Game 2 Forum Move 4 2-1 to play

[ah_clem]

Schig rolled 6-3 and opted to make our bar point.( via email while the cube decision was pending)

We rolled a mighty 2-1.  

Zk7wYCDg8+ABUA  : UYmoAAAACAAA

[ah_clem]

thanks to stog's tip, I can now participate in the match again.

Spoiler

Entering with the one is forced.  Where to play the 2?

The safe play 8/6 exposes zero blots and threatens exactly nothing. It's too early for cautious play, so we should to slot a point with the 2.

Normally, the best place to slot is the point you most want to make; in this case I'd really like to make an anchor, so I'll opt for moving a checker up to the 22.

bar/22

[close]

[stog]

Spoiler

bar 24, 6/4 ......i think we can best get an advanced anchor by having another helper - that is if we are hit, if not by slotting the 4 point we may actually get a board point! i am also wary of being bombed out here ( i voted for drop before btw).

[close]

[blitzxz]

Spoiler

This is a difficult move and they are not going to get any easier, which is another reason why I don't like playing positions like this. What are our goals? We need a advanced anchor and make more points to our home board so that we can contain checkers that we might hit. Our secondary winning plan is just simply overrun our opponent with series of doubles. Third and last winning plan is to get some kind of late shot from low anchor. At some point we need a correct redouble and our take decision was actually based on that fact that we might have powerful recube.

Our opponents double decision was based on the fact that he might get lots of gammons with closing us out. Secondary plan for him is to prime our anchor and/or run to home.

Based on these considerations we can eliminate the anchor split, bar/22. To get any kind of anchor is lucky and it will cut down our opponents gammon chances hugely. We are not in a position to make any kind of risky plays as we don't have home board at all and our opponent can go all out for closing us out.

Also 13/11 doesn't look good. We need to make home board points but 6/4 does this more effectively and is actually safer.

So the choice is between safe 8/6 and slotting with 6/4. 8/6 is inflexible and ugly looking move. But we really don't want anorher checker send back because this will ruin our long shot running chances. Also we are not in a hurry to get home board points immediatly, because we are not likely to get good shot for awhile and priming that one checker is hopeless. However, the ultimate problem is that after our ugly 8/6 we are likely to be forced to leave shot any way on next move and also still have horrible distribution. So I vote for 6/4, so that we get checkers out of those huge piles.

[close]

[ah_clem]

Spoiler

So thinking about this some more....

Stog has a good point that being hit makes for a better chance to anchor than splitting the backmen.  Also, Schig will probably hit loose in his home board if we leave a blot - so any 5 or any 3 puts us back on the bar with strong chances of him making a fourth point.  The splitting play isn't looking so good.

6/2 only leaves a 4 to hit, and if we're hit we have a good chance of making an anchor.  If not, we're better off making the four than making the eleven.  I don't particularly like keeping the ace point as it risks  transitioning into a passive ace-point game, but that's better than being gammoned.

So, I'm flipping over to bar/24 6/2. 

I don't see the point of candlestick making with 8/6 - we'll just leave a shot next time, possibly with Shig having a better board by then.

[close]

[roadkillbooks]

Spoiler

So we have an achor albeit the worst one.  We dont not want to allow our opponent to continue the hitting game/blitz.  For me the choice of the 2 is either 13/11 or 6/4.  Now which point is more important for us (ie which point would we rather make) and which one will be easiest to cover.  The answer to both these questions is the 4 point.  13/11 implies a builder while 6/4 is a slot.  Since our builder would also be exposed to a direct shot and in addition indirect shots.. that again tilts the scales the 6/4.  Another factor to consider is unstacking... well both of these points are equally stacked so there it is a tie. 6/4 all the way with the 2.  I must add that 8/6 looks like a very bad play (i just checked and saw that it has several votes.  We are not up against a 5 point board on the verge to be gammoned (which is one of the few times I can imagine making such a cowardly and awkard play. I advocated taking the double but I would have prefered to drop than make a play like 8/6. My guess is hat 8/6 is a blunder while 13/11 is a small error. Again 6/4

[close]

[dorbel]

Spoiler

I wouldn't mind if a partner went for 6/4, but I chose 13/11, on the grounds that I would rather he breaks his anchor to hit. If he does that, he loses a prime asset and leaves more return shots. I didn't really consider 24/22, an anchor is so valuable when you are behind. 8/6 is very weak, creating nothing. Being hit here isn't very important and may even help us to get the high anchor that we would like.

