@StupidGM, humans just can't think the way engines do. It's simply not possible unless Elon Musk does manage to connect the human brain to computer chips. Engines don't think; they brute-force scan every single move at a rate of millions of positions every second. That is simply not something the human brain can match.
No, the days of humans beating computers in an even match are long gone.
Tic-tac-toe is so simple that a human can analyse every possible move - i.e. think like a computer. On the first move, while there are nine legal moves, because the board is symmetrical, it's really only three unique legal moves (center, edge, or corner). Additionally, the game can never have more than a total of nine moves. As such, it's quite feasible to match the computer; you can simply calculate all the way to the finish after each move.
To match the computer in chess in this manner, you would need to memorise every single move. And, since there are on average about 30 legal moves in a typical chess position, the search tree expands exponentially. Even if you were to memorise the entire Chessbase Mega Database of almost 7 million games (which is not humanly possible), you would still be a long, long way from memorising every possible move.
What you might theoretically manage to do is memorise a line where you win against a particular engine. But what happens when someone steps outside that line? Then you're outside your memorisation and have to play on your own.
It's not that simple. You have to calculate 20-30 plies ahead to match Stockfish.
But if you think you can do that and become world champion, by all means go ahead. So far none of the top GMs are playing like that.
Assuming you try to compare it to human Elo ratings, a 4000 is simply not possible. Consider that human players already sometimes play perfect games and will draw a game against perfect play here and there. I would bet it to be somewhere around the 3400-3600 Elo range. Todays engines still are blind in some positions and play most likely at a 3100-3200 level I would estimate. 3300 sounds too high since that would mean about 9/10 scored points against 2800 players.
It's likely impossible to get to 4000 as Ken Regan has estimated that a perfect chess player plays at a 3600 level. Also, here's a key concept that some helpful member pointed out in a forum thread a while ago: every 400 point difference between two players represents being roughly twice as good at chess. Can you imagine a computer being four times as good at chess as current computers, and eight times as good as Magnus Carlsen? Here's a hint: it's not likely.
If humans played "perfect games" with any frequency, they'd at least draw Stockfish once in a while. They can't. Engines can keep hair-splitting until they find an edge to win.
They can. A player like Magnus Carlsen should be able to draw at least half (if not most) of his white games against Stockfish, he just needs to play out openings which quickly leads to dead drawn positions. This is not too hard for a GM of his strength to do. He can´t hope to outplay the engine though, that yes, but with a draw strategy he will draw a lot of games.
"It was believed back in the 1980s that GMs could "hold a draw" against machines, because the simply didn't make enough errors that were serious enough to lose."
lol, thats a good one, now we know you have no clue what you are talking about. Carlsen wouldn´t win one in a hundred games yes, but he would draw at least half of his whites and not only, that is >15 points that he would score out of 100. Todays engines won´t make too many mistakes to perfect play either, 4000 or 5000 is ridiculous to say the least.
I think in 100 games against the worlds best computer Magnus Carlsen would lose 99 and draw 1. I dont think he would win a single game, but of course it's possible. I dont think it matters if computers ever get to 3400, or 4000 or 10,000 because at some point chess is solved where there are one (or a couple) correct moves for any given position. Whatever that eventual number is depends on how the rating system works when that happens.
I think in 100 games against the worlds best computer Magnus Carlsen would lose 99 and draw 1. I dont think he would win a single game, but of course it's possible.
You need to understand it is not about winning. Carlsen shouldn´t be able to win a single game against todays top engines, unless some unexpected bug in that software occurs. He can easily score a lot of draws though, because of the opening knowledge that players have nowadays. They often can steer/simplify positions into dead drawn endings early on, which are not complex. With that alone a top player will be able to score a whole bunch of draws, especially with the white pieces.
I think in 100 games against the worlds best computer Magnus Carlsen would lose 99 and draw 1. I dont think he would win a single game, but of course it's possible.
You need to understand it is not about winning. Carlsen shouldn´t be able to win a single game against todays top engines, unless some unexpected bug in that software occurs. He can easily score a lot of draws though, because of the opening knowledge that players have nowadays. They often can steer/simplify positions into dead drawn endings early on, which are not complex. With that alone a top player will be able to score a whole bunch of draws, especially with the white pieces.
Maybe. I'm sure that's possible. But I dont think that's what would happen. I think playing for a draw would only increase his chances of losing. At some point he would likely make an incredibly small mistake, an imperfection. That is when the computer would probably have the advantage. Until it happens, and he does play the worlds best computer 100 times, I still think he would win zero and draw maybe one or possibly two. I think the rest would be all losses.
i wonder if the world champion will be able to beat an all brilliant bot. Like a bot who's every move is brilliant.
Also joe biden is president now, just saying, cuz you guys are from 2017 and im from 2021
I don't think that centaurs are any stronger than engines.
In 2014 Nakamura + Rybka played against Stockfish with no odds and lost 0.5-1.5.
Some things to consider is that engines improved somewhat since 2014; at that time Rybka was weaker than Stockfish, so it would be more fair to play Nakamura+Rybka vs Rybka, or Nakamura+Stockfish vs Stockfish; also, there were only 2 games played, who knows what would happen in, say, 7 games.
Nakamura was using an old version of Rybka. If you gave Nakamura Stockfish 8 and had him play against Stockfish 8, he would win more often than not. The top centaur players are better than the pure engine.
Also, if you try to play on ICCF (which permits engine use) and just play the top engine move, you will get destroyed by the top centaur players over there - even if you have better hardware.
(BTW, ICCF is a chess federation comparable to FIDE, not a competitor)