[pescetom]

This board was surprising to me.

This is the second time in three days I saw GiB supply MI, in the first case a missing explanation when it has an agreement and in this case an illogical and self-serving explanation when it has a different and more logical agreement.
Usually it explains the 3-level raise honestly as "The Law, 3+ cards".
Here it had 2 cards and decided to deviate from agreements ? ( perfectly legal, although opens up many discussions).
But it then explained as "The Law, 2+ cards", which is both illogical and misleading, above all seems based upon the actual hand rather than the agreement.

[pilowsky]

pescetom, on 2023-April-23, 14:58, said:

This board was surprising to me.

This is the second time in three days I saw GiB supply MI, in the first case a missing explanation when it has an agreement and in this case an illogical and self-serving explanation when it has a different and more logical agreement.
Usually it explains the 3-level raise honestly as "The Law, 3+ cards".
Here it had 2 cards and decided to deviate from agreements ? ( perfectly legal, although opens up many discussions).
But it then explained as "The Law, 2+ cards", which is both illogical and misleading, above all seems based upon the actual hand rather than the agreement.

It passes the "Law" with an 8 card fit.
You have 6, it has 2.
Ergo hic hoc.

---

[sfi]

pescetom, on 2023-April-23, 14:58, said:

But it then explained as "The Law, 2+ cards", which is both illogical and misleading, above all seems based upon the actual hand rather than the agreement.

"The Law" says that if there are 16 total trumps, you should bid 3 over 2 but not 3 over 3. So if E-W have 8 spades then "The Law" indicates a 3-heart call. Is this good bridge? You can read the various books and make of it what you will.

[smerriman]

sfi, on 2023-April-23, 17:51, said:

"The Law" says that if there are 16 total trumps, you should bid 3 over 2 but not 3 over 3.

I haven't seen it put that way before; most statements of The Law say that you should bid 3 over 2 only with 17 trumps, as then if they have 8 we have 9.

I guess your version would take vulnerability into account, but GIB doesn't; it uses the simpler interpretation anyway, so pescetom is right that its understanding of The Law shows 3 trumps here.

But the explanation is the same as the other thread; GIB has no way of saying 'or'.

It has a default rule that in any auction, if it knows the partnership holds 21-24 combined points, it can make a non-jump bid at the 3 level if that shows a previously undisclosed 8 card fit.

It has a second rule that after opening a weak 2, a 3 level raise can be used on any hand with 3 trumps that doesn't have enough values for game opposite a minimum, and must be alerted with 'The Law'.

3♥ thus combines the two rules, and means something like {"I have 14-17 points, 2+ hearts" or "I have 4-17 points and 3 hearts, alert The Law!"}. The point ranges, trump length ranges, and alerts are merged for the two rules via least common denominators into a single description that removes the word 'or'.

It's not deviating from any agreement or changing the description based on its hand; it's just not disclosing its entire logic which wouldn't come close to fitting in the description box (it also involves a lot of rules that eliminate other bids first; the above is a simplified version).

[thepossum]

The Law is a strange thing. Sometimes it works and sometimes not. But you (me) are obviously using it for the wrong type of hands. It only applies in situations when it works. I just had what appeared to be a law hand but used a different rule for a better result. But it was only out by one trick. Of course

[sfi]

smerriman, on 2023-April-23, 19:47, said:

I haven't seen it put that way before; most statements of The Law say that you should bid 3 over 2 only with 17 trumps, as then if they have 8 we have 9.

That's not quite what Cohen wrote (maybe he elaborated on it in later books). The way he phrased it in his first book is:

Quote

Never outbid the opponents on the three level with 16 trumps. (Chapter 2)

Try not to let the opponents play at a level equal to their number of trumps. (Chapter 5)

(Box and bold in original text)

In summary, he says you should bid 3 over 2 but not 3 over 3 with 16 total trumps. He has tables and everything to back up his thesis, but you're right that his approach doesn't quite work when you're vulnerable and they're trigger-happy.

