Match 1, Game 3, Move 5 Forum

[blitzxz]

6-2?

Score:
Green(forum) 2, White(factotum) 0, match to 5

Pips:
Green 149 White 176

GnuID:
4HnwACmM58gBIg:QQmrAAAAEAAA

[diane]

Spoiler

I have opted for an ultra safe, hardly exciting at all 13-11, 21-15.  My reasons are that I cannot see a way to capitalise on the situation we have here - 3 loose blots - without putting ourselves at huge risk.  Since we have recklessly tossed that cube across the board , I don't want to do anything that could result in it coming back - or worse still.....The move I suggest is a way of moving forward without doing anything too risky - let him make the next move or two, and lets see if we can get a good solid safe hit, before he gets properly back on his feet.

[close]

[playBunny]

Spoiler

Mine's the same as Diane, 21/15 13/11. 6/4* is out of the question for me so the alternatives are 24/18 13/11 and 21/13. The latter is tidy as far as the backrunners go but leaves that blot on 11 and it's got three hungry wolves approaching. 13/11 turns it from a victim into an 11-point backstop, which I like. Running 21/15 duplicates 3s in a minor way - a 3 could be used by factotum to anchor on our 4-point. More than that I like that 21/15 comes out far enough to threaten factotum's backrunners, should one take a leap into the void but fall short of his midpoint.

[close]

[lewscannon]

Spoiler

That I'm following Diane's lead causes me to doubt this decision, but I think it's the best move. It keeps the possibility of making the 5 point alive. The other option seems to be the 21-13, but I don't like that particularly, as a 7, 9 or 10 picks off the blot on the 11 and shifts momentum. Granted, if he somehow hits and makes his 21 point, this doesn't look as good, but there are only 4 rolls that can do that (4-2,2-4,4-4,1-1). He might pick up a 4-3 or 3-2, hit both blots and buy him a roll to make a point in our home board. Odds are he's going to make a point somewhere in our home board anyway with the next roll.

[close]

[ah_clem]

Spoiler

A tough one here, and I often mis-play these kinds of positions.  And after screwing the pooch on the cube decision, I'm reluctant to be very declaritive here.  That said, here goes:

The main decision is safe play vs bold play.  Applying Magriel's criteria we have:

1) we're ahead in the race -> safe
2) we have fewer men back -> safe
3) our opponent has no blots in his board - >safe
4) we have no blots in our board -> bold
5) our opponent has only one point in his board -> bold
6) we have two points in our board -> meh?

So, three criteria favor safe play, and two favor bold play and one is bold-ish.

Observe that we're not committed to the blitz here.  A quiet play like making the 11 and running a backman leaves us ahead by 35 pips.  Gammons are only worth 3/4ths of what they normally are due to the match score.  A simple win takes us to crawford with our opponent having zip.  So - is there any real point in playing agressively with a loose hit?

Arguing conversely, an exchange of hits would tend to favor us, but it's fairly volatile and we really don't want to eat a re-cube.

So, my hunch is to play it safe here.  If we're going to hit loose, I think the right play is to go all the way and hit both.  At 0-0 I'd say hit twice, but with a 35 pip lead and a 2-0 match score it's better to play it safe.  21-15 13-11 seems best - it makes another point vs the more passive 21-15.

So that's my opinion - but I'm going to see what others have to say before voting.

[close]

[stog]

Spoiler

mmm i went for it...... the hit 13,7 6,4*  (maybe accounts for my rating but i am wary that he will get an advanced anchor, and although i see what you are all saying, i think we can still go for a blitzzzzzz

[close]

[socksey]

Spoiler

Since my first choice is not up there, 11/3, I will go with 21/13. 

[close]

socksey



"We must all hang together, or assuredly, we shall all hang separately." - Benjamin Franklin

[diane]

That I'm following Diane's lead causes me to doubt this decision

Admit it, you are just a big copy cat 

[NIHILIST]

Spoiler

I favor hitting 2 checkers here for several reasons. I don't want him to consolidate his position, putting 2 of his checkers on the bar here is huge even if it does leave 2 of our own blots exposed, even though a gammon might be worth less than 100% it still wins the match for us.

