Actually you could be correct. My hypothesis is just that, a hypothesis. But for the reasons I gave, I believe that there is indeed a limit to human chess-playing ability, well below computer levels or accuracy. My view is that the maximum chess players thus far that humanity has produced are Lasker, Capablanca, Alekhine, Fischer, Karpov, and Kasparov.

Perhaps there are no new mating patterns out there, but chess has seen a tremendous exploration of positional ideas. Karpov's games can serve as an encyclopedia, and some of Petrosian's set-ups defy straight-forward understanding.

Wow! Thanks for confirming my suspicion. To me at least, if Lasker was playing training matches with Znovsko-Borovsky, presumably in the 1910s, he probably was secretly doing the same in the 1920s and 1930s. This could account for his incredible strength even in the 1930s, past 60 years old!

I think Lasker was a rather cunning individual. He always seemed to have tried to give the impression that he did not study chess. I think he did that so that his opponents would underestimate him. The fact is that in games were his opponents misplayed the opening, Lasker would blast them off the board.

Lasker's play in the 1930s has always befuddled me. It seems biologically and medically impossible. I would not believe it for an instant if some one were to tell me that some chess player could play Candidate level chess consistently above 60, but Lasker's 1930s games are undeniable facts; so the only obvious thing to do is to create hypotheses trying to explain the facts.

Botvinnik of course took this practice to the extreme in shaping the Soviet Union school of chess.

<Everett> I join you in your <All kudos to the oldies, but modern players are blessed with all their hard work and understanding "up front.">

Here,

<the chess players today certainly have more patterns to work with (in all phases) than their predecessors>

I believe it should be qualified. This is mostly true for the opening phase. For the middlegame and the endgame, 99% or all of the patterns we see in 1910 to 1940 games are the same as today's. At least, I cannot distinguish a 1910 to 1940 game from a master game yesterday if I were to eavesdrop in the middlegame; they look exactly the same.

For openings, I can easily distinguish a game if it occurred post WW2 if its opening employs the KID, or some lines of the Sicilian such as the Richter-Rauser or the English attack or the Dragon. For most other present-day openings, the moment they start deviating from theory, in a few more moves the middlegame that they produce is completely indistinguishable (to me) from pre WW2 games. As for endgames.. if one takes endgame positions from random master games here in CG.com without knowing the dates, there is no way of knowing in what era they were played.

For instance, if you take a look at this endgame, and were only given the fact that the two strongest players of the day were playing it, but not the dates and the identities of the players, how would you know what era it was played in? You can't. From all appearances, it could have been played yesterday.

Morphy vs Anderssen, 1858

Now you were allowed to peek at the opening. An Evan's Gambit between two of the world's strongest players? This must be a pre-WW1 era game.

With every new great player, the chess world is enriched by their particular intuition and insight at the board. Look at Kramnik with the Catalan, playing it like he was possessed by the spirits of Karpov and Smyslov on steroids. IMO I don't think he could play it with such mastery without studying the predecessors I've named. Every new chess genius will now have Kramnik's unique insight on a database, and this information would be unavailable to anyone from previous ages.

I guess what I'm trying to say is all the greats had chess wisdom, but the newbies have greater access, greater mass of quality information, and greater tools to assimilate efficiently.

We have to start somewhere. In any case, 0.8+ would be used for all games and for all subsequent WC matches that would be analyzed, if I understand <Bridgeburner> correctly. There shall be consistency.

Now that you say it, you have a point with these two. Here is what Petrosian said about himself and his chess style:

<I was brought up on the games of Capablanca and Nimzowitsch, and they became part of my chess flesh and blood. - Tigran Petrosian>

Yet almost every one would agree that Petrosian developed a highly unique style. Superficially his games might resemble some of the quiet but powerful positional games of Capablanca and Nimzovich. However, it's not really Capablanca IMO, and not really Nimzovich. He would make subtle prophylactic moves and then... Wham! His opponent discovers he has walked into a lost position.

Here is what Karpov had to say about himself and his style:

<The ideal in chess can only be a collective image, but in my opinion it is Capablanca who most closely approaches this... His book was the first chess book that I studied from cover to cover. Of course, his ideas influenced me. - Anatoly Karpov>

<We have to start somewhere. In any case, 0.8+ would be used for all games and for all subsequent WC matches that would be analyzed, if I understand <Bridgeburner> correctly. There shall be consistency.>

It isn't about consistency, what percyblackeney hinted at is that a thresholds of 0.8 and 1.4 are clearly a way too large.

If he really wants to cling onto a threshold-based system, instead of average error-based one, I'd suggest implementing four categories of mistakes: an inaccuracy 0.20 1p;
a mistake 0.40 2p;
a serious mistake 0.80 3p;
a blunder 1.60 4p;

After 17. Rxb7 Nxb7, we have the following position:

click for larger view

What is White's compensation? In a Pelikan set-up, White has a Tarrasch Knight in a hole of opposite color as the Black Bishop. No one would take anything away from the excellence of this game.

Question is, were top pre-WW2 masters playing such exchange sacs regularly? Yes! Even pre-WW1 masters knew all about the exchange sacrifice. In fact, Janowski was rather notorious for his exchange sacs. Even at the slightest opportunity, he did the exchange sac thingy. See

Game Collection: David Janowsky's exchange sacrifices (Congrats for <Karpova> for the collection.)

Just take a look at this game:

Dide vs Janowski, 1901

14. c3 Rxb6 15. Nxb6 Qxb6 16. cxd4

click for larger view

This is a Pelikan type set up like in the Kasparov vs Shirov (1994) game above; but this time it is ironically Black that does the exchange sac thing, and also straight out of the opening. The compensation is two powerful Bishops in an open position.

Did Janowski also do the exchange sac in order to obtain Tarrasch Knights? Yes!

