jryden: I'm guessing that what you mean by the current state is equivalent to the worst-case scenario, except in a few unusual cases that I've just thought of (e.g. White has borne off 10 checkers and has 5 on his ace point, while Black has 14 on his ace point and 1 on his 19-point: a backgammon is not possible, but one might say the game is in a backgammonish state).

But the game in question was not in a backgammonish state, nor was a backgammon possible, yet I was awarded 3 points. Therefore BK behaves differently than the way you and I believed it behaved.

Is it well understood how many points are awarded for a timeout in a cube match? In this game, I was trailing 4-2 in a 5-point match when my opponent disappeared. A backgammon is no longer possible on the board; in fact, if the game were played out, every possible sequence of rolls and moves would result in a gammon. I thus expected that when my opponent timed out I would receive 2 points for the gammon. Instead I received 3 points and won the match 5-4. I feel somewhat guilty for being credited with an impossible backgammon. Not very guilty, as I expect my opponent would also have timed out in the next game, but somewhat guilty.

I had thought that a timeout would be scored the same way as a resignation -- i.e. that the player who timed out would lose the maximum number of points possible from the final position, at the current value of the cube. But I can't find this written anywhere. Has anyone else had experience with this? Is this the intended result, or a bug?

joshi tm: This game doesn't appeal to me. The rules appear overly complicated, arbitrary, and inelegant, and I doubt that the strategy will be rich enough to compensate. Even the tagline doesn't fit: this game provides more, not less, incentive to hit. Has the game even been playtested?

Apart from my other criticisms, the rules need a thorough editing. Several points are unclear, and there seems to be at lesst one contradiction.

jryden:
It's true that arpa lost to a stronger player, but this is not relevant for the Sonneborn-Berger calculation. What matters is that alanback defeated a stronger player.

arpa's and redsales's S-Bs were computed incorrectly. The correct result is:

Section 1 	BKR	1	2	3	4	5	Points	S-B (?)	Order
1 mrbugs001	2040	 	 0 	 0 	 0 	 0 	0.0	0.0	5.
2 alanback	2004	 1 	 	 1 	 0 	 1 	3.0	5.0	1.
3 arpa    	1961	 1 	 0 	 	 1 	 1 	3.0	4.0	2.
4 Ebru    	1840	 1 	 1 	 0 	 	 0 	2.0	3.0	3.
5 redsales	1732	 1 	 0 	 0 	 1 	 	2.0	2.0	4.

Thad: Yes, it is. It's too bad BrainKing doesn't use the Glicko rating system -- we could get RDs into the mix.

Perhaps, rather than looking at BKRs directly, we should look at percentile ranks.

Ouch. I didn't mean to post that all in bold. If I weren't a pawn I would fix that.

Chicago Bulls: No, not zero. The default BRK is 1300.

Yes, of course it's a question of definitions. This whole thread has essentially been about how to define overall strength in these five games. I've proposed one plausible quantitative definition. It's obviously not perfect; it inherits all the flaws of the BKR system, and it may have additional ones. I'm not convinced it's the best definition, but I haven't thought of one which is clearly better.

SafariGal: But what if no one ranked #1 in one game has sufficiently strong credentials in the others to be considered the best overall player? As it turns out, of the five top-ranked players, only arpa has established BKRs in all five games. Do you really consider 54th, 4th, 10th, 1st, and 24th better than 5th, 3rd, 2nd, 5th, and 4th?

One reasonable measure of overall strength might be average BKR across the five games. I doubt you'll find anyone who can top alanback's 2160.

JMD & NIRVANA: Yes, a "new person" indeed.

Hrqls: A game of backgammon begins with each player rolling one die. If the two dice are equal, they roll again; otherwise the player who rolled the higher number moves first, using those two dice. Thus the game can never begin with doubles.