As for what the robots are doing, you're much better placed to answer that. But according to the description, it kind of looks like they were following Cohen's guidelines for once.

[smerriman]

I guess the catch is that it relies on knowing your opponents have an 8 card fit - leading to the adage of never letting the opponents play in a fit at the 2 level - which isn't the case here.

That's probably why it's more commonly described as bidding to the level of the number of trumps you have, and needing 3 card support to raise to 3.

Will have to look at the book but it would surprise me a lot if he suggests raising 2♥ to 3♥ in this situation just because you hold 2 card support and the opponents might have a fit. With this hand GIB is bidding it based on HCP instead.

[DavidKok]

Ignoring the actual hand for a second (I'm not familiar with GIB but this looks like a value bid, not a LAW-based bid, compatible with smerriman's explanation), Cohen introduces the "bid to the level of your total number of trumps" as a shortcut for "bid 3 over 2 with 16 total trumps, but not 3 over 3". The idea is that, conditional on us having an 8 card fit, the opponents are unlikely to have a 10 card fit or longer (if they have a 10 card fit that's 10 out of their 26 cards spoken for, and the other 15 have to break 5-5-6 to give us 'only' an 8 card fit, in which event we even have a double fit). Conversely, it is possible but unlikely the opponents only have a 7 card fit (our other 18 cards have to split 6-6-6). So with exactly an 8 card fit, assume the number of total trumps is 16 or 17, and act accordingly. Similar logic can be applied to a 9 card fit, but it gets less accurate with 10 card fits or longer.

To the best of my knowledge the statement of the LAW is that the total number of tricks (defined as the number of tricks NS can make in their optimal trump suit plus the number EW can take in their optimal trump suit - neither of which need be the longest trump suit) is (often) equal to the number of trumps (longest trump suit of NS + longest trump suit of EW). All rules on bidding X over Y, or bidding to level Z, are derived from this in combination with bridge score tables and some statistics, along with adjustments to make the law more accurate.

[nullve]

One often has to apply the Law in situations where one doesn't know, but can only estimate, the number of total trumps.

As North I would estimate total trumps to be ~ 16 on this deal.

[pilowsky]

Bit of a stretch calling it a Law.
How about the Possibility of Total Tricks?

[sfi]

smerriman, on 2023-April-24, 00:30, said:

I guess the catch is that it relies on knowing your opponents have an 8 card fit - leading to the adage of never letting the opponents play in a fit at the 2 level - which isn't the case here.

That's definitely a catch here, but only a moderate one if you assume that opener will never hold four spades. I was assuming that the robot did make that assumption and then simulated enough hands to decide that it was likely they were in an 8-card fit.

Of course, the law itself is poor enough that you really don't want to be making additional assumptions and then relying on the outcome. In short - I agree that it's not a good "law-based" decision, but it at least made sense to me that the robot was mostly applying it.

[pescetom]

My understanding of what the LoTT says is exactly as summarized by smerriman. I agree with Larry Cohen of course about the opportunity of bidding 3 over 2 with 16 TT at suitable vulnerability, but that is going beyond the Law itself as I see it.

What worries me here however is not so much the technical bidding choice or the logic behind it, but the misinformation about actual agreements. If GiB has decided ("agreed") that we bid 3 on either strong 8 card fit or any 9 card fit then it should describe the agreement to opponents this way: "The Law", which is jargon for only the second possible explanation of the call, is inappropriate.

The legal problem of course is that the robot will not explain better the next time even if caught and punished. This is arguably closer to a concealed partnership understanding than an accidental poor explanation.

[smerriman]

It should from a legal perspective yes. But it would be a nightmare from a technical perspective. Sometimes bids have 6 or 7 triggering rules of different priority; listing every single one one after the other would be a very long description.. and likely end up even more misleading / insufficient legally, because most of the rules are highly conditional on dozens of other rules *not* matching. And even in a single rule itself, the part that gets into the description is often very different from all of the conditions that it actually tests..