[close]

[ah_clem]

rollout

Spoiler

A tough decision.  I should have stuck with my original choice - bar/22 is best by a clear margin.  I can only surmise that sitting on the ace point and waiting to lose is not an effective strategy.  Other interpretations?

It's interesting that the ultra passive 8/6 is tied for second with the Forum's choice.


    1. Rollout          bar/22                       Eq.:  -0.299
       0.399 0.111 0.010 - 0.601 0.234 0.007 CL  -0.439 CF  -0.299
      [0.002 0.003 0.002 - 0.002 0.003 0.001 CL   0.007 CF   0.009]

    2. Rollout          bar/24 6/4                   Eq.:  -0.360 ( -0.061)
       0.368 0.099 0.012 - 0.632 0.219 0.015 CL  -0.501 CF  -0.360
      [0.002 0.003 0.002 - 0.002 0.003 0.001 CL   0.007 CF   0.009]

    3. Rollout          bar/24 8/6                   Eq.:  -0.360 ( -0.061)
       0.368 0.102 0.010 - 0.632 0.220 0.012 CL  -0.492 CF  -0.360
      [0.002 0.003 0.002 - 0.002 0.003 0.001 CL   0.007 CF   0.009]

    4. Rollout          bar/24 13/11                 Eq.:  -0.361 ( -0.062)
       0.367 0.098 0.010 - 0.633 0.219 0.019 CL  -0.497 CF  -0.361
      [0.002 0.002 0.002 - 0.002 0.003 0.001 CL   0.008 CF   0.010]
        Full cubeful rollout with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 842623946 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]

[close]

[stog]

do remember though that gnu would always play "perfectly" from that position - whereas even our combined efforts, might well not realise our best - this is probably the main failing using computer plays as a basis for individual moves -

[ah_clem]

do remember though that gnu would always play "perfectly" from that position - whereas even our combined efforts, might well not realise our best - this is probably the main failing using computer plays as a basis for individual moves -

Right.  And here "perfectly" means exactly like gnu 0-ply evaluation, which we all know is far from perfect.  (c:


Agree that sometimes you're better off by playing an less-favored move that leads to a game where you understand what's going on.  For instance, I'm reluctant to play tricky backgames against better opponents.  That said, kinwing what I know now, I'd opt for the bar/22 play; I think the collective effort is capable of playing that position pretty well.

[dorbel]

I don't see that any play here is going to be more or less difficult to play than another. Gnu's rollout probably does produce the best play, although I'm very surprised to see 8/6 second best! How might we have found 24/22 overe the board? We might have given more weight to the fact that he doesn't have the checkers ready to blitz and that fives are duped. Playing 24/22 is also thematic, bidding immediately for a better anchor, usually right when the opponent has one.

[stog]

good comments

but i feel schig's 65 would be us on the bar now (and in danger of gammon), if we had split our back

[roadkillbooks]

Ok my take on this is that after 24/22 schigs best plays will be to go all out for the blitz.. for which he is not that well equpped (all 8 pieces in the zone). So a lot of our equity probably comes from anchoring and surviving an overextended blitz attempt in which we would then begin making our inneer board and holding fort with the equity of a 3 point holding game. I often have problems with attempting to get into a bad holding game.. of course i didnt have problems getting into a bad ace point game.  But grapping the ace point we have ace point equity immediately and can choose other things to do.  Even after seeing this I think I might just make the same play again..except against a bot. 
All the other 2's are the same because onw I assume that most all of the equity is the ace point game equity and the other choices dont mean much.  Dorbel.. on the 13/11.. i normally do want the oppoent off my bar point to return hit but since we have no board at all... part of the overall strategy of 6-4 was to build a board and then get them stuck in the bar point holding game.. that was my thinking at least.. of course the numbers show us all wrong... im gonna roll it out in eXtremeGammon

[blitzxz]

best plays will be to go all out for the blitz.. for which he is not that well equpped (all 8 pieces in the zone)

This is propably the key point. I thought that there would be a big difference in gammons with split or not, but it turns out there is just 1 % difference.

But even slightest changes in this position will change the correct move. For example one more attacker in bar point or 16-point will make the split a blunder. However, extra attacker in 15-point will make not spliting even bigger blunder, then it is in the match position. But then again attacker in 14-point will make splitting or not almost equal. Understanding the dynamics of these small changes is out of my head.