I’ve set this thread up so spectators can make comments or ask questions without interrupting the flow of the game thread here:
To start with here’s a reply to Amarone’s post in the game thread:
Certainly computers are now absolutely essential in chess for professional players.
They can store millions of previous games, neatly categorised by player / opening / result etc.
They have established ending theory using ‘tablebases’ (databases which contain every possible position and thus ‘play’ perfect chess.) So far I think they have analysed all endgames with 7 or less pieces.
I think the main change in openings has been that more ‘tactical’ lines can be played, because the computer says they’re OK.
I don’t think any main opening lines have been rejected because of computers. Remember that players have been analysing openings for over 100 years (and publishing their results), so it’s hard to find anything too revolutionary.
I know a tiny bit about chess, but haven’t played in a long while. But all of the exchanges about quiet games or such-and-such gambit and whoziwhat’s defense make me grin. The remind me of nothing so much as the duel between Inigo Montoya and The Man in Black in The Princess Bride. And it’s outstanding.
How many moves does it usually take for various level players to no longer be using pre-memorized openings? I imagine it depends somewhat on what opening they’re playing, but just generally?
Six-piece tablebases are over a terabyte; I wasn’t able to find a reference that says seven-piece has been completely enumerated. If the size progression follows the same pattern, it would be nearly a hundred terabytes.
20 years ago, when a one GIGAbyte drive cost over a thousand bucks, that would have been ridiculous to contemplate. Today, or at least before the Thailand flooding, drives were down to about $30/TERAbyte, so it would be easy for Google or the like to host a 100 TB tablebase.
But even with quantum computing, it looks like it may be a while before we get a 64-piece tablebase that solves chess completely.
It’s actually just Mornington Crescent. Once you get good enough to read the secret codes in chess books, you realize there’s nothing encoded in them, and it’s all so the cognoscenti can exchange mysterious remarks over the heads of the uninitiated.
There are quite a lot of different ‘levels’ in chess!
I’ve seen games between world-class players that followed previous analysis up to move 30 (that doesn’t happen too often, but it shows what professional players with computer databases can do.)
I’ve got some lines analysed up to move 14 - but they don’t come up that much.
Watching club players, I guess they have 4-8 moves ready (but of course quiet openings are easier to prepare.)
In any case, ‘pre-memorized openings’ are no use at all. If you don’t understand the opening, then as soon as your opponent plays a new move, you’re in trouble.
I used to use the English notation (P-K4 etc), but to be fair it was confusing that each square had a different name depending on what colour you played.
Algebraic (e.g. e4) is easier to learn and is now accepted world-wide.
Endgames are straightforward for a computer to analyze, because you’re working to a well-defined endpoint. If you checkmate the other guy, well, that was a good line. If the other guy checkmates you, that was a bad one. If it’s a stalemate or both sides end up with insufficient power, that’s a good line for the weaker side and a bad line for the stronger side. If you get to a position that’s already in your table, you don’t need to look any further; just see where that one ended up.
Openings are trickier, because you don’t end up with a definite end of the game; you end up with a position, which you then have to evaluate, and evaluation of a position that’s not within sight of a mate is only an approximation (if we could do it exactly, computers would be unbeatable). So maybe the computer has lost every game it’s tried starting with a particular opening, but maybe there’s just some really good moves later on that it hasn’t tried that would make that opening amazing. It’s hard to be sure.
So is there another good (bloodless) defense to the position at 3. Bc4 besides pulling the bishop out (Hungarian)? What else might White do after 3. … Bc5?
Well that International Correspondence Chess notation is just using grid refences, so 1. 5254 5755 just means 1. e2-e4 e7-e5.
I suppose playing against someone who might not speak your language (or even use your alphabet) means that relying on numbers only makes sense.
I too followed Fischer-Spassky.
Nowadays you have live coverage with commentary!
I absolutely recommend the London Classic, which runs until Dec 12. The World’s best players take on the best English chaps - with very exciting chess already (see Anand - Nakamura in particular :D)
Tablebases are constructed by first generating every possible legal position with that number of pieces.
Then you find all the positions that are checkmate; then all the positions one move from checkmate; all the positions two moves from checkmate … and so on.
Then you create the tablebase form that information, which (because you have literally examined every position) plays perfectly.
You show it a position and it declares ‘mate in 15’. :smack:
It’s interesting that there’s no calculation involved, just storage (as brocks pointed out earlier.)
As Chronos says, it is indeed currently impossible to be so sure about openings - but I’m certain if a particular line leads to many quick losses that it won’t be worth playing!
I’m interested in following the London Classic and have a couple of questions if you don’t mind…looking at the current results:
Standings after round 4
No Name Win Draw Loss Score / games Tie break Rating TPR
1 McShane Luke J 2 2 0 8.0 / 4 2 black wins 2671 2942
2 Carlsen Magnus 2 2 0 8.0 / 4 2 white wins 2826 2909
3 Nakamura Hikaru 2 1 1 7.0 / 4 2758 2905
4 Kramnik Vladimir 1 2 0 5.0 / 3 2800 2886
5 Aronian Levon 1 1 1 4.0 / 3 2802 2709
6 Short Nigel D 1 0 2 3.0 / 3 2698 2654
7 Anand Viswanathan 0 2 1 2.0 / 3 2811 2583
8 Howell David W L 0 2 2 2.0 / 4 2633 2568
Adams Michael 0 2 2 2.0 / 4 2734 2510
(That didn’t line up like I thought it might…the standings can be found here).
It looks like 3 points for a win and 1 for a draw…is this typical (I was under the perhaps mistaken impression that a win was usually 1 point and a draw 0.5?)
The time controls is 40/2, 20/1, g/15’+30" - does this mean 2 hours for the first 40 moves, 1 hour for the next 20, then 15 minutes to finish the game with 30 seconds added per move?
One more thing, I noticed in the rules listed on the website the players cannot agree to a draw directly without a judgement by a tournament official - who will only grant a draw if the 50-move rule, the 3-fold repetition rule, or a “completely drawn position” which “will only appear in the far advanced endgame” applies.
In a typical tournament play can players not agree to a draw whenever they like? Is the above rule (and maybe the win counting for 3x the value of a draw) to encourage more aggressive/interesting play?
):
