I've mentioned several times in the past that analyzing a position means more than staring at a rollout and inventing a plausible-sounding theodicy. One has to interact with the position. I thought I'd give a small illustration of what I mean, using Jeremy Bagai's recently posted position as an example.
Like Jeremy, I botched the decision. The first thing I do in this sort of situation is to switch to a money game to see if the score makes a big difference. Here's what we get:
          
 
        
 | | White is Player 2
score: 0
pip: 138 | Money session
Jacoby Beaver | pip: 129
score: 0
Blue is Player 1 |
|
| XGID=-BB-B-BBa---dB-Aabbe---A-A:1:-1:1:43:0:0:3:0:10 |
| Blue to play 43 |
| 1. | XG Roller+ | Bar/21 23/20 | eq: -0.128 |
| Player:
Opponent: | 48.92% (G:18.58% B:0.52%)
51.08% (G:10.54% B:0.58%) | |
|
| 2. | XG Roller+ | Bar/21 7/4 | eq: -0.280 (-0.152) |
| Player:
Opponent: | 44.67% (G:14.49% B:0.29%)
55.33% (G:11.23% B:0.60%) | |
|
| 3. | XG Roller+ | Bar/22 15/11 | eq: -0.303 (-0.176) |
| Player:
Opponent: | 43.66% (G:13.77% B:0.15%)
56.34% (G:10.10% B:0.48%) | |
|
eXtreme Gammon Version: 1.21
Qualitatively, we're seeing the same story here: bar/21 23/20 is on top and everything else is a whopper plus. The pseudocubeless numbers also suggest that we don't have to worry too much about subtle win/gammon tradeoffs either. So from this point on, I stick to money games.
Now it's time to review what I was thinking and to come up with a theory about where I went wrong. I thought I had a strong board and wanted to maximize contact, to send a White checker to the bar. But I also noted that bar/21 23/20 was best for racing, and we're ahead in the race. So that suggests a variant: What if I give myself the best four-point board? That will cut into my race lead while simultaneously making a blot-hitting contest more attractive for me.
| | White is Player 2
score: 0
pip: 138 | Money session
Jacoby Beaver | pip: 139
score: 0
Blue is Player 1 |
|
| XGID=---BBBBBa---dB-Aabbe---A-A:1:-1:1:43:0:0:3:0:10 |
| Blue to play 43 |
| 1. | XG Roller+ | Bar/21 23/20 | eq: -0.081 |
| Player:
Opponent: | 51.13% (G:14.63% B:0.57%)
48.87% (G:7.15% B:0.29%) | |
|
| 2. | XG Roller+ | Bar/21 7/4 | eq: -0.172 (-0.091) |
| Player:
Opponent: | 47.72% (G:15.07% B:0.55%)
52.28% (G:8.13% B:0.41%) | |
|
| 3. | XG Roller+ | Bar/22 15/11 | eq: -0.218 (-0.137) |
| Player:
Opponent: | 46.14% (G:12.48% B:0.27%)
53.86% (G:6.54% B:0.24%) | |
|
eXtreme Gammon Version: 1.21
That made a little bit of difference, but not much. I still don't feel that I've hit on the key features of the position. The race matters, of course, but the play is not super-sensitive to the pip count. What other features of the position have I perhaps neglected? Well, there's White's prime. It might not look too formidable, but maybe a big part of stepping up is that we'll get primed otherwise. Let's restore my original board position but wreck White's prime somehow; for instance:
| | White is Player 2
score: 0
pip: 136 | Money session
Jacoby Beaver | pip: 129
score: 0
Blue is Player 1 |
|
| XGID=-BB-B-BBa---dB-Aa-de---A-A:1:-1:1:43:0:0:3:0:10 |
| Blue to play 43 |
| 1. | XG Roller+ | Bar/22 15/11 | eq: -0.012 |
| Player:
Opponent: | 52.75% (G:17.79% B:0.18%)
47.25% (G:7.46% B:0.32%) | |
|
| 2. | XG Roller+ | Bar/21 23/20 | eq: -0.045 (-0.033) |
| Player:
Opponent: | 52.20% (G:16.92% B:0.26%)
47.80% (G:8.50% B:0.41%) | |
|
eXtreme Gammon Version: 1.21
Now that did something! We're further ahead in the race than we were in the previous variant, but now 15/11 is the XGR+ favorite. Now I feel I'm starting to understand the original position: Picking a fight is not without merit, but we can't neglect White's priming threat while doing so.
I could go on, because this is really just the beginning of what I would consider a careful and thorough analysis of the position. For example, I'd try playing with White's prime in different ways to examine the effect. (I see now, after writing all this, that Daniel Murphy has already posted some variants.) Another thing I'd look at is what it would take to make 7/4 or 7/3 float to the top.
I'll stop here because I think I've said enough to illustrate my point. Analyzing a position isn't just about inventing an Aesop's fable to "explain" How The Rollout Got Its Equities. You must not only formulate hypotheses but also test them, and repeat the process until you've exhausted your ideas (or your patience!).