Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Best 4?

32 views
Skip to first unread message

Chase

unread,
Nov 9, 2015, 12:08:24 PM11/9/15
to
XGID=-a--aCCABA-AcA---b-bbbbB-A:0:0:1:41:0:0:0:1:10

X:Player 1 O:Player 2
Score is X:0 O:0 1 pt.(s) match.
+24-23-22-21-20-19------18-17-16-15-14-13-+
| X O O O O | | O X |
| X O O O O | | O |
| | | |
| | | |
| | X | |
| |BAR| |
| | | |
| | | |
| X X | | O |
| X X | | X O |
| O O X X | | X X X X O |
+-1--2--3--4--5--6-------7--8--9-10-11-12-+
Pip count X: 160 O: 136 X-O: 0-0/1
Cube: 1
X to play 41

eXtreme Gammon Version: 2.19.211.pre-release

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

Bradley K. Sherman

unread,
Nov 9, 2015, 1:11:17 PM11/9/15
to
Chase <now...@nowhere.com> wrote:
>Score is X:0 O:0 1 pt.(s) match.
> +24-23-22-21-20-19------18-17-16-15-14-13-+
> | X O O O O | | O X |
> | X O O O O | | O |
> | | | |
> | | | |
> | | X | |
> | |BAR| |
> | | | |
> | | | |
> | X X | | O |
> | X X | | X O |
> | O O X X | | X X X X O |
> +-1--2--3--4--5--6-------7--8--9-10-11-12-+
> X:160 O:136, Cube:1, X to play 41

Bar/24,11/7

--bks

Tim Chow

unread,
Nov 9, 2015, 2:03:52 PM11/9/15
to
On Monday, November 9, 2015 at 12:08:24 PM UTC-5, Chase wrote:
> XGID=-a--aCCABA-AcA---b-bbbbB-A:0:0:1:41:0:0:0:1:10
>
> X:Player 1 O:Player 2
> Score is X:0 O:0 1 pt.(s) match.
> +24-23-22-21-20-19------18-17-16-15-14-13-+
> | X O O O O | | O X |
> | X O O O O | | O |
> | | | |
> | | | |
> | | X | |
> | |BAR| |
> | | | |
> | | | |
> | X X | | O |
> | X X | | X O |
> | O O X X | | X X X X O |
> +-1--2--3--4--5--6-------7--8--9-10-11-12-+
> Pip count X: 160 O: 136 X-O: 0-0/1
> Cube: 1
> X to play 41

The "obvious" 4 is 11/7. The other two plausible candidates are 8/4* and 13/9. I'm not seeing how 13/9 is better, but 8/4* looks like it might be better at DMP even though it's asking for a gammon loss when it doesn't work. 11/7 leaves a lot of shots anyway, and more importantly, it also gives O a good 3 to anchor up. If we're going to win this going forwards then I think we have to knock O off our 4pt and make it ourselves. On the flip side of the coin, getting all our blots vacuumed up isn't necessarily fatal if we succeed in making a second anchor. I would try 8/4*.

---
Tim Chow

michae...@gmail.com

unread,
Nov 9, 2015, 3:59:46 PM11/9/15
to
Tim, it IS a DMP-->1 pointer match.

Just because it's a quiz and should have a surprising answer 8/4* as well. X enters a Kamikazi mode towards a backgame with the usual ~ 40% chances which looks better than what he has now.


bananab...@gmail.com

unread,
Nov 9, 2015, 5:20:09 PM11/9/15
to
The 'obvious' to me is 13/9. 'Every problem is a next to somebody.' - Stick circa who knows when.

Stick

Tim Chow

unread,
Nov 9, 2015, 8:15:02 PM11/9/15
to
On Monday, November 9, 2015 at 3:59:46 PM UTC-5, michae...@gmail.com wrote:
> Tim, it IS a DMP-->1 pointer match.

Yes, I knew that...sorry, I guess the way I phrased my response wasn't clear.

---
Tim Chow

Chase

unread,
Nov 10, 2015, 4:18:29 AM11/10/15
to
I've been missing some loose hits lately, so I took a long look at 8/4*,
but between the shots and the blots and the concession of the 8pt, it
didn't seem justified. I chose 11/7 over 13/9 for the solid four prime
and because I like securing the barpoint in opposition to O's back
checker. It's not obvious to me why 13/9 is better. Perhaps Stick will
enlighten us.

XGID=-a--aCCABA-AcA---b-bbbbB-A:0:0:1:41:0:0:0:1:10

X:Player 1 O:Player 2
Score is X:0 O:0 1 pt.(s) match.
+24-23-22-21-20-19------18-17-16-15-14-13-+
| X O O O O | | O X |
| X O O O O | | O |
| | | |
| | | |
| | X | |
| |BAR| |
| | | |
| | | |
| X X | | O |
| X X | | X O |
| O O X X | | X X X X O |
+-1--2--3--4--5--6-------7--8--9-10-11-12-+
Pip count X: 160 O: 136 X-O: 0-0/1
Cube: 1
X to play 41

1. Rollout¹ Bar/24 13/9 eq:-0.413
Player: 29.36% (G:6.02% B:0.81%)
Opponent: 70.64% (G:45.78% B:16.35%)
Confidence: ±0.007 (-0.420..-0.406) - [100.0%]
Duration: 6 minutes 16 seconds

2. Rollout¹ Bar/24 11/7 eq:-0.431 (-0.019)
Player: 28.43% (G:5.56% B:0.54%)
Opponent: 71.57% (G:46.03% B:15.92%)
Confidence: ±0.007 (-0.439..-0.424) - [0.0%]
Duration: 6 minutes 01 second

