[Bbradley62]

I posted this in the middle of another thread, but I think it got buried there. Here is another wrong-card-from-honor-sequence lead. http://tinyurl.com/79yz5l9

[Stephen Tu]

This is odd from a human standpoint, but this isn't one where there are equal honors and GIB is choosing some convention from equal honors and getting it wrong. It's probably dealing hands where the Q is just *better* than the ten or 9. Like for example if declarer had stiff J of diamonds and dummy has the K. But for some reason not getting hands where declarer has AJx/kjx and dummy has xxx, etc. So it's a bit weird that it does it, but not going to be something fixable, unlike lead from actual touching honors.

[Bbradley62]

GIB has it marked (yes, by default, but still marked) on his convention card that he leads T from QT9x against suits. If that's not what he actually does, then either his leads algorithm or his convention card should be editted. If it's too difficult to change the algorithm, it should be very easy to change the convention card. I guess you're saying this is the case for all interior-sequence leads.

[barmar]

Since you were North, this must have been in the MBC. Are you using the download or web version of BBO?

[Stephen Tu]

No, I'm saying that if it's going to lead the Ten or 9 from QT9x, that's a convention. If you catch the GIB leading the 9 from this holding, then that's a correctable bug. The ten or 9 are equals, it's just agreement which one a partnership leads.

But here, the Q and the Ten ARE NOT equals. It is *not* by convention that is is leading the Q here. If its calculations are showing that the Q is going to superior to the T or 9 on some auction, it's going to lead the Q. It's not going to lead the Q from QT9x *always*, only when it thinks the auction makes the Q score better. It's not a "different lead from agreed convention" here. It's "playing the best card" (from how it is thinking).

It's just like normally from Kxxxx playing in a trump contract, one normally leads 4th best or fifth best by convention. But if the auction is such that you have a ten card fit, and you choose to lead this suit, you might bang down the K in case you have to switch. That doesn't mean you change your convention card to say K is std from this holding! Same if your holding was KJTxx, you might lead K even though this isn't your convention card understanding.

[Bbradley62]

barmar, on 2012-May-04, 14:21, said:

Since you were North, this must have been in the MBC. Are you using the download or web version of BBO?

Always web.

[Bbradley62]

Stephen Tu, on 2012-May-04, 14:22, said:

It's just like normally from Kxxxx playing in a trump contract, one normally leads 4th best or fifth best by convention. But if the auction is such that you have a ten card fit, and you choose to lead this suit, you might bang down the K in case you have to switch. That doesn't mean you change your convention card to say K is std from this holding! Same if your holding was KJTxx, you might lead K even though this isn't your convention card understanding.

I think this is incorrect. If I occassionally deviate from the convention card in bidding or card play, that's one thing, but if I do it often enough that my partner might realize that there's a pattern, it becomes a partnership understanding and requires disclosure. It certainly appears that GIB does this often enough that he should have both the Q and T marked (and probably similar double-marks for other interior sequences).

Edit: Similarly, since GIB completely consistently leads low from three small against NT, his convention card should not say that he leads high from that holding.

[Stephen Tu]

So you are seeing it lead top card instead of from the interior sequence a lot? I hadn't noticed. Doesn't seem to me that it should happen a lot since the sims should often find it suboptimal vs. maintaining a tenace position over declarer.

[Bbradley62]

Yes, it appears to me that it's not an "only occassional" thing.

[barmar]

I'm trying to find out what's going on here. We know we messed up some of the NT leads in the system file that went with v21, and we fixed it in v22. But maybe something else got screwed up in the process.