Do you think a tournament based on the following format would be popular?

All winners progress to next round plus a certain number who lost but played with the lowest error rate?
I would suggest using XG XGR+ as the arbiter as it's the strongest useable (non-rollout) bot setting.
Participants would have to play their matches knowing that their error rate would be judged as such.

Every match would have to be recorded, then entered into XG. The directors would have to hire one recorder for every two participants, with their fees coming out of the prize fund (where else?). That amounts to a huge rake levied on the participants, not to mention a large time delay for all the matches to be entered and scored. Count me out.

Honestly, I think focusing so much on error rates would take too much away from the game itself. The excitement of backgammon comes from the dramatic swings, gutsy plays, and lucky rolls. Focusing on the error rates would take away from those moments somewhat. It would also slight some of the winners of matches by basically saying, "Well this guy lost his match, but he still advances with you because we know he's a better player than you are."

I think in the end it becomes a tournament to beat the computer, not the other player.

Fair comment. In principle, if there were no costs to the prize fund, would you take part?

No. Scoring by error rate rather than actual result is just another step on the road to taking the backgammon out of backgammon. If one finds the luck and randomness of backgammon somehow offensive, there are excellent alternatives available. One is an obscure game called "chess", said to be popular in Russia. There is also an even more obscure game called "Go", rumored to be popular in Asia. As for me, I find the luck and randomness of backgammon rather exciting.

Would you say that backgammon is your favorite game, Bill? Obviously you've had great success with it, but I know you got your start in chess and have written quite a bit about it. If you could only play one game, which would you choose?

I've invested years of my life studying poker, but I'm finding I like backgammon even more for several reasons. It's really a beautifully complex game that you can't really win just with blind luck. I played chess and go in my teens, but didn't like the lack of luck in those games. I was in chess club in high school, but didn't explore go all that much because I found it difficult to find opponents.

No prizes for telling me who said this:
"What makes Backgammon 2000 a particular interesting variation of the game is that it contains a built in "skill-o-meter". By changing the ratio of the challenge stake to the basic stake, you can move the skill versus luck ratio of the game from the normal backgammon ratio (when challenge stake = 0 to any arbitrarily more skillful ratio (as the challenge stake approaches or exceeds the basic stake)."
From an article proposing using rollouts to settle bets on errors you think your opponent made during money sessions.

Why not just all play XG a certain distance and each round discard the half with highest error rates?

That's not what I was proposing.
As an example. Let's say there's 48 entries so 24 matches in round one.
Put through the 24 winners plus the 4 lowest error rate losers.
And continue so on until maybe the semi finals when it's pure knockout.
I haven't worked out the exact numbers but that's my thinking.

It should be more heavily weighted towards the winners than the lowest error rates.

I said it, in the last issue of Inside BG (1999). The idea was a new way for generating props during money sessions. It was a bad idea, and it died an appropriately natural death, as it should have.

I personally would like playing in this style tournament. My checker play is very close to the bots (still working on cube play), so that would give me an advantage. I like having an advantage.

I'm well aware of this. That's why there was a question mark at the end of my sentence.

Just seems a bad compromise between two very different things. If you want skill only, go for rates v XG and lose the bottom half of the field each round; if you want regular BG play regular BG.

All a format like this does is give top pros a free-roll through each round, scares away the fish due to said free-roll, and penalises many of the human EV elements of BG (bluff doubling etc), which the bots mark as mistakes.

Something just occured to me, I will often generate games in GnuBG where my play is "Supernatural", where I have a Snowie error rate of 0.00, and a rating of 2050. These are often games where GnuBG is cleaning my clock. It is hard to screwup a move when all you are doing is dancing and then racing your men around the board in an effort not to get gammoned.

So I'm not so sure that a low error rate is always the best measure of skill. The type of games being played will also have a huge impact on error rates. For example, a failed blitz results in a complex backgame with two Giants battling it out. One of the Giants makes a small timing error that costs him the game. Meanwhile, me and geneftw both roll 65 twice and end up in a quick race. Geneftw proceeds to roll several doubles then doubles the cube and I resign. I lose but don't screwup. Good news for me is that I advance to the next round while the Giant goes home.

well, you vs giant and you roll 65 twice followed by doubles. You win while giant goes home. There's luck involved either way.

Lol, we could buy into this tournament with our sklansky bucks.

Hi,

my favourite tournament format would be as follows:

between every two players there's a horizontal flat touchscreen with a gnu/snowie/XG (depending on the sponsor of the event) backgammon game on it. There are no problems with illegal moves or illegal dice (except when there is a problem with the randomgenerator).

Now for the match-format:

I favor a sort of swiss tournament: The first round is paired randomly. We play for instance 7-point matches. After the first round the winners will play against other winners, but the error rate is used as a secondary tool for the pairing (the player with the lowest error rate plays against the player with the highest error rate). After a few rounds there will be the need of some shifting in the lower scoring regions to make sure that you dont play the same player twice.

5 rounds will be enough to know a conclusive winner with 32 players. 6 rounds for 64 players and so on.
Extra rounds will give players a comeback chance. Then the highest player will play the first player in the list that he didn't play against yet, and so on.

Each player will leave the tournament with a personal matchbook (on a tournament USB-stick) with all his games analysed. Ofcourse all the games will be available online as well,

greetings k.