saying such a thing about a move Leela made that he described as making no sense, the. In the mirror game (each plays the same opening on both sides) Stockfish turned around and played the exact same move. Maybe I can find it. I think that was a French as well.

I remember stockfish once converted it into a regular french and won the game 

I found it. It wasn’t a French, it was a Sicilian. I could post it (the starting position was from a Karpov-Korchnoi match), but I don’t think anyone cares. But my point was that what GMs often call a positional mistake, computers often think are the best moves. Saying that Leela lost because she played suboptimally is rather tautological. 🤷‍♀️

Qd7 has been proven to be worse than the main c5 lines with the help of computers. 

 

Computers are not the end all, be all of chess.  Are they overall better than humans?  Yes.  But that does not make them perfect.  I play correspondence at ICCF, where computers are allowed.  If you just blindly follow computers, you will lose!  Is a computer's top choice correct 9 times out of every 10?  Probably!  But all it takes is one slight misevaluation and you are dead!

 

Also, computers are horrible at evaluating openings.  If your 1 CPU computer says 1...c5 is +0.18, 1...e5 is +0.23, 1...e6 is +0.24, and 1...c6 is +0.26, does that automatically make the Sicilian the strongest move?

At @thriller - of course you are correct, computers are not the be all and end all. But 9 out of 10, or even 5 out of ten still makes them vitally important in opening preparation.

When doing prep, elite players don’t just look for the best lines, but just as importantly lines that will surprise opponents and give the person you’re facing opportunities to slip up. After all, they all know all of the main lines, so opening prep is also about coming up with something new that doesn’t just blunder the game away.

Computers can do this infinitely better than humans can, and just because Leela lost by playing “suboptimal” moves does not mean her lines can’t be useful to humans in a practical sense. After all, SF is rated 3700+ now and much more able to exploit this suboptimal move much better than the best humans who are rated 900 points lower. I’d bet any amount of money that Leela would handily destroy any human after making her “mistake”.

These are actually the best kinds of lines that prep teams like to explore, right? If one side thinks that a line is bad because theory shows that it can be punished easily, but it’s shown that it actually has possibilities, it can be a great weapon.

The most obvious example is the Berlin defense. Time was that it was considered bad. Then it was used as a surprise weapon at the WCC as a drawing weapon. And nowadays computers have advanced Berlin theory to the point where practically every Ruy you see at the highest levels is a Berlin.

So yes, computers aren’t everything, but their value doesn’t lie in their ability to confirm what we already know, but to find holes in the conventional wisdom.

NikkiLikeChikki wrote: "These are actually the best kinds of lines that prep teams like to explore, right?" 

Not right at all. Teams like to explore lines the computer and player likes, not lines the computer likes but seem horrible positionally for the human.   

NikkiLikeChikki wrote:  And nowadays computers have advanced Berlin theory to the point where practically every Ruy you see at the highest levels is a Berlin.

And finally he betrays that he has no idea what he's talking about. Brighter than the usual ygnr, but no cigar by a longshot. 

All I know is what I’ve read. This is a snippet from an article I read on chess games.com:

For most of that time, the closed variations beginning 3... a6 4.Ba4 Nf6 5.O-O Be7 have been considered the main line of the Ruy Lopez, but ever since Vladimir Kramnik used it to win the chess crown from Garry Kasparov in 2000, the Berlin variation - the so-called "Berlin Wall" - has become increasingly popular at the highest level to the point that it could now be considered the main line of the whole Ruy Lopez.

Shall I provide the link?

Well you should be more discerning about the source. That is definitely not true. 

I did a google search online and the only webpage with those words is this article for beginners by an anonymous person that attracted 3 comments. 

https://www.chessstrategyonline.com/content/tutorials/introduction-to-the-chess-openings-ruy-lopez

If you're going to make statements about what is going on in the chess world you could try looking what is actually being played. eg. Tata Steel, Hastings (coincidentally enough the first ruy lopez in tata steel this year was actually a berlin defence). 

https://www.chessgames.com/perl/chess.pl?tid=101980

https://www.chessgames.com/perl/chess.pl?tid=101840

Not just look at a beginner's article that could have been written by a 16 year old boy.

And even that article doesn't claim anything like what you did. It said it could be considered the main line, you claimed almost nothing else was played other than it.  

So why try to act like you know things and are really knowledgeable? 

Please. I’m not allowed to engage in hyperbole? Of course others are played and I follow all the tournaments, and it does seem to me that the Berlin is the most played. I looked that up after some people on a twitch chat started saying that it seemed that the Berlin was about all you saw anymore.

I tried to do a statistical breakdown by year and compare it to others, but I could find no such resources.

The only thing I could find was a little graph beside the usage of the Berlin on chessgames.com that showed a steady increase, but that’s not comparative.

To say that I’m a fool or an idiot is just wrong. I didn’t pull numbers from a dark orifice.

Besides, you have no “gotcha” moment. My initial claim was that computers have revolutionized what is played and made a hyperbolic claim about the Berlin to support my argument. The essence of the claim remains. And I’m not wrong when I state that the Berlin is extremely popular these days.

Please show me where I can get a year by year breakdown of variations, and I’d be happy to do a breakdown so we can come to an evidenced conclusion to this argument rather than settling it by calling me a noob. If I’m wrong, I’ll gladly admit it. If you are wrong, you will apologize. Deal?

What the hell are you talking about? 

The "evidenced conclusion to this argument" is that there is more Ruy Lopez than almost all the Berlin that is played at the elite level. In fact it is almost all not the Berlin of the Ruy Lopez that has been played. I posted the links to this above - Tata Steel, Hastings, those are the last two big tournaments played this year. You can see them on the left of the front page of chessgames.com  I dunno what's wrong with some people here but I am going to block myself from coming back to this site. 

First. Again, you are purposefully missing my broader point. Second, two tournaments is a small N and doesn’t constitute a sufficient sample size to make broad generalizations about usage. If you looked at the WCC you would think that the Russian Game was super popular and that the Sveshnikov was the most popular Sicilian variation. Idiosyncratic decisions in a small sample do not constitute a trend. You need to choose a relatively long period (a few years at least) and separate games by rating and variation.

That’s what the hell I’m talking about. I’m trying to settle this using data rather than bombast.

     I am frequently amused that these forums veer off into arguments between two or more people, becoming heated and personal, with continued arguments concerning minor points with no relation to the original topic. The original question was "Why did Korchnoi use the French Defense less often in his second world championship match vs Karpov?" The obvious answer is that they had played the French so often previously that Korchnoi knew Karpov and his team would prepared some lines they thought would have a great chance of success, so a better strategy for Korchnoi was to prepare some lines himself in different defenses to take Karpov out of his comfort zone and leave him at a bit of a disadvantage.