How many moves are you capable of calculating in advance?

Chunking doesn't apply here. Of course GMs can see 20 moves ahead, even I can. There is no mental storage, you can look a board and move the pieces around however you like. It just doesn't help to calculate 20 moves down a single line when there is no chance that line will come to fruition, and you have to pick the right moves to calculate, not just calculate a lot of them (if you are human). The important element of calculation is whether you are creative enough to see every possibility, not how deep you can calculate.

Not everyone can remember changes made to the board in their head that far. This is why I am much worse at blindfold chess than regular chess.

Daily rating 1650, moves ahead 5 maximum (in forcing lines). Some GM said it's not important how many moves you see ahead, it's your evaluation of the resulting position.

I agree strongly with all the answers that say it depends on the position. Put two kings + one rook on the board, and visualize just the rook moving around. You can go on indefinitely, since you need to remember only one temporary thing at each ply, so the answer in that case is infinity. Put one more rook on the same board and suddenly it becomes much harder because you have to remember two things with each ply.

I want to clarify what I've said, to build on sqod's post.

I am not some sort of blindfold or chess visualization genius. I cannot even play blindfold chess yet. I have plenty of visiluaization blocks that make it hard or impossible for me to visualize things, I just want to emphasize that depth is not very important. Here are some factors that get in my way when I try to play blindfold chess:

1: Board size. 90% of the difficulty for me is visualizing the board, not the pieces. On a 4 x 4 or 5 x 5 board I could play blindfold chess indefidately with perfect recall of the position. For blindfold visluazion in a normal 8 x 8 board, I have to use shortcuts. Usually, this means I focus on a single sectinon of the board (sometimes 5 x 5, but sometimes maybe 8 x 4 or irregularly shaped) and I approximate everything else by being on this or that file, rank, or diagonal. This for the most part works, so there aren't a lot of holes. I'll a remember a 5x5 position plus a rook on the h file that would be relevent. however this doesn't work to play a full blind fold games because too many pieces are sort of floating around in the nexus. As my board visualization range gets better (it already has) this will stop being a problem.

2: Number of pieces involved. The more pieces that get involved in a calculation the harder it is to keep track. This is where chunking may apply, but I am still skeptical. I think it is possible to remember the position of all 32 pieces on the board, it's a skill that can be learned. I can remember very long melodies or even pieces of music that have way more than 5 - 7 notes after a single listen. That's because it's a skill, not just a short term memory trick. That being said, my skill is short for chess and when it comes to indefinate visualization I am a bit of a tablebase (6 piece limit). However, keep in mind that this is the number of dynamic pieces in the position. Pieces not relevent and pieces that don't move are easy to remember, as long as they don't get involved in the calculation. This is especially true when I am visualizing with a board and pieces in front of me.

3: Accuracy. It's not how many moves you consider, it's which moves you consider. Analyzing 1: Bxh7 Kh8 20 moves deep doesn't matter when my opponent plays 1: ...Kxh7. This is definitely what GMs excel at. It's not that they can't visualize 20 moves deep, it's that they don't (except for practice) because there's no point. Your opponent is never going to play all those moves unless you know they're the best moves or its forced. It's better to calculate "wide" and make sure your move works in every variation than to calculate deep. And calculating wide isn't just about visualiztion, it's about accuracy and it is the entire challenge of chess.

• Maybe irrelevant to the op but to xmans 'I think it's possible to remember all 32 pieces'. there was a study done where masters were challenged against novices. They were shown a position and asked to recreate it without sight of the board. Initially they were shown positions fromm real games and the masters predictably creamed the novices. Next they were shown positions where pieces were entirely randomly placed. Masters and novices were equally bad.
• I believe the point was that pattern recognition is way more important than visualisation.
• I hope this is pertinent and informative
• Sorry for the bold type and bullet points. My phone did it and I don't know how to change it back

Hooray , thanks chesspro. Maybe I will pretend in future I did it on purpose to add gravitas

Excellent!

That's strange, I was curious about that very question some time ago and everyone said that the masters indeed performed better than the novices. Can you link to the study?

In a thread about a year back, I questioned Carlsen's ability to recall boards set up with entirely random piece arrangements. However, the consesus on the thread was that this was indeed possible and people said that this had been tried and Carlsen was able to.

