[Wentys]

AI made an interesting (maybe wrong?) point in response to my prompt "When is a 4-4 trump fit better than a 5-3 fit"?
It gives the standard reasons why 4-4 is preferred, but also teasingly suggests that our 4-4 holding gives better
split possibilities of finding the missing cards 3-2.

Is there any theory that the "conditional probability" of a 3-2 outstanding split might not be the same if hold 4-4, vs 5-3?

Thanks
Peter

[DavidKok]

No, the probabilities are exactly equal.

[Cyberyeti]

Theoretically the probabilities are equal, I'm not sure if there are variances in this if the boards are hand shuffled.

It is well known that you get flatter hands after inadequate hand shuffling, but if there's any effect it will be tiny.

[thepossum]

Cyberyeti, on 2026-January-25, 04:23, said:

Theoretically the probabilities are equal, I'm not sure if there are variances in this if the boards are hand shuffled.

It is well known that you get flatter hands after inadequate hand shuffling, but if there's any effect it will be tiny.

Gut feel suggests even in poorly shuffled hands there is no (statistical) difference - I assume the deck has been cut too lol - can I prove it. Of course not

The only reason something like Goulash produces weird hands is the dealing

[wuudturner]

Consider two hand shapes for the NS pair: 5=3=3=2 (for North) opposite 3=5=3=2 (for South), compared to 4=4=3=2 opposite 4=4=3=2. All I've done is to swap a heart for a spade between the N-S hands. And that means the E-W hands will have been untouched. There is no reason to think the probabilities for the E-W holdings between the two cases should now change. AI tools are not always correct. In fact, they are often just flat out wrong.

[mike777]

 wuudturner, on 2026-January-25, 18:43, said:

Consider two hand shapes for the NS pair: 5=3=3=2 (for North) opposite 3=5=3=2 (for South), compared to 4=4=3=2 opposite 4=4=3=2. All I've done is to swap a heart for a spade between the N-S hands. And that means the E-W hands will have been untouched. There is no reason to think the probabilities for the E-W holdings between the two cases should now change. AI tools are not always correct. In fact, they are often just flat out wrong.

So are humans

Give it time to learn please

[Cyberyeti]

thepossum, on 2026-January-25, 17:38, said:

Gut feel suggests even in poorly shuffled hands there is no (statistical) difference - I assume the deck has been cut too lol - can I prove it. Of course not

The only reason something like Goulash produces weird hands is the dealing

What I was suggesting was that 4-4 might be more common if the hand was poorly shuffled, therefore the chance of it being poorly shuffled might increase, therefore the chance of a 3-2 might increase, but the effect would be slight at best.

[AL78]

If the fit is 4-4, there may be a slightly enhanced probability that suits have been dealt more evenly around the table making a 3-2 break slightly more likely, but I would expect this effect is miniscule if it exists at all.

[AL78]

wuudturner, on 2026-January-25, 18:43, said:

AI tools are not always correct. In fact, they are often just flat out wrong.

I agree it is unwise to take everything at face value, but my gut feeling is AI get criticised more than it should. I have been having a go with it with positive results. Recently I asked it how I can reign in my frustration during bad bridge sessions with some information on why I might get frustrated more than I should. It came back with some good mental techniques which I am practicing with the aim of some self improvement.

[mike777]

 AL78, on 2026-January-26, 07:54, said:

I agree it is unwise to take everything at face value, but my gut feeling is AI get criticised more than it should. I have been having a go with it with positive results. Recently I asked it how I can reign in my frustration during bad bridge sessions with some information on why I might get frustrated more than I should. It came back with some good mental techniques which I am practicing with the aim of some self improvement.

So AI is influencing your behavior...

[P_Marlowe]

Hi,

No.

If you know, that a pair of 2 hands has a 44 fit, and another pair of hands has a 53 fit,
you basically have fixed 26 cards, in both case you know, that the remaining 26 cards contain
the same number of cards in a given color 5.
Randomly dealing those remaining 26 cards delivers the same probability for a 2-3 split in
both cases.

What I think the AI suggests is, that the 44 enables you to easier find out, how the cards are
distributed.

With kind regards
Marlowe