smerriman

I don't know the exact definition of preempting; it differs from the older version of GIB since in 2017 BBO released an update and said "Preempts have been redefined to be much more in line with what most bridge players would expect.". But it seems this isn't quite accurate and they also introduced gaps between preempts and opening bids. (Opens your hand 1♠ with a point more, 3♠ with a jack less.) And in this case, an advanced robot doesn't help at all, since robots never simulate the opening bid (which is a good thing; it would be a disaster if it tried to predict the entire auction based on a small set of simulations).

Yes. The ACBL document says to alert all artificial bids except ones listed. It lists this as non-alertable: But that's not the case here.

Is it? 0 seems possible to me: [hv=https://www.bridgebase.com/tools/handviewer.html?n=S97652H753D65CT54&e=SQJT3HT8DT8CJ9763&s=SKHKQJ62DKJ7CKQ82&d=w&a=1dppx2dpp]400|300[/hv] Of course, there are times you should assume partner has a certain minimum even if they have done nothing but pass, like when balancing, and 2NT is silly here, but that's way too much for a bidding database and the fallback rule has some logic to it. (And to be honest, I'm not even sure 2N is defined for me.)

This should probably be in the robot forum, but GIB will virtually always use Blackwood if you start cuebidding - even if missing controls you denied - because it doesn't know how to cuebid properly and treats it as an extremely strong hand instead (25+ points). In particular, after the 5NT bid all it has promised is the ♠A, ♥Q, and a single extra 'total point' anywhere in its hand. It will bid like above with Axxx-Qxx-xx-xxxx. So any inferences that it may have 'extras' are unfounded. (It would have also bid 3♥ and 4N with Jxxx-Qxx-Qx-Qxxx, or even Jxxx-Qxx-x-Qxxxx, though not 5N of course).

In most places in the database, the number of HCP that a natural 2NT invitation shows is defined to be exactly {24 minus the minimum number of HCP partner has shown to date}. (Since it wants partner to pass with a minimum, and accept with more than a minimum). This is no exception - just that the minimum partner has shown so far is 0.

Is there a specific sequence you're referring to? Neither option uses up an entire bidding level - they're only a single bid apart - so I'm probably misunderstanding altogether.

I was probably thinking of the difference between non-serious and serious, where they have been some pretty strong arguments on BW about how much the information leakage is important, and much more than incidental. But compared to not playing 3NT at all, then definitely.

If partner wanted to know about a club control, 4♦ over 3NT will make that pretty clear. The general idea behind any type of non-serious 3NT bid is to prevent having to reveal extra information to the opponents if partner didn't really have any slam interest after all.

You misquoted me again. I said 12 total points; GIB has 13. The parts you should have quoted were all the ones where I said don't play with the free bots, because there are a huge number of situations in the bidding database where there are (sometimes intentionally) small gaps that the advanced robots are designed to fill in. Like the basic robot needing 13 total points to bid 4SF, but 13 HCP to have enough for 3NT. The advanced robot has no such problem, and raises your no trump bid. So once more, the mistake was playing with the version of GIB which has most of its logic turned off, rather than GIB itself.

[hv=https://www.bridgebase.com/tools/handviewer.html?lin=st||pn|smerriman,Argine,Argine,Argine|md|3SA32HA762DAT65CA6,S5HK854DQJ742CQ97,SKJ74HT9D83CKJT82,SQT986HQJ3DK9C543|sv|o|rh||ah|Board%201|mb|P|mb|P|mb|1D|an|4+!D;%20HCP%2011-23;%204-card%20!D%20opener|mb|P|mb|1S|an|4+!S;%20HCP%204-11;%20natural|mb|P|mb|1N|an|2+!C;%204-5!D;%202-4!H;%202-3!S;%20HCP%2015-17;%20balanced|mb|P|mb|2N|an|4!S;%20HCP%207-8;%20invitational%20hand|mb|P|mb|3N|an|2-4!C;%204-5!D;%202-4!H;%202-3!S;%20HCP%2015-17;%20to%20play|mb|P|mb|P|mb|P|pc|H4|pc|H9|pc|HJ|pc|H2|pc|HQ|pc|H6|pc|HK|pc|HT|pc|H5|pc|D3|pc|H3|pc|H7|pc|CA|pc|C7|pc|C2|pc|C3|pc|C6|pc|C9|pc|CT|pc|C4|pc|CK|pc|C5|pc|D5|pc|CQ|pc|CJ|pc|S6|pc|D6|pc|S5|pc|C8|pc|S8|pc|S2|pc|D2|pc|S4|pc|S9|pc|SA|pc|D4|pc|HA|pc|H8|pc|S7|pc|D9|pc|DA|pc|D7|pc|D8|pc|DK|pc|S3|pc|DJ|pc|SK|pc|ST|pc|SJ|pc|SQ|pc|DT|pc|DQ|]400|300[/hv] West's heart plays at trick 2 and 3 look ludicrous, and I figured was surely an Argine bug. But when I ran this through a double dummy sim, overtaking the king only cost a trick on 2 out of 500 deals, but gained on 55 occasions, so is "correct" according to double dummy. But the gain only occurs when you have a killing switch. After you've overtaken with the king, double dummy analysis tells you a heart continuation works more often than a switch. So the overtake ended up being a no-win play after all, but not one that a robot can foresee. The perils of double dummy analysis.

Why the king? You wouldn't lead the ten from T98xx(x) or T97xx(x)? The robot knows it's not from shortness, and that declarer has spades stopped too. It's all of those cases where the heart picks up extra tricks.

