Maths, mnemonics, chess

or, How to memorize chess openings (long version)

Watch Arthur Benjamin, “Mathemagician,” multiply two five-digit numbers IN HIS HEAD:

The scene in question starts around 10:30, although the whole video is quite interesting.

He explains his methodology just a bit, and then attempts to verbalize the whole thought process as he solves the problem, which he breaks down into three segments. After finishing one segment, he remembers the resulting number by repeating a phrase: “cookie fission”.  Substituting words for numbers. The mathemagician’s mnemonic aid. After he finishes another of the three segments, he will recall his first segment result with “cookie fission”, combine those two segments, and continue on from there.

Now stay with me:

There is a wonderful, ridiculous, ultra-sharp, ultra-theoretical chess opening called the Botvinnik variation of the Anti-Meran line in the Semi-Slav defense against the Queen’s Gambit. (Yeah.) Starts 1.d4 d5 2.c4 [Queen’s Gambit] … c6 3.Nc3 Nf6 4.Nf3 e6 [Semi-Slav, versus a …Bf5 setup] 5.Bg5 [Anti-Meran, rather than 5.e3 Meran] …dxc4 [Botvinnik, as opposed to 5…h6 Moscow variation].


Botvinnik Anti-Meran
Botvinnik Anti-Meran: let the fun begin!

Now I don’t have a great chess memory. I have to keep looking up these names in Wikipedia just to know what a Meran is and what an Anti-Meran is. And the Botvinnik line is so notoriously well-researched that GMs often play more than 20 moves they’ve memorized before making any original decisions. So this is not a logical line for me to play, with either color.

However, it’s fascinating and fun. Either side could get mated, or lose or win an ugly endgame. So after taking it up in blitz for a time, I had a chance to play it over-the-board for the first time against a 2350+ player this weekend. I half-expected to get embarrassed by making some basic well-known theoretical blunder. And I did make a bonehead mistake immediately upon forgetting the book line on move 14 (but actually we wound up making a draw!).

Listening to IMs talk about this kind of line, I was always blown away that they not only remember reams of analysis, but even remember specific games between other players that have taken this course or that course within the opening. “Shirov-Van Wely, Wijk An Zee 1994, continued 27…Qg8! and won” they might say.

Holy mackerel! They remember who played it and where?! Astonishing!

But does Arthur Benjamin give us a hint as to why that might be? I think so.

What I failed to grasp previously is that the names and locations are actually HELPING the chessplayer remember the analysis. I know who Shirov is, and I know van Wely – what they look like, where they are from, how they play. I can attach their names to particular strings of chess analysis just as one might store a string of numbers as “cookie fission”.

Now that I have grasped this probably obvious fact, I may actually be able to remember some of my King’s Indian lines.


3 thoughts on “Maths, mnemonics, chess

  1. Some years ago, I came across a technique of multiplying 2 digit by 2 digit numbers which made mental calculation feasible for most people. With practice, I could do 3×3’s (with nowhere near the speed or fluency of the mathemagician, of course!) His “words to numbers” technique was surprising, different and cool.

    I wonder if the effect you describe helps people with synesthesia with certain memory tasks. Which also leads me to wonder how many GM’s might have the condition.

    While multiplying 5 x 5 digits is amazing, I was a bit more impressed with the mathemagician being able to get the day of week from a date in the 19th century. Taking into consideration the irregularities of the calendar (leap years and differing days in the month).

    1. I kinda get the words to number thing. Essentially you might have to memorize a sequence of 999 words, but you could help yourself by using alphabetical order to some degree (all words 1 to 99 start with A, for example, with the second letters also proceeding by alpha). So a five-digit number is a two-word sequence: 88,462 = word #88 (cookie) and #462 (fission). Not saying I could do it, but that’s an approach anyway.

      Agree that the calendar thing is absurdly cool.

Comments are closed.