I've often said there are 2 emotions that rule all bg players, FEAR and GREED. This is a pure greed play. I think aggressive action is called for while he has no board. The potential payoff is too big to ignore.

[close]

Bob

[diane]

Spoiler

I favor hitting 2 checkers here for several reasons. I don't want him to consolidate his position, putting 2 of his checkers on the bar here is huge even if it does leave 2 of our own blots exposed, even though a gammon might be worth less than 100% it still wins the match for us.

I've often said there are 2 emotions that rule all bg players, FEAR and GREED. This is a pure greed play. I think aggressive action is called for while he has no board. The potential payoff is too big to ignore.

[close]

Bob

I think it is the 4-4, 2-2 return rolls that put me off doing that - suddenly we would be in a big pickle, with our opponent holding a pretty crucial cube.

[factotum]

wow, blitzxz gave you 10 (ten) choices!    you'd better pick the right one....

[NIHILIST]

Spoiler

How about the 4 rolls that fan completely ? And rolls that enter only 1 checker on the 1 or 5 point without hitting anything ?

[close]

Bob

[Zorba]

Specially for socksey, I added an 11th move to the list, 11/3  Hope that's okay. Everyone can still change their vote!

[ah_clem]

Specially for socksey, I added an 11th move to the list, 11/3  Hope that's okay. Everyone can still change their vote!

I'm not changing my vote until I hear what you have to say.  So what gives?

[dorbel]

Spoiler

The team doubled on their gammon threat so they may as well go for it. The correct way to utilise an advantage in home board points is to hit and having hit on the 4pt, you may as well put a second checker in the air IMO.
Diane is worried about the jokers 4-4 and 2-2 after this play (5.5%) but these are excellent rolls for White in any event. Besides, that 5.5% chance is more than balanced by the 11% chance that White will fan! I don't think that playing "safe" here is actually safe. You must always ask your self, "What does my opponent want me to do?" Then do the opposite. As White here, I would so like to be left alone to make my next roll in peace, so hit twice! It's actually safer than hitting once.
Two on the bar, better by far.[

[close]

[diane]

Spoiler

How about the 4 rolls that fan completely ? And rolls that enter only 1 checker on the 1 or 5 point without hitting anything ?

[close]

Bob

Perhaps I have been on fibs too long...my opponent never rolls those 

[ah_clem]

Spoiler

The team doubled on their gammon threat so they may as well go for it. The correct way to utilise an advantage in home board points is to hit and having hit on the 4pt, you may as well put a second checker in the air IMO.
I don't think that playing "safe" here is actually safe.

Ok.  I'm going to change my vote.   Hitting twice has a lot to recommend it, I'm just not sure it's really sound.  If you think it's sound, I'll go with that.

Reasons for hitting twice:

First, it's thematic - we've doubled and started a blitz.  Hitting twice is the logical continuance.

Next, except for 22 44 and 55, hitting twice makes it impossible for white to make an advanced anchor.
OTOH, if we play quietly, any 2 any 3 plus 41 makes an advanced anchor.  63 65 and 53 allow him to make the bar point.  That makes him an overwhelming favorite to consolidate his position and enter a holding game with good timing.  Letting him collect himself presents a bigger danger than hitting one of our inner board blots.

He's a 55% chance to leave a checker on the bar, so even if he hits one of our blots we'll get another shot at him.   If he enters both, they'll likely still be a blot or two to attack. 

Since we own two inner board points and he only owns one,  an exchange of hits favors us. Also, the double hit slots two checkers, threatening to make a third inner board point.  Yes, it's likely that one of our blots will be hit, but white has no board.

In short, he's down - hit him again and don't give him a chance to get up and fight back.

[close]

[diane]

Spoiler

The double hit I would do without a second thought, if we hadn't cubed.  The joker rolls then have soooo much more impact.  I guess this is fear playing in my case - but that is the way it feels to me.  That cube made me feel more cautious - we don't really need to gammon enough to risk being gammoned.

