[straube]

I was reading through Kerr's relay structure...and others modeled after his...and I noticed that his penultimate pattern is the least frequent pattern. So...

3D-5332
3H-6331
3S-7330
3N-7(32)1, 2 controls
4C-7(32)1, 3 controls
etc

but his "super-accept" starts at 5 controls.

After 4D end signal
.....4H-2-4 controls
.....4S-5 controls
.....etc

So don't these ideas conflict? Why doesn't he end...

3D-5332
3H-6331
3S-7330
3N-7(32)1, 2-4 controls
4C-7(32)1-5 controls
etc

In fact, how does he follow up after pd has shown 7330 and captain rebids 3N?

3N-
.....P-2 controls? Or 2-4 controls?
.....4C-3 controls? Or 5 controls?

This gets even more critical when we're dealing with QPs instead of controls. Say our QP base is 6. Do we really want to play

3S-7330
3N-7(32)1, 6
4C-7(32)1, 7
etc.

I would think 3N would be 6-9 or 6-8 or whatever.

[rbforster]

From a probability and space conservation perspective, if you can afford to zoom into the next question, you want to flip the last two answers relative to the "natural" descending frequency as bids rise. Essentially, you are optimizing not for the resolution level of the first question (which would be strictly in probability order), but for average resolution level of the second question. For this, a little thinking will convince you that this last flip is indeed an improvement.

However, when you starting running out of space below a likely game contract, practical concerns will have to take over from theory.