Game 3, move 6: roadkillbooks 6-5 NEXT, move 7 Forum 2-1

[Zorba]

On request, I'll now add diagrams for NEXTs too.

After our hit, roadkillbooks rolls 6-5 (first diagram) and enters B/20 and return-hits with 24/18*. We now roll a 2-1 from the bar (second diagram):

jGfwICGMZ2HAQg:cImoACAAAAAA

[diane]

Spoiler

Bar-24, 9-7 for me, gets that one of the bar and returns the hit without leaving any extra blots.  Can't think I would want to do anything else.  In some circumstances, the 6-5 hit would be appealing, but not with all the blots we already have back.

[close]

[Zorba]

Spoiler

This is becoming a very interesting and complex game, with our four backcheckers against rkb's three. The pipcounts are high and pretty close. That means it's not time to think about any backgame yet, for either side. The battle ahead might be long and mostly positional, aiming at advanced anchors, outfield control and creating blocking points, hoping for that ugly roll for opponent or great doubles for us, or a hit followed by a dance or ugly entering roll, in order to make real progress. It's worth spending enough time to think about choices in these games.

I eliminate two moves from the choices: B/23 8/7* is just unnecessary risky, breaking the useful eightpoint and leaving three blots and lots of shots. B/23 9/8 gives rkb a free hand to make either the great 20pt anchor, the strong barpoint or perhaps something else that's good. This move is way too safe. We have the 20pt anchor ourselves, so we can afford quite a lot of risk. Also, with four checkers back already and rkb's unthreatening board, getting hit is not so bad anymore. Structure is more important than pipcounts for now, with a long battle ahead.

The choice is between the "obvious" B/24 9/7* and the riskier, but also more interesting alternative B/23 6/5*. First, what about the difference between entering on 24 or 23? I think it doesn't matter much here. We already have the 20pt anchor, and the other backchecker on 21, so flexibility looks good after either move. Should we get hit upfront and get another backchecker, then making the double anchor on either 24 or 23 is not such a great combo with the 20pt, with 23 probably slgihtly better. Making the 21pt would be much stronger, denying rkb the chance to make the two points he wants to make most. Anyhow, I consider the difference between B/24 and B/23 mostly irrelevant here.

So, the choice is about the hit upfront. Both plays leave quite a lot of returnshots, but 6/5* should leave more (I might count them later!). Also, 6/5* creates a new blot, whereas 9/7* just moves the blot. Clearly, 6/5* is the riskier move. So does it have enough advantages over 9/7* to make it worth it? Yes, I think so. 6/5* slots the point we really want. It also knocks rkb off one of the points he really wants. It unstacks our sixpoint, instead of just moving a builder. It gives us a better structure upfront, with much better covers and attack potential next turn: A strippped midpoint, a builder on 9, the eightpoint, a spare on the sixpoint and slot on 5 is more powerful and flexible than the stripped midpoint and eightpoint with a builder on 7 and the sixpoint spares with no real place to go yet. If the barpoint slot survives, we have no comfortable way to cover it; if the fivepoint slot survives it looks much better. Not to mention that the fivepoint is much stronger than the barpoint here, also because of the made threepoint (we really want the points in between 6 and 3).

Then there's also the matchscore to consider, which basically means we like more gammonish games. Gammons won are more valuable than usual, gammons lost cost less than usual. The fivepoint slot is riskier and could lead to more men back and thus cost more gammons, however, going for the fivepoint ourselves and trying to knock rkb off it increases our own gammon wins. So, 6/5* should suit the matchscore.

At a normal score, it looks like a close decision to me; the many advantages of 6/5* are counbtered by the clearly increased risk. At this matchscore, 6/5* should be clearly better then. Doubled gammons, here we come (but it will be a long journey from here!  )

[close]

[ah_clem]

Spoiler

These are the positions I ofter play poorly.  That said, here's my best shot:

With more men back, behind in the race, and behind in the match, "safe" plays like bar/23 9/8  or bar/23 21/20 are the wrong idea.  We need to return hits.  But where to hit?

I don't think it matters much whether we enter on the 24 or 23, so let's pick our hitting play first and then put the bar checker wherever it needs to go.  The candidates are:

8/7*: very agressive and play well if rkb fans.  But he probably won't fan (1 in 9 chance), and probably will hit us if he enters.  I'm not adverse to leaving a blot or three, but 5 is excessive in this case.

6/5*: slots the 5 point, which is good if he misses, but he probably won't miss.  And four blots seems a lot....

9/7*:  only leaves three blots, doesn't expose a blot to a direct hit from the bar, and slots the bar point.  This looks like the best move to me.  So

bar/24 9/7*

[close]

[Zorba]

GnuBG 2-ply rollout:

Spoiler

The first serious error by the Forum in quite a few moves. 6/5* wins more games and gammons. It loses about 2.5% gammons more too, but given the matchscore situation those aren't so important here. At a "normal" matchscore, the plays would be quite close probably but it looks like 6/5* would still be best then.

    1. Rollout          bar/23 6/5*                  Eq.:  +0,1527
        50,06  15,92   0,78 -  49,94  17,99   2,32 CL  +0,0231 CF  +0,1527
      [  0,14   0,13   0,03 -   0,14   0,20   0,20 CL   0,0047 CF   0,0089]
    2. Rollout          bar/24 9/7*                  Eq.:  +0,0881 ( -0,0646)
        49,14  13,99   0,72 -  50,86  15,43   1,80 CL  +0,0224 CF  +0,0881
      [  0,13   0,13   0,06 -   0,13   0,21   0,17 CL   0,0044 CF   0,0082]

        Full cubeful rollout with var.redn.
        2592 games, Mersenne Twister dice gen. with seed 860183004 and quasi-random dice
        Play:  2-ply cubeful prune [world class]
        keep the first 0 0-ply moves and up to 8 more moves within equity 0,12
        Skip pruning for 1-ply moves.
        Cube: 2-ply cubeful prune [world class]

[close]