[close]

[Zorba]

Spoiler

I won't be of any help here, as I'm unsure about what to do. Two pretty big errors earlier in this game (double hit before the double was wrong, and the double itself was wrong, both heavily influenced by the score) made me reconsider things at this matchscore. At a balanced score (money) I'd play 6/4* 8/2* quickly, as it will win a lot more gammons. It will probably lose more gammons too though, and this is a bad trade-off at the score. Still, the position mostly argues for bold play, even if the matchscore doesn't.

The only alternative I can see is 21/15 13/11, making the elevenpoint and a good escape attempt, basically hoping to bring the position home against a very likely anchor, profiting from our solid racing advantage. It also duplicates a lot of factotum's threes to hit or make an advanced anchor. How likely is it that the race plan will succeed? We have the threepoint and elevenpoint for extra structure, and likely just one backchecker left to escape with hardly any obstacle. But we'd still face clearing the mid and outfield points against an anchor, most likely.

I voted for this last safer and less gammonish play because of the matchscore, but I'm ready to learn. I could very well see the double hit as correct, maybe even by a good margin. Preventing factotum from consolidating his position is a strong argument for the double whammy.

N.B. In short, 21/13 looks wrong as it still leaves a lot of shots and doesn't make the 11pt, a really nice asset here, as it's aiming at our open fivepoint which we'd really like to have. Should factotum make our fivepoint, then the 11pt nicely blocks his sixes from there. It can also help us as a landing spot for checkers coming around. 6/4* without 8/2* leaves a TRIPLE direct shot from the bar for not much good. Hitting the second blot actually makes it safer, as it's now just a double direct shot, with of course the added benefits of putting two on the bar instead of one. 11/3 moves in a checker too deep, too early. It may not look so bad at first, but this kind of play usually hurts you for the rest of the game. The checker on 11 can be used to make a lot of very useful points, it's in a great position. Moving it to 3, it becomes a nearly useless checker that can only be moved with ones or twos, and even then mayt not be too useful. It's almost like playing with just 14 checkers for the rest of the game, and that is not a good thing so early in the game.

Now I'll go check the bot and also find out what our real gammon values ("prices") are at this score and cube level, and how that may influence this decision (or not)!

[close]

[NIHILIST]

Checking the bot shouldn't influence your decision since you can't use bots in real play. I'm now beginning to wonder if those disturbing rumors of how you got back above 1900 are true.




Bob

[Zorba]

I won't change my vote in any case, I'd rather have my mistakes and blunders recorded for history!

[lewscannon]

Spoiler

I voted otherwise, but the arguments here have convinced me that the double hit is a good idea. If this goes wrong, however, I'll be blaming everyone else.

[close]

[stog]

yes i'm willing to change my vote too

Spoiler

4 the double hit!

[close]

if we are quick i wonder if we can change the outcome of the vote - do u need a moderator to change your vote lews? or others if u need us to change your vote - post your request quickly!

[playBunny]

Spoiler

I'd probably do the double hit if I was feeling like some reckless fun and it was a longer match and if my opponent wasn't formidable factotum. ;-)

There are things that I don't like about the double hit here. One is that one of the hits is deep and another is that we strip our midpoint in order to do so. Next roll we'd have only one builder to capitalise on the potential gain (although a 2 would give us 4/2 and a deep home point, whoopee) and might have to bust the 8-point. The 2-point hit is almost sacrificial but I don't think that it gains enough. Another point against hitting is that if we get double-tapped in return, we're looking to have our 11-point blot zapped while we scramble to our feet over in our opponent's home. I know I'm focusing only on the negatives here but they do seem fairly substantial to me.

[close]

[ah_clem]

do u need a moderator to change your vote lews? or others if u need us to change your vote - post your request quickly!

There's a "Remove Vote" link that works.  After you've removed your vote, you can vote again.  No need to bother the mods. Thanks to blitzxz for implementing that functionality.

