Position # 52

[PersianLord]

Position information:

Pip-count: Red 134-96 White
Score:      Red    0-0   White
Match:                5-pointer

White on roll. Proper cube decision?

[PersianLord]

Spoiler

Position ID: bNuCAWC2bS0AIA , Match ID: MAGgAAAAAAAA

[close]

[sixty_something]

what did GNUbg say about this position?

[PersianLord]

what did GNUbg say about this position?

Spoiler

Cube analysis
2-ply cubeless equity  +0.521 (Money:  +0.535)
   55.8%  45.6%   1.8% -  44.2%   5.5%   0.1%
Cubeful equities:
1. Double, take         +0.750
2. Double, pass         +1.000  ( +0.250)
3. No double            +0.572  ( -0.178)
Proper cube action: Double, take

[close]

[dorbel]

Spoiler

Pl has found an interesting position for us, from the class known as prime/prime. These are famous for being difficult to play, both with checkers and cube. I would fully expect a cross section of decent players to offer a range of opinions as White, from not being good enough to double right up to too good. How should you approach this over the table? As ever, Woolsey's rule can be your guide. Are you sure from Red's point of view that this is a correct take? If not, it must be a double. I don't see how Red can be certain that he should take here, so it's good enough to cube. Is it too good? Now you can apply dorbel's rule. If you're sure that it's a double but not certain that it is too good, turn the cube. Sometimes you will lose a bit of equity when they correctly pass but you will put a point on the scoresheet. Occasionally however, they take and then your equity goes through the roof! The main rationale for this plan of action is that doubling gives your opponent the chance to make a mistake, whereas leaving the cube in the middle denies him that opportunity.
This process leads us to conclude that White has a solid double. Should Red take? Personally, I would be glad to give this one up and would pass fairly quickly. Gnu 2-ply thinks that the take is quite easy, but as we have seen so often before, it is wrong. Snowie gets this one right calling it a solid double/pass and rollouts with Gnu and Snowie agree with that. White is on course to win this about 61.5% including about 49.5 gammons and of course a couple of backgammons as the icing on the cake. Even with his 38.5% wins and a few gammons of his own, Red should pass.
Some points to note. Owning the cube is worth a lot to Red here and that is typical of prime/prime positions as a whole. This is because primes are immensely strong but capable of crashing in a single roll. To give a single non-trivial example here, after White double/Red take, if White rolls 6-6, Red can redouble to 4. It's a take for White, actually quite an easy take, but it's not what you want to see! Cube ownership isn't enough to make this a take for Red, but always keep it in mind in these positions.
White also needs to know how to play this with the checkers to get the most out of it. Her first job is to kill sixes in order to keep the board as long as possible and allow time for her straggler to step up an out. Thus if White doubles and Red, a fan of the big turnround takes, a 3-2 for example should be played 8/5, 8/6. Any other play is wrong and anything with 9/7 in it is a humungous blunder. If Red knows enough about White's play to know that she is capable of making a mistake like this, then that might be enough to justify a take! You can also talk yourself into taking if you know that White handles 4 cubes badly, either being prone to weak passes of a recube or alternatively being a rash taker of cubes that should be dropped. Note that at this score you need about 3% more than usual to take a 4 cube, around 25% as an average rather than the more normal 22%. These figures assume no gammons by the way.
So to sum up, as White ship it in and let Red do the worrying. As Red, give it up and move on to game 2.

[close]

[PersianLord]

