BGonline.org Forums

Bot content etc etc

Posted By: neilkaz
Date: Monday, 9 February 2009, at 5:16 p.m.

In Response To: Snowie and GNU don't agree (neilkaz)

OTB I quickly played 24/18 6/3 as 18/9 looked to be a bit too hard to win from, noting I rarely hit next turn and sometimes still can't win it later.

I have my Snowie setup to analyse my matches I play at GS and my GNU setup for SHG (although I can Snowie a SHG match if I want) I prefer to use GNU for match analysis (this pos will show you why) but am too lazy to switch it back and forth from SHG and GS imports and esp for quick timed 3 pt matches on GS vs weaker players.

Anyhow, in going over the match, Snowie surprised me by preferring 18/12 6/3 by .022 over 18/9 (weird) and by .029 over my 24/18 6/3. What I presume is happening is that Snowie plays the checkers in lookahead like it is for money..ie wants a G and thus thinks that I win more games by leaving a checker back. This is likely true if Snowie always attacks it.

Anyhow here's GNU's rollout.

The score (after 0 games) is: Opp 1, neilkaz 0 (match to 3 points)

Move number 6: neilkaz to play 36

Pip counts: Opp 75, neilkaz 107
Position ID: trvDAABt2wFgIA Match ID: QYl5ABAAAAAA

#		Ply	Move	Equity
	1	R	24/18 6/3	-0.7005
			0.1480 0.0036 0.0002 - 0.8520 0.0622 0.0000 -0.7005 -0.7005 0.0008 0.0005 0.0002 - 0.0008 0.0078 0.0007 0.0016 0.0016
			Full cubeful rollout with var.redn.
			616 games, Mersenne Twister dice gen. with seed 816125413 and quasi-random dice
			Play: world class 2-ply cubeful prune [world class]
			keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]
	2	R	24/18 4/1	-0.7142 ( -0.0136)
			0.1418 0.0023 0.0001 - 0.8582 0.0660 0.0003 -0.7142 -0.7142 0.0008 0.0004 0.0001 - 0.0008 0.0076 0.0001 0.0015 0.0015
			Full cubeful rollout with var.redn.
			616 games, Mersenne Twister dice gen. with seed 816125413 and quasi-random dice
			Play: world class 2-ply cubeful prune [world class]
			keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]
	3	R	18/9	-0.7547 ( -0.0541)
			0.1205 0.0046 0.0000 - 0.8795 0.3380 0.0115 -0.7547 -0.7547 0.0014 0.0007 0.0035 - 0.0014 0.0153 0.0026 0.0028 0.0028
			Full cubeful rollout with var.redn.
			616 games, Mersenne Twister dice gen. with seed 816125413 and quasi-random dice
			Play: world class 2-ply cubeful prune [world class]
			keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]
	4	R	18/12 6/3	-0.7573 ( -0.0568)
			0.1194 0.0042 0.0001 - 0.8806 0.3448 0.0120 -0.7573 -0.7573 0.0014 0.0006 0.0000 - 0.0014 0.0145 0.0026 0.0028 0.0028
			Full cubeful rollout with var.redn.
			616 games, Mersenne Twister dice gen. with seed 816125413 and quasi-random dice
			Play: world class 2-ply cubeful prune [world class]
			keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]
	5	4	18/12 4/1	-0.7270 ( -0.0264)
			0.1330 0.0075 0.0000 - 0.8670 0.2924 0.0087

Here's GNU 2ply eval. Not perfect, but unlike Snowie it finds the best play.

The score (after 0 games) is: Opp 1, neilkaz 0 (match to 3 points)

Move number 6: neilkaz to play 36

Pip counts: Opp 75, neilkaz 107
Position ID: trvDAABt2wFgIA Match ID: QYl5ABAAAAAA

#		Ply	Move	Equity
	1	2	24/18 6/3	-0.6929
			0.1515 0.0044 0.0000 - 0.8485 0.0232 0.0005
	2	2	18/12 6/3	-0.7047 ( -0.0119)
			0.1427 0.0105 0.0001 - 0.8573 0.2729 0.0093
	3	2	18/9	-0.7056 ( -0.0128)
			0.1422 0.0107 0.0001 - 0.8578 0.2741 0.0079
	4	2	24/18 4/1	-0.7106 ( -0.0177)
			0.1430 0.0038 0.0000 - 0.8570 0.0246 0.0007
	5	2	18/12 4/1	-0.7275 ( -0.0346)
			0.1316 0.0099 0.0001 - 0.8684 0.3164 0.0123

Messages In This Thread

• Snowie and GNU don't agree
  neilkaz -- Sunday, 8 February 2009, at 9:35 p.m.
  • Snowie and GNU don't agree
    Stick -- Sunday, 8 February 2009, at 9:57 p.m.
    • Snowie and GNU don't agree
      Chuck Bower -- Sunday, 8 February 2009, at 10:26 p.m.
      • Snowie and GNU don't agree
        Chuck Bower -- Sunday, 8 February 2009, at 11:00 p.m.
  • Snowie and GNU don't agree
    Rich Munitz -- Monday, 9 February 2009, at 12:54 a.m.
  • Snowie and GNU don't agree
    John O'Hagan -- Monday, 9 February 2009, at 3:32 a.m.
    • Snowie and GNU don't agree
      Rich Munitz -- Monday, 9 February 2009, at 4:41 a.m.
      • Snowie and GNU don't agree
        garyo -- Monday, 9 February 2009, at 6:32 a.m.
    • Snowie and GNU don't agree
      Tad Bright -- Monday, 9 February 2009, at 4:50 a.m.
      • Snowie and GNU don't agree
        Stick -- Monday, 9 February 2009, at 5:04 a.m.
      • Snowie and GNU don't agree
        neilkaz -- Monday, 9 February 2009, at 7:06 a.m.
  • Bot content etc etc
    neilkaz -- Monday, 9 February 2009, at 5:16 p.m.
    • GNU-bg 3-ply eval...
      Chuck Bower -- Tuesday, 10 February 2009, at 10:26 p.m.
  • Snowie and GNU don't agree
    Frank Berger -- Monday, 9 February 2009, at 9:44 p.m.
    • Snowie and GNU don't agree
      Matt Cohn-Geier -- Monday, 9 February 2009, at 9:52 p.m.
    • Snowie and GNU don't agree
      neilkaz -- Monday, 9 February 2009, at 9:55 p.m.
      • Snowie and GNU don't agree
        Frank Berger -- Monday, 9 February 2009, at 10:23 p.m.