[Zorba]

After playBunny's analysis, I suggest everybody changes their votes once more  

[ah_clem]

Don't tempt me.  I just might change my mind again!  (c:

Anyway, I see that the votes for safe plays are spread across four different moves while the bold voters are stacked up on one.  If anybody's going to change votes, I'd suggest that the "safe" voters consolidate their support behind one move.  Or go over to the dark side  (c:

[diane]

Spoiler

I'd probably do the double hit if I was feeling like some reckless fun and it was a longer match and if my opponent wasn't formidable factotum. ;-)

There are things that I don't like about the double hit here. One is that one of the hits is deep and another is that we strip our midpoint in order to do so. Next roll we'd have only one builder to capitalise on the potential gain (although a 2 would give us 4/2 and a deep home point, whoopee) and might have to bust the 8-point. The 2-point hit is almost sacrificial but I don't think that it gains enough. Another point against hitting is that if we get double-tapped in return, we're looking to have our 11-point blot zapped while we scramble to our feet over in our opponent's home. I know I'm focusing only on the negatives here but they do seem fairly substantial to me.

[close]

This is very much what I am feeling here - and couldn't see what dorbel meant by his comment that 'either of the two jokers were excellent rolls for white'...if he could get back in time to explain that to me, I might be more persuaded.

[dorbel]

If White is not on the bar, he has no bad rolls and 4-4 and 2-2, the rolls that you fear both play very well. He'll probably make the 9pt and 4pt with 4-4, while 2-2 anchors on the 21pt and makes the 4pt. It's an interesting paradox that the thing that you most fear is being on the bar, yet you are reluctant to put your opponent there!
I don't actually think that 21/15, 13/11 is a bad play by the way. For all I know it may be best! If that's what you like go for it!

[diane]

Spoiler

It isn't a fear of being put on the bar as such - for us to put him on the bar, we have to leave two blots - if he rolls either of those two jokers, he puts two of our men on the bar against a two point board and leaves us nothing - with the option on taking our third one, as we waste our roll coming off the bar [if we dont fan]. It is just how bad this potentially is, whilst he holds the cube, that turns me away from it.  I also really don't like giving him time to consolidate his position - but I also think we are ahead and safeish, and that we don't need the hits to take these two points.  I think I will stick with my current vote.

[close]

[lewscannon]

Spoiler

It isn't a fear of being put on the bar as such - for us to put him on the bar, we have to leave two blots - if he rolls either of those two jokers, he puts two of our men on the bar against a two point board and leaves us nothing - with the option on taking our third one, as we waste our roll coming off the bar [if we dont fan]. It is just how bad this potentially is, whilst he holds the cube, that turns me away from it.  I also really don't like giving him time to consolidate his position - but I also think we are ahead and safeish, and that we don't need the hits to take these two points.  I think I will stick with my current vote.

[close]

It looks like Diane's league matches against me have psychologically scarred her.

[ah_clem]

Rollout:

Spoiler

The safe play is the best, but not by alot.  As I was watching the rollout progress, the double-hit went into the lead several times, even though the safe play won in the long run. And the double-hit was the winner in the evaluation.  No wonder so many of us had a hard time with this one.


    1. Rollout          21/15 13/11                  MWC:  52.05%
       0.595 0.177 0.009 - 0.405 0.099 0.003 CL  54.50% CF  52.05%
      [0.001 0.001 0.000 - 0.001 0.000 0.000 CL   0.04% CF   0.04%]
        Truncated cubeful rollout (depth 11) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 858129907 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    2. Rollout          8/2* 6/4*                    MWC:  51.77% ( -0.28%)
       0.564 0.236 0.015 - 0.436 0.121 0.007 CL  54.52% CF  51.77%
      [0.001 0.001 0.000 - 0.001 0.001 0.000 CL   0.05% CF   0.04%]
        Truncated cubeful rollout (depth 11) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 858129907 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    3. Rollout          21/13                        MWC:  51.47% ( -0.58%)
       0.587 0.161 0.007 - 0.413 0.095 0.003 CL  53.99% CF  51.47%
      [0.001 0.001 0.000 - 0.001 0.000 0.000 CL   0.03% CF   0.04%]
        Truncated cubeful rollout (depth 11) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 858129907 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    4. Rollout          21/15 6/4*                   MWC:  51.01% ( -1.04%)
       0.565 0.192 0.011 - 0.435 0.115 0.005 CL  53.69% CF  51.01%
      [0.001 0.001 0.000 - 0.001 0.000 0.000 CL   0.04% CF   0.04%]
        Truncated cubeful rollout (depth 11) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 858129907 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    5. Rollout          24/18 13/11                  MWC:  50.99% ( -1.06%)
       0.576 0.161 0.009 - 0.424 0.104 0.004 CL  53.58% CF  50.99%
      [0.001 0.001 0.000 - 0.001 0.000 0.000 CL   0.04% CF   0.04%]
        Truncated cubeful rollout (depth 11) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 858129907 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]