Spoiler

Pl has found an interesting position for us, from the class known as prime/prime. These are famous for being difficult to play, both with checkers and cube. I would fully expect a cross section of decent players to offer a range of opinions as White, from not being good enough to double right up to too good. How should you approach this over the table? As ever, Woolsey's rule can be your guide. Are you sure from Red's point of view that this is a correct take? If not, it must be a double. I don't see how Red can be certain that he should take here, so it's good enough to cube. Is it too good? Now you can apply dorbel's rule. If you're sure that it's a double but not certain that it is too good, turn the cube. Sometimes you will lose a bit of equity when they correctly pass but you will put a point on the scoresheet. Occasionally however, they take and then your equity goes through the roof! The main rationale for this plan of action is that doubling gives your opponent the chance to make a mistake, whereas leaving the cube in the middle denies him that opportunity.
This process leads us to conclude that White has a solid double. Should Red take? Personally, I would be glad to give this one up and would pass fairly quickly. Gnu 2-ply thinks that the take is quite easy, but as we have seen so often before, it is wrong. Snowie gets this one right calling it a solid double/pass and rollouts with Gnu and Snowie agree with that. White is on course to win this about 61.5% including about 49.5 gammons and of course a couple of backgammons as the icing on the cake. Even with his 38.5% wins and a few gammons of his own, Red should pass.
Some points to note. Owning the cube is worth a lot to Red here and that is typical of prime/prime positions as a whole. This is because primes are immensely strong but capable of crashing in a single roll. To give a single non-trivial example here, after White double/Red take, if White rolls 6-6, Red can redouble to 4. It's a take for White, actually quite an easy take, but it's not what you want to see! Cube ownership isn't enough to make this a take for Red, but always keep it in mind in these positions.
White also needs to know how to play this with the checkers to get the most out of it. Her first job is to kill sixes in order to keep the board as long as possible and allow time for her straggler to step up an out. Thus if White doubles and Red, a fan of the big turnround takes, a 3-2 for example should be played 8/5, 8/6. Any other play is wrong and anything with 9/7 in it is a humungous blunder. If Red knows enough about White's play to know that she is capable of making a mistake like this, then that might be enough to justify a take! You can also talk yourself into taking if you know that White handles 4 cubes badly, either being prone to weak passes of a recube or alternatively being a rash taker of cubes that should be dropped. Note that at this score you need about 3% more than usual to take a 4 cube, around 25% as an average rather than the more normal 22%. These figures assume no gammons by the way.
So to sum up, as White ship it in and let Red do the worrying. As Red, give it up and move on to game 2.

[close]

Spoiler

Thanks for the through analysis. Would be nice to provide full Snowie analysis as well. (I was quite surprised to know that Snowie thinks this is a pass)

[close]

[blitzxz]

Now you can apply dorbel's rule. If you're sure that it's a double but not certain that it is too good, turn the cube. Sometimes you will lose a bit of equity when they correctly pass but you will put a point on the scoresheet. Occasionally however, they take and then your equity goes through the roof!

I disagree with this complitely. It's sure way to lose equity and the practical reasoning that you give opponent change to error is also wrong unless you know for sure that the opponent will make huge cube errors. If you continue to play too good position, you will very often end up in position that is double/pass or double/take and then of course you can give correctly the change for error without losing any equity yourself. In my opinion it pays lot more (in practice also) to double later in a position which is trickier or sometimes you just simply gammon the opponent.

[dorbel]

Yes Blitz, I see your point. If you are fairly sure that it is too good then I agree you should back your judgment and go for the gammon. If you are not sure, as in this position for example, then turning the cube is a good practical option. The fact that you are not sure means it should be fairly close, so the amount of equity lost when it is technically too good and you get a pass should be fairly small. The occasional incorrect take should cover that. You don't need to know that your opponent is certain to make a blunder, only that there is a chance of it. Also, this is very important, the point that you gain when you cash a "slightly too good" position is a concrete point on the scoresheet. You don't have to use any skill to get it and it is there, written down. The equity from correctly playing on is for the moment only equity, theoretical and up for grabs. You have to play to get it and sometimes you have to exercise a lot of skill or it won't be worth what you think it is. In my opinion, in practical play, it is often worth paying a small price for a real gain as oposed to a theoretical one. Think of it as selling your house, the market value is $250,000, but a cash buyer offers you $235,000. You may prefer to have that in your hand rather than risking a fall in the market and not being able to get your theoretical price in the future.
As to giving your opponent the chance to make a mistake, of course you can only do this once. Picking your moment is a key skill. You say "If you continue to play a 'too good' position, you will very often end up in position that is double/pass or double/take and then of course you can give correctly the chance for error without losing any equity yourself." What you don't say is that you can also find yourself very suddenly not good enough to double at all! Thay can happen here on the very next roll for example. This isn't a too good position of course, but the reason for doubling now is partly that you will lose your market with a one and partly to avoid the horror sequence where you roll a big crashing double and are suddenly not able to double at all.

Sorry PL, I didn't save the Snowie analysis, but the approximate figures are in my text. Thanks for your kind words.

[playBunny]

Spoiler

As ever, Woolsey's rule can be your guide. Are you sure from Red's point of view that this is a correct take? If not, it must be a double. I don't see how Red can be certain that he should take here, so it's good enough to cube.

