lichess.org
Donate

3.Nc3 is an inaccuracy, what?

[Off original topic]

@drmrboss You judge a computer assessment by their rating compared to humans. Keep in mind I was referring to near-2800 elo players.

Anyway I'm talking about play between humans.

Computers are surprisingly bad at first moves, as they are proven bad at some crowded positions, and, even if you argue they are still scientifically better than top humans at openings, we are still not talking about playing computers. Human players of all levels want to create the most possibilities. Whereas a position analysis going on for ages on a monster hardware may find very specific lines that justify an opening. But you aren't a monster computer with access to ages of data. You can try playing the computer way, but see how you go. Probably give wide options to your opponent and dig yourself a hole of critical decisions. You can't compute, you can use chess knowledge intuitively.

Most things the computer says are helpful, but some aren't, or at least you still need to truly understand why that works, before trying to apply it.

Finally, who helped bust the king's gambit was Fischer in 1961, it didn't take a computer to find the computer refutation in this case.
Which computer, who operate, what hardwares? What do you mean first moves? First 2 moves, first 20 moves?

Just show me some games you play or any human play vs powerful Stockfish or Leela Zero in properly set up computers, where people get advantage vs computers? May be top 2700+ might play better in 5% of games but likely worse in 95% of games.

Btw, in 2000, in msdoc family games ,chess programs played 1.Nf3 . 2. Nc3 first moves. And also there are 2000 rated CCRL programs today as well. And also there are 99% of people who dont know how to set up program properly( how much hash memory to use, how much time is necessary for good analysis, how to set up EGTB etc).I am not refering to them .:)
I said more than once that I wasn't referring to humans playing against computers. That just doesn't matter. Computers can compute and always find the win through brute force. That has nothing to do with human play, although it always beats humans.

What makes it different from a quick math dispute between humans and calculators, is that chess positions have tree branches so many that no computer can reach the end. That's a serious deficiency when computing is all you can do, while humans can do other things, including things logically unexplicable.

Of course, despite so different, they are still useful for human play, but only when you humanly get things out of such brute force. Therefore it's just a tool, an amazing tool, but never a chess-solving god.

All my problem with your comment is how it puts them above everyone's work. Probably a fantasy by the developers, related to being superior to top players by creating something that knows everything about chess, when that's just not possible.
"This line in the Queen Gambit Declined:

1. d4 d5 2. c4 Nf6
and then
3. Nc6?

Which is the most popular move according to chesstempo."

This line is the Marshall Defence not the Queen's Gambit Declined. The QGD is 1.d4 d5 2.c4 e6. The problem with the 2...Nf6 is that white can exploit the knight move winning a tempo with 3.cxd5 Nxd5 4. e4 or with 3.cxd5 Nxd5 4.Nf3 avoiding the break e5 and pushing e4 later with tempo gain (this last line is a finesse and was considered the best by various repertoire books) in my modest opinion 4.e4 is also good and has sense for the players that plays e4 in the Queen's Gambit Accepted.
Appreciate all the replies.

I originally thought that this was just a "typo" on Stockfish's part, as "Computers aren't the best at opening moves" etc...

I also might have made a mistake inputting Nc6 instead of Nf6 into chesstempo, leading into the chigorin defense, woops.
But this was quite informative nevertheless, since the Queen Gambit is what I mainly play as white.

@Spartako,
"The chesstempo database seems to show not the number of times the move was played, but the number of occurrences for the position reached after that move is played."

I didn't know that, seems odd... Though it shouldn't matter THAT much for early-game? I think?
I should really get into the habit of using lichess database then, seems interesting.
(Also, I would appreciate it if you could hand out a source for that statement)

So basically, 3.cxd5 is best because it wins a tempo, open the c-file and exchanges c-pawn for a center-pawn, grabbing more control of the center.

My statement is from a fact that you can easily check yourself. For example, go input this very dubious line on analysis:

1.e4 e5
2.Nc3 dxe4
3.Nxe4 Qd5

Chesstempo will say nearly 10 thousand instances of 4.Nc3, which is actually because it will reach the main line of the Scandinavian: thousands of games with 2.exd5 Qxd5 3.Nc3.

Lichess will correctly say just 46 instances of 4.Nc3 and, only after you input this move and definitely reach the main line position, it will show the thousands of games.

Of course you can go further and test something even crazier:

1.e4 d5
2.exd5 Qxd5
3.Nc3 Nf6??

Of course this move just blunders the queen, but see Chesstempo give dozens of instances for the 4th move (none of which is NxQ) because it only cares about the position it's gonna reach. Of course no master has ever played 3...Nf6??
Here is the game mentioned above

Karpov did not dare 3 cxd5, probably because after 3...Nxd5 4 e4 Nb6 balck has ceded the center, but gets counterplay against it, just like the main line Grünfeld 1 d4 Nf6 2 c4 g6 3 Nc3 d5 4 cxd5 Nxd5 5 e4
Normally Nf6 is not bad move ıf black play e5 ( e6 is bad move every theoretical move is not so good )
here my puzzle
Nb6 is theoretical move ( Nc3?! theoretical move Nf3! give advantage for white and block e5 ) but not best Nf6 equalize game quickly and black have no problem after e5 ( notice Nb6 lose tempo Nf3! but good move black can equalize game )

This topic has been archived and can no longer be replied to.