[close]

[playBunny]

 
Rollout 1296 trials, 2-ply for 12 plies, 0-ply thereafter, 2-ply cube throughout

Spoiler

# .. Move .......... Equity . (Diff) ... Win .... WinG ... WinBg -- Lose ... LoseG . LoseBg
1 .. 21/15 13/11 ... -0.110 ............ 59.7% .. 25.4% ... 4.7% -- 40.3% ... 9.9% ... 1.0%
2 .. 8/2* 6/4* ..... -0.148 (-0.038) ... 58.0% .. 28.3% ... 6.9% -- 42.0% .. 11.1% ... 2.1%
3 .. 21/13 ......... -0.162 (-0.052) ... 58.2% .. 23.1% ... 2.7% -- 41.8% ... 9.5% ... 0.9%
4 .. 24/16 ......... -0.188 (-0.079) ... 57.6% .. 22.8% ... 3.9% -- 42.4% .. 10.2% ... 1.0%
5 .. 24/18 13/11 ... -0.214 (-0.104) ... 56.9% .. 24.6% ... 5.6% -- 43.1% .. 10.8% ... 1.1%
Std Err ............. 0.008 ............. 0.2% ... 0.5% ... 0.4% --- 0.2% ... 0.1% ... 0.2%

Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 858261885 and quasi-random dice
Play: supremo 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 16 more moves within equity 0.32
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]
Different evaluations after 12 plies:
Play: 0-ply cubeful prune [expert]
Cube: 2-ply cubeful prune [world class]

[close]

[dorbel]

Interestingly, a 2-ply cubeless Snowie rollout (equivalent to Gnu 1-ply), supported the double hit that Snowie 3-ply will make. The win rates were broadly the same but Snowie won a lot less gammons (than Gnu) with either play. However, for those who didn't want to play the double hit because they thought that it lost too many gammons, as you can see here, Snowie agrees that the gammon losses are more or less equal.
I will do another rollout using cubeful 3-ply, but it is worth noting that Gnu is usually reckoned to be better for this sort of position, because of its ability to rollout according to score.

[dorbel]

And rollout fans may be interested to know that a long 3-ply cubeful rollout gave the same result.
The double hit won 57.3 games with 23.4 gammons, 21/15, 13/11 won 59.5 games with 19.7 gammons. Both plays lost 10 and a bit gammons. This gives 8/2*, 6/4* a 0.033 advantage cubeless, but only 0.01 cubeful, both results lying well within the margin of error in any case.
It isn't clear to me why Gnu wins so many more gammons. I would be slightly more inclined to believe Gnu's results from experience, but I have no evidence for that. Perhaps the two plays are about equal.

[Zorba]

My 2-ply GnuBG rollout is still running, but lower settings rollouts and evaluations all pretty much agreed so far that the double hit is about a 0.06 error. I'll post it later.

BTW, ah_clem's rollout looks like it was done with the wrong score or cube level, our MWC should be much higher, also, on my GnuBG all evaluations prefer 21/15 13/11 by a good margin.

That Snowie cubeless has the plays close makes sense, I'm sure GnuBG would get a similar result. The difference is the matchscore, where our gammons don't have full value on the 2-cube, and zero value once the cube reaches 4 (with factotum's gammons always valuable). Since factotum should be quick on the redouble here (again because of the score), relatively many games will get played on a 4-cube.

One that cube reaches 4, factotum can afford to stay back with  his backcheckers forever. That's why we will win many more gammons when this is played out according to score, with a live cube, as GnuBG plays it. It's the kind of gammons that are not useful for us, so that explains the apparent paradox: Gnubg wins many more gammons after the double hit, but it doesn't show up in the cubeful equity, more like the opposite. It's because once gammons don't count, we will win many more of them.

The effect is even clearer when looking at backgammons, that already have lost any value for us on a 2-cube. We win much more of them, because of that.