[close]

Spoiler

I thought it was a huge take. Obviously the 1-6 as a unit would be wonderful for White but it's the dream rather than the expectation. The awkward-to-disastrous large doubles have more chance than that nice roll. I don't think that White has an awful lot of timing in which to get a 1 and then a 6 so pretty soon it would be crunch time.

I ummed and aahed between Double/take and No double/take and finally stabbed the No Double, my usual caution gaining the upperhand again.

[close]

GnuBg 4-ply analysis agrees with my thinking about the opponent's response to a cube:

Spoiler

4-ply cubeless equity  +0.517 (Money:  +0.537)

Cubeful equities:
1. Double, take         +0.752
2. Double, pass         +1.000   (+0.248)
3. No double             +0.546    (-0.206)

Proper cube action: Double, take

[close]

Which is great - except that the 2-ply rollout tells me not to be such a fool!

Spoiler

Cube analysis, 1296 games

2-ply Rollout cubeless equity  +0.689 (Money:  +0.676)

Cubeful equities:
1. Double, pass         +1.000
2. Double, take         +1.220  ( +0.220)
3. No double            +0.743  ( -0.257)

Proper cube action: Double, pass

[close]

[PersianLord]

I agree with dorbel's rule, specially when playing against stronger opponents. When you offer a cube in a position in which taking is a slight error, there will be 2 scenarios:

1- The opponent passes: in this case, as dorbel said, you lose equity but not without any reward. You gain a score equal to the current cube value and that will add to your current score. As the old Persian saying advices us: "Serkeye naghd beh az halvaye nesye" ( The available vinegar is better than the promised sweetmeat ). We also should take into account the fact we're playing at FIBS with it's truely rigged dice  .

2- The opponent takes: in this case, you gain tremendously. Apart from equity gain, it's a known sterategical advise that if you're the weaker player, try to shorten the matches as far as possible.

But against weaker players, I would say playing on might prove more profitable.

Regards

[blitzxz]

In my opinion, in practical play, it is often worth paying a small price for a real gain as oposed to a theoretical one. Think of it as selling your house, the market value is $250,000, but a cash buyer offers you $235,000. You may prefer to have that in your hand rather than risking a fall in the market and not being able to get your theoretical price in the future.

My first rule is:

When gambling you should always maximaze the expected value no matter what the risk or volatility is otherwice you are simple giving chances (or money) away.

Of course I mean here practical equity not theoretical bot equity. (So you should cash marginal too good if all your moves are lot harder then opponents. However if you're better player you should especially play too goods to the end because it makes games longer and gives more chances to error for opponent.) But in general you take the risk away simple by playing over and over again and in the end the expected will become the real value no matter what the volatility or risk is. If you can't play over and over again or can't stand the risk it means you're playing too high stakes (like for your house).

In real life however you sometimes have to take too big one time risks so you want to minimaze it and end up paying the risk premium for some one else (like insurance company) who has the big bucks to play it over and over again and take the full expected value.

So sometimes too good will turn out to be not double at all. That happens. But the equity for the long run is still there and that's what you should always go for.

[dorbel]

Absolutely correct Blitz, if you KNOW that you are too good. My advice applies to situations where you think that you might be too good, but not certain.

[lewscannon]

I disagree with this complitely. It's sure way to lose equity and the practical reasoning that you give opponent change to error is also wrong unless you know for sure that the opponent will make huge cube errors. If you continue to play too good position, you will very often end up in position that is double/pass or double/take and then of course you can give correctly the change for error without losing any equity yourself. In my opinion it pays lot more (in practice also) to double later in a position which is trickier or sometimes you just simply gammon the opponent.

I'm with Dorbel on this argument in that in a close situation, give your opponent a chance to screw it up, of course depending on the quality of your opponent. I happen to think this is a double/take, but that's based on a hunch, as are a lot of my decisions of this kind. If I'm red, and white doesn't get out, I've got a good chance of getting 4 points out of this. If I'm white, I'm happy with the point if red declines the cube.

lews

[blitzxz]

Absolutely correct Blitz, if you KNOW that you are too good. My advice applies to situations where you think that you might be too good, but not certain.

To my experience if you think you might be too good you're propably clearly too good already. Doubling when it is too good used to be one my most common errors and it's a common error for my opponents too so I wouldn't give such advice.