That being said, supposing it is true that masters can't memorize completely random positions, there is still a lot of room for visualization in blindfold chess. At best masters do how I described- Raw visualization combined with shortcuts. This is different from pattern recognition everything and chunk the last 5 - 9 pieces in your short term memory. It is a skill that can be practiced and learned and it is possible to visualize more than 5 - 9 pieces, just like how it is possible to memorize melodies more than 5 - 9 notes long.

At the most three moves and that takes me awhile.

This is an interesting question. In general I think it is FAR more important to calculate 3-4 moves ahead accurately than it is to see 10-20 moves ahead. However, this is where the top 10 in the world excel - you can see them going into the tank to calculate this deeply in an important tournament game.

But a lot of what they are doing is not only envisioning the position that is on the board after so many moves, they are trying to form an evaluation of who is better afterward, why they are better and by how much.

 As others have said, it is mostly pattern recognition and so forth, but there is no way to deny that practicing deep calculation will make you a better player overall. It will not necessarily make you a better player at blitz and fast time controls, because noone has the time to calculate in those time controls anyway, they are relying on intuition and their ability to spot the critical moves quickly. IE not missing tactics and strategic blunders and so forth ~

Afraid not. I have hundreds of chessl books and I'm not sure I even read it there or somewhere else. I am struggling to think what I should Google to find it online.

I promise I am being absolutely honest in that I read it somewhere. In the randomised test the masters were very slightly better but the difference was negligible. I know I said they were equally bad but I think this was explained by some pieces randomly appearing on their 'normal' squares or patterns. I will search and post if I find something concrete.

Well, first off, I'm interested in what a truly random board looks like. That would give me some insight into how I memorize positions, and possibly this question about how to best learn to play blindfold chess. I mean think about it, it doesn't even matter if the position is legal, just one of 4.8 * 10^53 possible positions.

EDIT: that's only the positiosn with 32 pieces. But the position could have anywhere between 0 and 32 pieces, if it were a complete database.

Degroot did this a long time ago (psychological tests that found masters remember real positions easily but do no better than non-players are random positions).

More recently they did it with one of the Polgars in that video documentary where they had a truck drive by with a large chess position on the side (so she could only see it a few seconds).

It's a pretty well known result (err, for chess players that is lol)

Here are some random chess positions I found to show how much of a monster these things are to memorize. Unfortunately, the program is incomplete so no pawns or kings yet. They would get much much harder once those are added.

Thanks for the source.... guy with numbers in his username... there is no reference on that particular wikipedia page but it gives me something to go off of.

That being said, I still think that its not a fair comparison- to say that blindfold chess is mostly memorization and not visualization. I think it's like if you showed a skilled pianist a sheet of computer generated random music and asked him to sight read. Then, after he failed, you concluded that "piani sight reading is memorization, not skill." Obviously, a pianist who practices sight reading gains skill at reading music, not just pattern recognition of particular harmonies and melodies. I think trying to learn blindfold chess just by patterns (as you might do after learning this result) will get you nowhere as you need the visualization skill, just like learning to sight read piano by learning all the harmonies and melodies of western music will not get you anywhere. The vast majority of the task is an element of skill and a pianist's ability to memorize music more quickly or more slowly is an element of skill. (I am of course speaking in the short term... hearing a melody and then playing it back after being given a starting note.)

EDIT:

DavidPeters2,

Perception and memory in Chess: Studies in the heuristics of the professional eye

Check for that book on your book shelf.

Randomised means proper randomised. Pawns on 1st or 8th rank,illegal positons.  Literally any piece can appear anywhere.

Anyway, I found a link! theinvisiblegorilla.com , search for how experts recall chess positions.

I'm with you xman,  it's fascinating stuff, seems we shouldn't worry about all positions only the ones that actually appear 

@xman

Honestly if those are the positions they used in that study its pretty much nonsense. They aren't even "chess positions" IMO, because they give you things like 4 queens, 4 knights or bishops, and there arent even any pawns on the board.

I would say those types of memory tests might be useful for something, but they definitely have nothing to do with chess.

011101110011001 beat me to it. Only I can lose on time in the forums ☺

Israil - I think that they have nothing to do with chess is the whole point