3. Rollout¹ Bar/24 8/4* eq:-0.470 (-0.057)
Player: 26.49% (G:5.37% B:0.77%)
Opponent: 73.51% (G:54.05% B:29.93%)
Confidence: ±0.008 (-0.478..-0.462) - [0.0%]
Duration: 6 minutes 40 seconds

4. 3-ply Bar/24 5/1* eq:-0.542 (-0.129)
Player: 22.89% (G:3.56% B:0.14%)
Opponent: 77.11% (G:47.35% B:8.17%)

5. 3-ply Bar/24 7/3 eq:-0.560 (-0.147)
Player: 21.99% (G:2.34% B:0.15%)
Opponent: 78.01% (G:50.51% B:8.21%)


¹ 1296 Games rolled with Variance Reduction.
Dice Seed: 69314136
Moves: 3-ply, cube decisions: XG Roller


eXtreme Gammon Version: 2.19.211.pre-release, MET: Kazaross XG2

Paul

unread,
Nov 10, 2015, 6:41:33 AM11/10/15
to
On Tuesday, November 10, 2015 at 9:18:29 AM UTC, Chase wrote:
> I've been missing some loose hits lately, so I took a long look at 8/4*,
> but between the shots and the blots and the concession of the 8pt, it
> didn't seem justified. I chose 11/7 over 13/9 for the solid four prime
> and because I like securing the barpoint in opposition to O's back
> checker. It's not obvious to me why 13/9 is better. Perhaps Stick will
> enlighten us.
...

I doubt that any human can consistently avoid 0.02 errors so I don't think 13/9 is "obvious" to anyone. Perhaps Stick meant the most natural play -- the play that would most readily leap to the mind of most world-class players?

X doesn't want to be hit here. 13/9 probably does best with regard to minimizing the expected number of blots that O hits over the next two rolls. It's hard to tell them apart, in terms of shot jeopardy because both leave the same number of blots and both leave an immediate double shot. However, after 11/7, if O hits with a 5, O has another potential double shot. This makes 13/9 likely to be safer over the next two rolls. As an intermediate, there's no way I would play 13/9 over the board. I think (and hope) I would play 11/7 which is second-best, but I'm not sure.

Paul

bananab...@gmail.com

unread,
Nov 10, 2015, 10:34:40 AM11/10/15
to
13/9 leaves the opponent with good 3s and 6s, a sort of double duplication. 11/7 leaves the opponent with good 1s, 3s, and 5s. Assuming everything else is relatively the same this is what jumped out as the clear decider for me.

@Paul - remember this is set up at dmp so even though it's 'only' a .020 error that equates to 1% match winning chances which to me is a big deal. It's likely that at a normal score the error is bigger. I count .020 errors at dmp as blunder.

Stick

michae...@gmail.com

unread,
Nov 10, 2015, 10:47:37 AM11/10/15
to
in addition to Stick's factors I think there is a strategic factor too. It's a hollow prime alright, but when O hits and stays at the 7 point, then there is more ammunition to hit back, and better distribution to do point and hit inside the board.

Chase

unread,
Nov 10, 2015, 6:43:07 PM11/10/15
to
Helpful. Thanks.


>
> @Paul - remember this is set up at dmp so even though it's 'only' a .020 error that equates to 1% match winning chances which to me is a big deal. It's likely that at a normal score the error is bigger. I count .020 errors at dmp as blunder.
>
> Stick
>

I hadn't quantified it but, since I've been focusing on dmp, my
threshold for what is an acceptable error has dropped considerably. I
used to waive off .025 errors and instead focus on the large blunders.
Now, I'm finding even 0.010-0.015 errors annoying. I didn't do this
intentionally, but it seems to have come as a byproduct of my studying.

Paul

unread,
Nov 10, 2015, 7:23:04 PM11/10/15
to
So how did you do in the Othello quiz? I only got 1 right.

Paul

Tim Chow

unread,
Nov 10, 2015, 7:49:40 PM11/10/15
to
On Tuesday, November 10, 2015 at 6:41:33 AM UTC-5, Paul wrote:
> I doubt that any human can consistently avoid 0.02 errors so I don't think
> 13/9 is "obvious" to anyone.

While it's true that everyone is going to make errors > 0.02 once in a while, even a top player at DMP, it doesn't follow that, for a *particular* 0.02 decision, nobody will find it obvious. For example, it's easy to concoct a bearoff decision at DMP where the equity at stake is tiny but almost everybody will regard the top play as obvious.

---
Tim Chow

Chase

unread,
Nov 10, 2015, 7:56:47 PM11/10/15
to
I think I got four right. But to be clear, my point wasn't about my
skill level, which is nothing to write home about (~5.0 at dmp and ~7.25
for match play). I was just sharing my observation that prioritizing dmp
study has tightened up my perception of what constitutes an unacceptable
error considerably.

Paul

unread,
Nov 11, 2015, 8:15:13 AM11/11/15
to
I agree with what you say on this post. However, in chess commentary, the phrase "obvious move" has two very different meanings. It can mean "obviously best" or it can mean "the play which would occur most readily to experts." An instance in which these two definitions differ could be a case where I have checkmate in four, and I also have an opportunity to immediately win a queen for nothing. In such a scenario, if someone asked "What's the obvious move?", I think the intended answer would be the inferior play of taking the queen.

Assuming that "obvious" means the same in backgammon as in chess, I thought it was possible that Stick meant my second meaning, and wasn't implying certainty about the rollout. On the other hand, maybe he finds the play "obvious" in both senses.

Paul

0 new messages