chess no theory

#40
No Fischer and Carlsen were exposed to theory, but also practiced play against themselves and say they progressed by that. AlphaZero got to 3000 playing 700,000 games against itself exclusively, without any theory at all. AlphaZero thus reinvented theory like the Berlin and the Grünfeld and also invented new theory. So maybe the method of playing against oneself is more potent than one thinks. Stockfish was fed theory and AlphaZero only practiced against itself with no theory, only the Laws of Chess.

For humans its pointless to point how how AZ learned chess.  Humans are not capable of doing that.  Now if we could harness a fraction of that?

This glorification of an engine is detestable

I think chess engines like most things that advance technology and mankind's understanding does more harm than good.

citing playing against oneself as a potent way of learning and giving examples like carlsen, fischer, alphazero. i doubt these entities form even a fraction of 1 percent of the chess-playing population in terms of representing a control group. they are the exceptions to the normalcy. you can't use them as yardsticks to justify your point.

Did they say how much it helped them progress?

I'd say it probably helped Fischer more than Carlsen.

AlphaZero and Stockfish are programs used on expensive computers...people are not.

My opinion is that you can learn the most by playing much stronger players often in order to learn from your own mistakes quicker. Playing equal or lesser strength players can help you learn from theirs.

I'm sure you can learn some things playing against yourself. It won't lead to a very fast progression (imo) without any outside help.

Fischer and Carlsen did not learn chess by playing against themselves. Carlsen has frequently said that his first burning ambition in chess was to beat his sisters! And it took him quite a few tries before he could. After that he played constantly on the internet, so that he had amassed a lot of experience at an early age

Fischer lived before the internet, but from the time he fell in love with chess, he played other, stronger players as often an he could, as well as devouring every chess book and magazine he could.

Both Fischer and Carlsen read a lot of books and went over a lot of grandmaster games. 

#47
Both Fischer and Carlsen started by playing their sister.
Both Fischer and Carlsen used other means besides playing against themselves.
In an interview Carlsen said he progressed most when he played against himself.
Stockfish was fed theory. AlphaZero did not get any theory and just played against itself.
So apparently playing against oneself is a powerful means.

All of the great players played a lot of chess against other people from a very young age. No human can play three million games against himself in 48 hours like AlphaZero, so how AlphaZero learned chess seems quite irrelevant

#46
"My opinion is that you can learn the most by playing much stronger players often in order to learn from your own mistakes quicker."
That is true, but if you are Fischer, Carlsen, or AlphaZero, then how do you find much stronger players to play? It is no coincidence that both Fischer and Carlsen are from the west. A player from the Soviet Union or now Russia had it easier to find strong opponents. Beating up weaker players is not as useful as playing games against yourself.

#49
AlphaZero only played 700,000 games against itself. That it could get to 3000 strength that way is relevant, as it proves that it is possible to excel at chess without any theory. A human cannot play 700,000 games against himself, but a human cannot reach 3000 rating either, even with all available resources.

AlphaZero even created new theory. In the Sinquefield cup we see 1 d4 Nf6 2 c4 g6 3 h4. This would have been frowned upon a few years ago, but is a direct result of games by AlphaZero.

#51
Magnus would not laugh if AlphaZero kicked him. He hates losing.
#52
No, I have spent a lot of time on openings, but now I realise it was all a waste of time.