[c_corgi]

iviehoff, on 2012-October-22, 06:29, said:

The thing you hope to be true requires campboy to be wrong, I think. It is campboy who is advocating comparing the expected results immediately after the infraction with the expected result without the infraction. This is what results in the 50%/51% sensitivity. You avoid this by doing what the law says instead, ie, comparing the table result with the infraction to the expected result without the infraction.

campboy, on 2012-October-22, 06:57, said:

This sensitivity exists no matter who is right. If the chance of a spade lead was 51% without MI then (I hope) everyone would adjust; if the chance of a spade lead was 49% without MI then (I hope) no-one would adjust, so whatever you do when it is 50% there must be a point where 1% makes a difference.

In practice of course there is no way you can distinguish between the three cases. Unless I was confident that the chance of finding the right lead was greater with the correct information than without I would not adjust.

Mine was a shoddy post based on muddled thinking and expressing myself poorly, so I'll try again.

My hope (which I think is in line with campboy) is that when rectifying against damage we don't have to also rectify against iviehoff's slings and arrows of fate.

Using campboy's approach with a more sophisticated formula for weighting the adjusted score could eliminate the 51% sensitivity by arranging for the NOS expectation to always return to what it was before the infraction should it fall below that level as a result of the infraction. The formula would be something like:

P(S) - P(Smi) = the amount of 3NT-1 given to NOS when a spade lead was not found and the MI made a spade less likely.
[1-P(Smi)]

where P(S) is the probability of a spade lead without MI
P(Smi) is the probability of a spade lead with MI

[sailoranch]

campboy, on 2012-October-22, 06:52, said:

I'm not even sure any more what you are claiming that I know perfectly well. Let me outline what I believe a little more clearly than perhaps I did before.

In order for there to be damage you need both

i) the table result was worse than the expectation immediately before the infraction
ii) this was caused, at least in part, by the infraction.

You seemed to be saying that we only need i), and since i) was clearly satisfied there was damage. This is not true: we also need ii) and in my opinion ii) is not satisfied unless the correct information would have made West more likely to find the correct lead.

It is in my experience completely normal in MI cases to rule that "correct information would not have made <successful action> any more likely, therefore there was no damage".

I disagree with "ii) is not satisfied unless the correct information would have made West more likely to find the correct lead." I think (ii) is satisfied if the correct information would make it more likely that West would play differently.

So in scenario (ii), we compare the table result from a heart lead to a 50/50 guess between spades and diamonds. If we didn't adjust, we'd basically be saying that because West chose the losing option of a heart, we're making him lead the losing option of a diamond with the correct information. I don't see why a diamond lead without the MI would be perfectly correlated with the heart lead at the table.

It's true that a 50/50 guess in spades and hearts has the same expectation as a 50/50 guess in spades and diamonds. But the player chose a heart at the table, when he would not have chosen a heart without the MI. That was caused by the infraction. Yes, the possibility of an adjustment would give the NOS a better EV than they would have if they had been correctly informed, but I don't think that's a problem.

I don't understand why the fact that West flipped a mental coin to choose a heart at the table is relevant at all to assessing damage in 12B1.

[iviehoff]

campboy, on 2012-October-22, 06:52, said:

I'm not even sure any more what you are claiming that I know perfectly well. Let me outline what I believe a little more clearly than perhaps I did before.

The definition of damage seems to be the same for MI cases as for UI cases and other cases. Are you arguing that in practice it is implemented differently for MI?

[campboy]

In UI cases the infraction is choosing a particular call (or, much more rarely, play), and therefore it is generally clear that the infraction caused the final result. In fact it is very rare in UI cases that we cannot say "the result would definitely have been different without the illegal call".

The fact that "because of the infraction" is usually so clearly satisfied in UI cases that we don't even bother thinking about it does not mean we can ignore it in other cases.

[gordontd]

iviehoff, on 2012-October-22, 06:24, said:

When have you ever seen a judgment ruling based upon the expected result immediately after the infraction, as opposed to the table result? Answer, never.

Not so. This is Kaplan's "consequent" vs "subsequent", which I believe is still quite commonly used in the ACBL, and I have seen it used in other cases.

[gordontd]

I've just found this paper in which at the top of p2 the Chairman of the WBF Laws Committee appears to take an approach at odds with the wording of the 2007 Laws.

[iviehoff]

gordontd, on 2012-October-23, 07:31, said:

I've just found this paper in which at the top of p2 the Chairman of the WBF Laws Committee appears to take an approach at odds with the wording of the 2007 Laws.

But this is a SEWOG case is it not? That is why he is considering the "normal result after the infraction", in order to assess what the damage of the "utterly undisciplined" 3S call is, as opposed to the damage from the infraction.

[gordontd]

iviehoff, on 2012-October-23, 09:16, said:

But this is a SEWOG case is it not?

No, I don't think so, which is why he didn't use that wording. He did give another paper on the same course where he discussed how bad a play had to be before being considered to be a serious error.

But in the paper I linked to, he sets out a procedure for judging damage, which seems not to follow the laws and makes no mention of SEWoG.

[Sjoerds]

scenario 1.
3SAC There is no correlation between the infraction and the result.

scenario 2.
50% 3SAC and 50% 3SA-1. Different information can lead to a different choice.

Scenario 3.
75% 3SA-1 and 25% 3SAC. The same as scenario 2 but with a different outcome