Back to "The Zare Formula" written by Douglas Zare

Cube Vs. Checker Play Errors

by Douglas Zare
25 August 2003

Douglas Zare

What distinguishes experts from each other and from advanced players, checker play or cube play?

Many players say that it is the use of the doubling cube. Some have told me that they understand checker play, but need to work on their cube actions. It is also the case that individual cube blunders tend to be larger than individual checker play mistakes. It is memorable when someone hands you a 6-sided token of their esteem for your position, and after you miraculously save the gammon, the kibitzers remark, "I can't believe you took that!"

However, when we look more closely, it appears that competitive players give up much more in checker play errors than in cube errors. Is that what makes a champion? No, most intermediate players make more checker play mistakes than cube errors, while making plenty of both. Further, measuring errors in EMG (Equivalent to Money Game) tends to overstate the importance of cube errors. Checker play mistakes are even more important when errors are measured in MWC (Match Winning Chances)

The rest of this article (15.72 K) is premium content.

GammonVillage Subscriptions:

You must be signed in to rate articles.

Feedback about this article:

Subject: Re: Cube Vs. Checker Play Errors

From: Zaphod1001

Date: 26 Aug 2003 04:45 EST

Interesting stuff.

I have been observing this myself. From most of my online matches i see that most of my own, and my opponents errors are checker play errors.

Your comment about going through every move of your matches so that you get positive reinforcement is usefull. I have started doing this lately, and i think it really helps. I used to only look at my errors earlier, but then i easily got depressed, and forgot about all those decsisions that i got rigth. Now i try to go trough all my moves. The only downside of this is that it is very timeconsuming, and i get less time for playing..

Another idea is to try to organise the error that you do and classify them. This can be a bit difficult. Myself i found out that i frequently missed moves that jumped into the outfield. Another blind spot for me was not seeing the right move..

A feature that i would be hoping to get on a bot someday is some sort of database, where you can collect difficult desciscions that you had and somehow classify them.(this might not only be problems that you got wrong, but also the one where you found the best move..). Later you could then quiz yourself, and see if you had improved since you last had the problem.

Subject: Re: Cube Vs. Checker Play Errors

From: DBrthrtn Member has a photo available

Date: 26 Aug 2003 18:30 EST

Here's a way to get practice at checker play for different match scores. Play money games against Snowie with the move panel active, making the move you think is right for money but also considering what move to make at DMP/gammon- go/gammon-save. Once you make the money move and the money equities come up, check how the moves rank at the other three scores with the buttons on the move panel (labeled 11, 21, and 12) before continuing. It doesn't slow things down as much as you may think it would. I wish GNU would add the buttons to their annotation window....

Subject: Re: Cube Vs. Checker Play Errors

From: plasticmillion

Date: 27 Aug 2003 14:35 EST

A few of other techniques that I (should) use to improve checker play

Review recorded matches of other players. This is a good way to get acquainted with positions that might arise less often with your style of play. Also, I find it is very useful to see the kinds of mistakes a world-class player makes. Finally, it is positive reinforcement to see a great player make a play that Snowie agrees with but that you know you would never have found, since it proves that we humans really can master this game.
Play nackgammon. I don't do this as much as I should, but it is a great way to get practice with complex positions since they come up more often.
Play one-point matches. I don't do this much either, but it is great for developing intuition about how much role gammons play in a given decision. Sometimes I know from experience that a given move is better, but never realized that another tactic entirely is called for when gammons don't count and the cube is out of action.

Matt

Subject: Re: Cube Vs. Checker Play Errors

From: MSquonk

Date: 05 May 2004 13:06 EST

I don't doubt your conclusions, but I think your data set cannot answer your introducing question. According to the data, the intermediate group gives up ~10% MWC relative to the expert group. Nearly half of that difference is due to the cube handling. Now, from my intermediate point of view, I have no hope of improving beyond expert status, so my potential for improvement in cube handling seems to be nearly as large as in checker play. Practically, to me the question what distinguishes experts from advanced players is more relevant than what distinguishes advanced players from perfect players (ah, Snowie evaluations, actually). Oh well, probably I'm just confused.

Maik

Subject: Re: Cube Vs. Checker Play Errors

From: zare

Date: 31 May 2006 04:47 EST

Good point, Maik. Improving cube handling does seem to account for over 40% of the improvement from the advanced/intermediate group to the expert group. So, this data does not convincingly answer the question of what distinguishes experts from intermediates.

This data does support the idea that experts and intermediates alike give up more equity through checker play errors than cube errors. The experts gave up 4 times as much equity through checker plays as they did through the cube (10.5% vs. 2.6%), and the intermediate/advanced players gave up 2 times as much (16.0% vs. 7.2%). I meant that to be the main point, rather than the introductory question.

Douglas Zare

You must be signed in to post comments.