[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

31P-62D-xx and 31P-52D-xx -- comparison

Posted By: Nack Ballard
Date: Sunday, 6 June 2010, at 4:13 a.m.

In Response To: 31P-xxx-21 data, and 32D-62D-11 analysis (Timothy Chow)


2O ' ' ' '5X1X3X ' '1X5O

2X ' ' '2O4O '2O ' ' '3X

31P-62D-xx: Blue to play 53, 32 or 21


Nack, I assume you have no data for 31P-62D-53 or 31P-62D-32 either? Would you predict that S rolls out best, as the eval suggests?

You assume correctly: I have no data for any of the 31P-62D-xx rolls (see position diagrammed above). As I stated in my other post, I would not split but it seemed close. So, yes, I would confidently play S (Split, 24/21 13/8) with the larger (more racy and less well-prime-timed) roll of 53.

For the in-between roll of 32, I would play D (Down, 13/11 13/10) only because as a matter of style I lean towards long-term complications (leading to positions that are more difficult to play for both sides) when it is this tough to judge which move is theoretically better.

From your other post:

Looking at GNU 2-ply, I see that it prefers priming to splitting with 31P-62D-21. However, I still think that there is more to be said for splitting than most people have indicated. For instance with 31P-62D-53 or 31P-62D-32, GNU 2-ply slightly prefers splitting (it's very close though, so a long rollout is definitely needed in these cases). Although there is some danger of getting attacked, I think that if 13/11 6/5 is superior to 24/21 then it's because 13/11 6/5 is such an effective developing move and not because splitting is conceptually wrong here.

That does seem to be what the other respondants as a whole were saying or implying, doesn't it? I agree with them in the sense that White has ten checkers in the zone and a 5-pip (i.e., 1 pip + half roll) lead in the race. However, anyone who fails to distinguish between the two positions below is missing a vital element:


2O ' ' ' '5X1X3X ' '1X5O

2X ' ' '2O4O '2O ' ' '3X

31P-62D-xx: Blue to play 53, 32 or 21


2O ' ' ' '5X '4X ' '1X5O

2X ' ' '2O4O '2O ' ' '3X

31P-52D-xx: Blue to play 53, 32 or 21


In the left-hand position (the one Mochy posted and you're addressing, and repeated from above), White has a 7pt blot. In the right-hand position, that blot is instead stacked on the 8pt. Blue has more incentive to split in the left-hand position: the greater danger of succumbing to a successful attack is outweighed by the value of

(1) The second direct-shot threat on White's blot (which poisons her 61 31 53 52 rolls), and
(2) The headstart on anchoring or escaping from White's impending prime (which is stronger in the left-hand position).

In other words, Blue's incentive to split is reduced from Opp having brought two checkers down (and by enough to make it slightly better to develop offensively than to split with the small roll of 21) but the incentive is still there, and it is greater on the left than on the right.

For 31P-52D-xx (the right-hand position), I also have no data, but I believe that it's right to bring two checkers down with 53, and even more so with 32, and 21 should be played along similar lines with 13/11 6/5.

Counting the number of checkers in the zone is worthwhile, but it certainly doesn't account for everything. Routinely going by that alone would cause one to play the appropriate 53 (and 32 and 21) in the right-hand position but the wrong 53 (and the right 21 but perhaps by luck, and maybe or maybe not the right 32) in the left-hand position. That method would also cause one to err if the position is 31P-32D-xx (move the variable checker from White's 7pt/8pt to her 10pt, not diagrammed), where a splitting play may well be right will all three rolls!

"Don't split against stacks" is often a more reliable early game axiom. It usually refers to a fourth checker on the 8pt (or more precisely to nine or more checkers combined on the 8pt and 6pt). Also, the value of plays can be quite sensitive to a small change in the race; it affects the probability of winning an ensuing priming battle.

One minor point: You said, "...if 13/11 6/5 is superior to 24/21 then it's because 13/11 6/5 is such an effective developing move..." That seems overstated to me: I don't see the 11pt builder as particularly effective. (A checker brought down to the 10pt, 9pt or 8pt is significantly better. Also, the 6/5 ace helps less than people seem to think; in fact, I mistrust the evals when they claim that 13/11 6/5 is so much better than 13/10, I'd bet it's within .01.) Perhaps it's semantics and I'm nitpicking, but rather I would say that 13/11 6/5 is the strongest way to play the weakest roll, and a key reason that it is best is that it includes an elevated chance of winning a priming contest due to the racing deficit.

That being said, if Blue is propped/offered/allowed to add a free 1-, 2- or 3-split to his properly played 13/11 6/5 move (even in the right-hand position, but especially in the left-hand position) he should certainly do so, which means not only that there is an inherent gain from splitting (it's not conceptually wrong in that sense, and that's probably what you meant), but also by implication that the 11pt builder is far from worthless.

Nack

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.