Exact, Exacting: Who is the Most Accurate World Champion?

Are the engines trained using a database that includes prior games, moves and outcomes by the same players being considered? If so, the accuracy metric and argument might be circular, and some careful cross-validation, leave-out, etc methods must be also be seriously considered.(?)

A very interesting attempt to quantify difference in accuracy of chess play historically, Thank you!

What I like most about it is it stimulated this discussion which brings out the limitations of a single number ACPL in describing what humans regard as accurate play.

There are caveats to this number as people have pointed out: the difference between sharp and dull drawish games, the later getting generally better scores for example. Then there are the apparently in some cases inaccurate scores used: the reliability of conclusions depends on your input data! Then the inherent biases of the analysis tool (chess engine) which is the same tool used by players to prepare in the computer era, and finally there are those who immediately leap on these numbers to justify their opinions regarding certain players and their standing relative to others (for example Capablanca, although this may be due to playing style and the length of drawish games he played in the WC).

An often repeated adage by Mark Twain immediately comes to mind:

"There's Lies, damned lies, and then there's statistics"

@manyways-play said in #69:
> I think lichess silently filtered it for everything but me. Make your way through my bio on the profile :).
I don't know German, but I'm happily listening to your piano :)

This survey speaks very well for some of the older, pre-computer era players. As we have seen in the current WC, today they are memorising computer lines and drawing games. No cigar there. How good would Capa, Alekhine & Lasker have been with engines and databases?

5 days ago, it was hard to imagine that we would see a more accurate game. And yet here we are. Game 7 was even more accurate than game 3, making the game the most accurate world championship game ever. What makes this especially surprising is that both players were exhausted both physically and mentally and still played better than ever before. Maybe I should stay awake until 2 AM before my next chess tournament...

it was a boring drawish and known line and therefore got a very low ACPL because these top players make no mistakes there. It is exactly what I was talking about in my last post, and what many others have also stated in this thread. This one number gives an incomplete picture.