|
| ||||||||||||||||
Including your contribution, here is data for 32Z-52:
Money
Snowie .. [S K7 X9 H26 D36 x47] 31**20 10 5
GnuBG .. [S K1 X12 H13 D26 x38] 46*15*5*
XG 3-ply [S K5 X11 H19 D22] 62**15*
XG 4-ply [S K4 X6 H28] 46**3
Other scores
GnuBG .. d[X S4 H6 K7] 5
GnuBG .. g[K H3] 15
Key: S = Split (24/22 13/8), K = Kill (8/3*/1*), X = hit-and-split (24/22 8/3*), H = Hit (13/11 8/3*), D = Down (13/11 13/8), x = alt hit-and-split (24/22 6/1*). Error sizes are in thousandths.
Before the brackets, d = dmp, g = gammon go; else assume money. After the brackets is the number of trials in thousands (rounded down), and an asterisk repeats a number. For example, "[S K4 X6 H28] 46**3" means that (for money), S is best, K is -.004 and X is -.006 after 46000+ trials, and H is -.028 after just 3000+ trials.
24/22 13/8 play is the only splitting play, so it can be nactated either S (Split) or Z (reverse split). That is, by "Z," David meant 24/22 13/8.
Compare the left-hand position 32Z-52 to the right-hand position 52S-52. On the left the quiet D play is -.028, whereas on the right D is slightly ahead of S by .002 (bot aggregate). Most notably, the hitting plays of K, X, H, and x are a lot closer with 32Z-52 (-.004, -.010, -.020, and -.042, respectively) than with 52S-52 (-.043, -.061, -.054, and -.071). Consistent with David's observation, hitting in the left-hand position prevents White from using her 10pt builder to make a key offensive point; also that same blot may become a liabililty in an ensuing blitz or exchange of hits.
If Claude, Neil, David, Alexandre or anyone else would like to participate in the XG 4-ply rollout project, here is the 4-ply queue. (The only changes are that 43D-44 and 32Z-52 are now completed.) If you let me know when you start the rollout, I'll add your name next to your chosen position and post an update.
Nack