[BLML] once again, UI from allowed source

Steve Willner swillner at nhcc.net
Fri Aug 30 03:37:01 CEST 2013


On 2013-08-28 11:28 PM, Jerry Fusselman wrote:
> If you don't mind, I'm going to focus just on your statement that, if
> I may paraphrase, says that partner always has a tempo that provides
> no UI to you about his thinking.

That's fair.  I'll back off to the extent it's the bridge version of 
"always," not the mathematical version.

I'll snip most of the argument.  I understand it despite my mathematical 
deficiencies and agree it is logically sound.

> You know that if he is thinking that 3NT is natural, then he will bid
> it somewhere between 4 and 5 seconds. Let's use the simplest
> distribution and assume that s|N (read this as "s given N") is U[4,5],
> i.e., uniform on the interval [4,5], and the density is 1 there.
>
> You also know that if he is thinking that 3NT is artificial, then he
> will bid his 3NT somewhere between 4 and 10 seconds, and here we'll
> assume that s|A is U[4,10], with a density 1/6 on [4,10].

Where I disagree is that in real life, I won't know either of these 
things.  Even if partner is sure 3NT is natural, it might be far from an 
obvious bid (shaky stopper, perhaps), and he might easily take 10 s or 
30 or more before coming up with it.  And if 3NT is artificial, he might 
take some time to remember that, but a couple of seconds (quite a lot in 
bridge terms) is likely to be enough.  If it is artificial, it shouldn't 
take too long to decide whether it's the right bid.  That is to say, the 
distributions will be the same within the uncertainties of my own 
knowledge, and if the tempo is near the ideal value, it won't change my 
perceived posterior probability distributions.

You do have a point, though.  In well-tuned partnerships, where each 
knows the other's tendencies perfectly, they can transmit UI to each 
other with no chance of detection.


More information about the Blml mailing list