Three Away Two Away

From: PhilSimborg   

Excellent...I love having the 3-1/5-3 reference position.

You stated that the leader's take point is 26.5. I have heard others state that they believe the take point to be close to 30 percent.

What number do you believe it to be; what number do you apply over the board; and will we know sometime soon, for sure, what the number really is, or are we just guessing?

I hope you will write another article dealing with the leader's cube action ATS. You might recall in a match with MCG last week as the leader at this score he gave me a cube which could easily have been either too good or not good enough, but in either case, it was a take. He cubed, hoping to take advantage of the possibility that I might make a mistake and drop, but my clever decision to flip a coin diluted that ploy.

Stick,

In the first reference position you say, "There's a fair bit of volatility in the position even though you have two on the roof and the inferior board you have a big cube! "

But in this position no one is on the bar!!! In short, the position is unclear and what you are saying about it and from whose perspective you are discussing it needs to be clarified. (E.G. You say that GNU is cubing and that you need to decide what to do...but yet appears the polar opposite that "guest" is cubing and GNU needs to decide)

Thanks,

Jason

From: neilkaz   

Very nice article, Stick !

Re: take point for the leader and Phil's question. Phil, I think you're confusing the TP at -3-3 which is 30% to the TP at -2-3 which, using the very reasonable 32% figure for -2-1C is 26.5%.

GNU 0 ply rollouts give somewhat less than 32% for -2-1C while 2 ply rollouts give somewhat more and my 3 ply rollout was slightly more than 32%. I use 32% when playing world class opponents.

If you use an inferior and old MET with the very low 20% gammon figure which results in -2-1C being 30%, then the TP is 20/70 or 28.6%, but in any event, it isn't terribly likely that one can judge a position that closely OTB and if one can judge that MET matters here with the decision, said decision isn't likely to be a big error.

Hi Stick,

Thanks for another great article! I have a general GNU question.. In the first two cube decisions in the article I noticed that the winning (and gammon) percentages go up after giving away the cube (centered 1-cube vs Gnu owns 2- cube). I would think they would in general go down since you cant double out your opponent. Can you explain whats going on.

Thanks,

John

And some more information on the takepoint at 30%: http://www.bkgm.com/articles/met.html

Tom Keith there derives the 30% TP when explaining how to compute a MET.

I'm reading this great article for the second of what I expect to be numerous times.

Reading the comments, I see now what Jason (ROQREGA) was confused about. First I think he was assuming that the pattern of your article was "diagram 1, discussion of diagram 1, diagram 2, discussion of diagram 2, etc." So he was confused because you started your discussion of diagram 2 *before* showing diagram 2, and he missed the words "shown below." Although I did not find it confusing myself, perhaps some readers would prefer it to be more visually obvious where you stop discussing one position and start discussing a new position, without having to catch words like "above" and "below" in the text.

The second thing is that he was confused by the fact that in your first diagram, the players are labeled "gnubg" and "Guest" but in the text you say that "gnubg" is actually you and "Guest" is GNU Backgammon. Again it didn't confuse me but if one person was confused then others might be too.