Janowski vs Kupchik, 1913

click for larger view

Here, Janowski pushes 22. e4, inviting Kupchik, one of the strongest players in the USA and who also was of GM caliber, to plant a Knight on e3 with 22...f4 23. Qf2 Ne3. The willy Janowski did it on purpose, for then he does the exchange sac straightaway. 24. Rxe3 fxe3 25. Qxe3 Nc8

click for larger view

Now his Knights located on the dark squares are virtually immune to his opponent's light-square Bishop.

Not content with that, he creates another hole on f6 which his Knight soon occupies by 26. Ng4 Rg6 27. e5 Rg7 28. Bc4 Bf7 29. Nf6

click for larger view

These are not one-game flukes. Janowski did the exchange sacrifice against World Champions and Almost World Champions from Steinitz to Alekine:

Steinitz vs Janowski, 1898

Lasker vs Janowski, 1909

Pillsbury vs Janowski, 1904

Capablanca vs Janowski, 1913

Capablanca vs Janowski, 1916

Janowski vs Alekhine, 1914

He beat Steinitz, Lasker, Pillsbury, and Alekhine in the above games.

I would like to point this out in justice to the often underrated Janowski. Just because Lasker massacred him in two matches does not mean he was a patzer. There is no doubt that the games above could only have been played by a GM caliber player.

chess engines are intrinsically dumb about chess. they have little or no understanding per se, and in some types of positions they will remain clueless forever. hence, the idea about gauging <humans'> absolute chess strength by comparison to diffs in evaluations by chess engines without plans, understanding or awareness about their <human> opponents - i consider that a flawed idea from the beginning. <even if computers indeed had solved chess> (in terms of a hypothetical 32-piece table base), judging human vs human strength relationships by whether a human perfectly follows the theoretically optimal game-path from any given position would be nonsense in my opinion.

the purpose of a game of chess is to win. a great number of positions can only be won by accepting some temporary "sacrifice" (mostly positionally) in order to create a scenario where the chances to achieve a decisive result increases.

a generally stronger player that plays another very strong, but slightly weaker player could choose between two strategies in a hypothetical 6 game match:

a) always try to play the objectively best and soundest moves

b) occasionally try to unbalance the position, for instance by making some positional "sacrifice" (inviting/accepting a doubled pawn, going for a manouver that temporarily decreases his/her space advantage - the kind of things that make engine heuristics produce <lower> values for the player if there is no visible compensation within the horizon of its evaluation).

it's not hard to imagine that the end result for strategy a) would usually be 6 draws (assuming the players are quite close in strength, as was my assumption). for strategy b) the result would often be, given our assumptions, for example 2 wins and 1 loss for the stronger player.

what we do know almost for sure is that the "computer comparison method" would consider the player stronger if he chose method a) over b) - <regardless> of the outcomes of the games.

in conclusion: the point of a game of chess is to <win it> - that's how you win a tournament or a match. a strong and smart player does what it takes to achieve that, including playing moves he/she knows involves some risk and temporary weakening.

the computer eval comparison method implies a philosophy about what <strength> and <best> mean which isn't neutral or even natural - and i consider it to be <in conflict> with the strategy current day top players need to follow in order to succeed at the highest level. (consider how people complain about leko's fundamental strategy.)

this change in the <shear number> of highly skilled and technically adept players also explains why nobody can have insane 20 game win-streaks against fellow top players today. it's not because they are significantly weaker than fischer that they can't repeat such feats, but because their opposition is significantly stronger.

solid, technical players like leko and kramnik would've completely destroyed capablanca's opposition, probably more devastatingly than capablanca did himself. if we could've sent karpov at his peak back to face capablanca's typical opposition, it would've been a massacre too.

augalv, thank you very much - that's generous of you.

<but I really don't see the need to resort to sarcasm and provocation in your comment.>

what's the provocation? that i don't consider capablanca's white play an effective way to win chess games against today's elite? that's an honest opinion and those who are provoked by that opinion own the problem.

I would like to point out to other kibitzers that are truly interested in the analysis of the Lasker-Schlechter World Championship Match (1910) that if I believed that the hypothesis above:

<If a master lived before WW2, he cannot be as strong as present day masters.>

is an axiom, I would not even bother to participate in <Bridgeburner's> ongoing computer analysis. Why? Because even if it, or any other analysis, shows that Lasker and Schlechter played perfect errorless chess throughout the match, I would still adhere to the axiom that <If a master lived before WW2, he cannot be as strong as present day masters.>

No 'proof' will ever be sufficient to make me change my mind.

So for those who believe that the above proposal is an untestable axiom, what I have to say is that I disagree with you, even if I respect any of your sincere thoughts on this.

nimh, that's true. it was mostly meant as a humorous comment to engines' fallible evaluations.

however, seeing how differently different engines assign values (even if they agree about the move, as quite often is the case between fritz and rybka) it's hard to decide on any "numeric levels" or eval-ranges to categorize mistakes.

also, in quite many endings (when still outside tablebases) it's <usual> that engines are clueless about which moves increase winning chances and which don't. and in lots and lots of possitions their evaluations indicate winning advantages when a strong human player immediately sees that the position is a draw.

hence, in order to have anything that resembles a meaningful method, these kinds of positions must be known. but do note my comment above:

that looks like a challenge! i'll see if i find time for collecting some sample positions. maybe i'm prejudiced about the middle games played between 1910 and 1940 - so i need to play through a notable number of top level games from that period - but i find it hard to believe that you find anything similar to the number of "messy" middle games from that period compared to the number of such games i know and can dig out from say the last 15-20 years.

This game is interesting:

Capablanca vs Janowski, 1916

Capablanca vs Janowski, 1913

Capablanca vs Janowski, 1916