[for the rollout results scroll to the bottom of the post]
Let's start with GNU & SW. GNU 2 ply thinks it's a ND by .04, SW thinks it's much closer to a double but still has it as a bare ND. Was I sure it was a proper double when I cubed? No...was I sure that in practice it's an enormous double? Of course. This has chouette pass written all over it.
If after this post you still fail to see why it's such a huge take I'd suggest what I always do, set the position up in your bot of choice and play it out until you're content.
- The race is even, though not likely to turn into a race, it's always something I pay close attention to.
- I have two men still to extricate while you only have one albeit that one is on the bar.
- Our boards are currently equal theoretically in strength. Obviously mine is better since it's a straight prime with no out of play checkers but you still have a 4pt board, nothing to snub your nose at since there are exchanges where you'll have a change to hit either a direct or a flyer from the bar.
After that let's use our 1 ply human analysis to see what is likely to happen. The most likely scenario is that White (me) rolls a number and extends his prime in one way or another. In fact a full 29 rolls are played offensively only, while just 7 (21, 51, 54, 55) are used to play both offense and defense. What kind of effect does this have? This means that while I do have 20 escaping numbers, all 2s & 5s, I will likely be running out of time to extricate those back checkers. Furthermore, on the next roll I may have to continue my attack, be it by choice (good dice) or my hand being forced (you enter). This impedes the escape of my back men even further. It's true that I can't do everything at once and releasing those back men may be harder than it seems. [eg, how many times can you remember dancing time after time after time trying to come in against a 4pt board? same thing]
Let's take a look at GNU's temperature map on 2 ply:
If you're unfamiliar with the temperature map it simply lists all upcoming rolls for the player on roll, how they should be played, and the equity of the person on roll after each roll. By default it's set to 0 ply but it's quick to change it to 2 ply as I've done here. It's also a handy tool for graphically showing in plain color the upcoming good/bad rolls.
It may seem like a lot of red but remember my equity after D/T is somewhere in the ballpark of +0.66 so while almost all numbers leave me positive equity wise, that doesn't necessarily mean I wouldn't throw them back for another shot at a better roll. The only disastrous rolls I have are the obvious ones, 44 & 66.
The last thing I noted while playing this out is that it seemed when things went your way you go in a recube fairly quickly. (usually recube + take) This may be inaccurate, I only played out so many games, but an impression I was left with.
The score (after 0 games) is: Stick 0, Readers 0 (match to 15 points)
Move number 13: Stick on roll, Stick doubles
| Stick | 116 |
|---|---|
| |
| Readers | 113 |
Position ID: M5sHA0DsbiUAGA Match ID: MAHgAQAAAAAA | |
Cube analysis
Rollout cubeless equity +0.4273 (Money: +0.4219)
Cubeful equities:
1. Double, take +0.6647
2. Double, pass +1.0000 ( +0.3353)
3. No double +0.6444 ( -0.0204)
Proper cube action: Double, take
Rollout details:
Centered 1-cube:
0.6362 0.2476 0.0047 - 0.3638 0.0990 0.0040 CL +0.4273 CF +0.6444
[0.0011 0.0017 0.0003 - 0.0011 0.0010 0.0003 CL 0.0031 CF 0.0078]
Player Readers owns 2-cube:
0.6512 0.2525 0.0032 - 0.3488 0.0966 0.0043 CL +0.9434 CF +0.6647
[0.0015 0.0024 0.0004 - 0.0015 0.0013 0.0005 CL 0.0088 CF 0.0128]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 1050236534 and quasi-random dice
Play: 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 12 more moves within equity 0.16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]