No arguments there. My "waste of time" was referring to in-person bridge, which is what all of the alert regulations were written for (ie where 'alerting' and 'providing a description' are independent).

Too early to answer any of those questions. Of course, Argine has been around for quite a long time on Funbridge, and I've seen people say it is "better", but that may be due to bias against GIB and / or because it hasn't been put to the heavier scrutiny of BBO. The only thing I do know - based on Funbridge, posts on BridgeWinners, and elsewhere - is that the developer is active and improving it, which is a big plus.

I guess that's precisely why there are regional alerting procedures, and why the WBF policy says it's not intended to replace those.

Fair enough. Just seems like it would be a big waste of time alerting bids like that where everyone knows the meaning, and then there's the 1 pair who plays it differently, and you'd never know unless you asked your opponents to explain what every single one of the obvious bids meant every time they bid it. Seemingly going against the whole idea of alerts in the first place.

I thought alerts were invented to alert opponents to the fact that your bid may not have the meaning they'd expect. I don't see any difference between this and having to alert Blackwood because it's not a natural no trump, or having to alert a strong 2♣ opener because it's not a natural club bid. You of course have to choose what the most common 'expected' meaning is, but I'm sure if you polled this on BW as to what the bid meant without agreement, support for spades would be near unanimous.

Really? ACBL alert procedures 2022:

It's the NMF and "cue shows inv+ support" rules getting mixed up.

*free I've been misquoted :) Though sadly most of these are issues with the advanced bots too :( The first older GIB finds the trivial 4♥ raise, but modern GIB apparently doesn't even simulate. The latter appears due to the 2♣ bid only promising 3 clubs, so it wants 5 to raise (modern GIB also can't raise so doubles 2♠ instead).

North is bid by Argine too, so if that were the case, it would be reasoning against its own logic. GIB (while potentially simulating a slightly different set of hands due to the difference in North's bid meanings) comes up with the J every time at matchpoints, which is the format being played here, so I don't think you can put this down to simulations. It is true that at IMPs, GIB will also play the King for the reason you mentioned. So if Argine is mistakenly playing IMPs.. which I guess is actually a possibility given the integration has only just been set up. GIB also concludes that heart is the best continuation based on double dummy simulations at MPs, so the robots seem to be in agreement here.

No, but it subtracts a point if there's an honor in a short suit.

You're crazy. OK, now that that's out of the way, time to look at the hand.. :) But it appears you're going to need AK AK in both minors, or AK AQ might do half the time, or rarer cases when you can set up a long minor winner.. but these seem too long odds to use RKC. Ran a super basic sim where all it assumes is that North has 5+ spades, <= 1 heart, and 12+ HCP (perhaps generous). - 26% of the time partner has 2+ aces and slam is makeable - 21% of the time partner has 0-1 aces and 5S is makeable - 30% of the time partner has 2+ aces and slam goes down - 23% of the time partner has 0-1 aces and 5S goes down It doesn't look good to me.

[hv=https://www.bridgebase.com/tools/handviewer.html?lin=st||pn|~~v3fakebot,~~v3fakebot,smerriman,~~v3fakebot|md|1SH982DAJ9654CAKT9,SJ654HA54DQ32C754,SAQ32HQT76DCQ8632,SKT987HKJ3DKT87CJ|sv|e|rh||ah|Board%203|mb|1D|an|4+!D;%20HCP%2011-23;%204-card%20!D%20opener|mb|P|mb|2C|an|1+!C;%20HCP%2010-27;%20natural%20or%20fit%20in%20!D%20-%20Forcing|mb|P|mb|3C|an|4-6!C;%204+!D;%20HCP%2012-23;%20fit%20-%20Forcing|mb|P|mb|3N|an|1+!C;%20HCP%2012-15;%20to%20play|mb|P|mb|P|mb|P|pc|ST|pc|D4|pc|S5|pc|SQ|pc|C2|pc|CJ|pc|CK|pc|C4|pc|H9|pc|HA|pc|H6|pc|H3|pc|H5|pc|H7|pc|HK|pc|H2|pc|HJ|pc|H8|pc|H4|pc|HQ|pc|HT|pc|S8|pc|D5|pc|C5|pc|SA|pc|S7|pc|D6|pc|S4|pc|C3|pc|D8|pc|CA|pc|C7|pc|CT|pc|S6|pc|C6|pc|D7|pc|DA|pc|D2|pc|S2|pc|DT|pc|C9|pc|D3|pc|CQ|pc|S9|pc|C8|pc|DK|pc|D9|pc|SJ|pc|S3|pc|SK|pc|DJ|pc|DQ|]400|300[/hv] Note declarer is North, Dummy is South. Check out the heart plays on tricks 3, 4, 5.

Not strange at all; exactly as described above. Not something that needs fixing; just stop playing with the free bot :P

Small typo, since you meant 17. A simpler way of looking at the aces part is: produce 10000 condition hcp(north)+hcp(south)==32 action frequency "aces" (aces(north)+aces(south),0,4) Frequency aces: 0 0 1 0 2 50 3 3472 4 6478 Even if you knew you had exactly 32 HCP combined (rather than 32-34 like here), the chance that the 8 missing points are two aces is about half a percent, so pretty negligible. (Shape doesn't have an impact on that figure.) What is quite fascinating from your numbers is when you do have a 17 count opposite a 1NT opener, partner will have the minimum 55% of the time..