smerriman

I would have bid 3♦ instead of doubling. But I'm probably wrong.

Aw, sorry to disappoint, I messed up :( It wasn't what I thought, though still something very interesting. Contrary to what I believed, it appears GIB does take into account leading rules in its simulations. In this case, the normal lead for GIB from T973 is the T; when it leads the 3, it assumes it doesn't have that holding, which means you don't have AJ8. So after the A and 8 are played, you must have two diamonds left, so what it plays at the end is irrelevant. Still going to follow this up more to see how exploitable it is..

Pretty sure I've figured it out, and it's another mammoth fundamental GIB bug. And if I'm understanding correctly, also what's causing some of those ones late in the play where it simulates hundreds of deals and misses the one actual important one. Stay tuned for part 3 of my 0% plays series (though need to put in a lot more analysis / testing first - if I'm right, this is the biggest bug yet and completely breaks GIB.)

There's definitely extremely strange going on here. At trick 10, at the point you lead the ♣3 from dummy, East knows you have one club left, and thinks there is a 36% chance you have 1 diamond left, and a 64% chance of 2 diamonds left, which makes sense. The 64% where you have 2 diamonds left, it sees J9, JT, and JT all equally likely, as you'd expect. The 36% where you have 1 diamond left.. it thinks you have the T 100% of the time, and the J never. GIB isn't being clever; quite the opposite - once you play the J, it thinks that guarantees you have the last diamond, so plays a random card. Still trying to figure out why it rules out the actual case.. it's not like it thinks 1NT by South doesn't fit with that hand; it bids it itself..

If your partner had the club ace instead of the diamond ace, would she have also cuebid 4♦ to show the diamond king? (Otherwise you miss an equally good slam.) If so, perhaps describing it as first round controls isn't entirely accurate.

Hardcoding a completely arbitrary figure like that in doesn't really make any sense at all to me, and wouldn't fit into the entire concept of the way it bids, so I can't agree with you there.

Right. But there is no chance a robot could ever figure out what humans at other tables would do; trying to add that in would more likely cause disasters than solve anything. Assuming the same contract at the other table is very reasonable given the robots' consistency.

Bid 3♦, then if GIB bids 3♠, undo and bid 2♠ instead.

No, all of that part works perfectly fine. It doesn't need to use human methods for working out what percentage it needs to make to be profitable; all that is computed from the simulated hands and theoretically should work better than the human methods. (But of course, as usual, a low number of simulations is not always 100% accurate).

Not too relevant here, since it also plays 4♣ as a help suit slam try rather than a cue bid. But it's the exactly the same as discussed in previous threads; the GIB database will never, ever, bid a grand slam unless holding the appropriate combined number of points. North can tell that's impossible in this situation, so every single simulated bid will lead to an identical result of 6♥. (And indeed, over the help suit slam try, 5♥ ties 4N and 6♥, it's only picking 5♥ based on an arbitrary ordering. It didn't consider 7♥ then because it wasn't allowed to). If enough hands where you can make grand turn up, then it will leap to 7♥, since that will work better on average. It is not capable of realising that if you go slow, you can simulate to find 7♥ if it's best later.

Just another 50% board in a challenge. [hv=https://www.bridgebase.com/tools/handviewer.html?bbo=y&lin=pn%7Csmerriman%2CRobot%2CRobot%2CRobot%7Cst%7C%7Cmd%7C1ST842HKTDAQ95CT87%2CSAJ96HA943DKJ2CA6%2CS75HJ7652D7643C93%2CSKQ3HQ8DT8CKQJ542%7Csv%7CB%7Cah%7CBoard%207%7Cmb%7CP%7Cmb%7C1N%7Can%7Cnotrump%20opener.%20Could%20have%205M.%20--%202-5%20%21C%3B%202-5%20%21D%3B%202-5%20%21H%3B%202-5%20%21S%3B%2015-17%20HCP%3B%2018-%20total%20points%7Cmb%7CP%7Cmb%7C2N%21%7Can%7CMinor%20transfer%20--%206%2B%20%21C%7Cmb%7CP%7Cmb%7C3C%7Can%7CTransfer%20completed%20--%202-5%20%21C%3B%202-5%20%21D%3B%202-5%20%21H%3B%202-5%20%21S%3B%2015-17%20HCP%3B%2018-%20total%20points%7Cmb%7CP%7Cmb%7C3N%7Can%7C6%2B%20%21C%3B%2013-15%20HCP%7Cmb%7CP%7Cmb%7C4N%7Can%7CBlackwood%20%28C%29%20--%202-5%20%21C%3B%202-5%20%21D%3B%202-5%20%21H%3B%202-5%20%21S%3B%2017-%20HCP%3B%2018%20total%20points%7Cmb%7CP%7Cmb%7C5D%7Can%7COne%20or%20four%20key%20cards%20--%206%2B%20%21C%3B%2013-15%20HCP%7Cmb%7CD%7Can%7C6-11%20HCP%3B%20rebiddable%20%21D%3B%20%21DKQ%3B%2012-%20total%20points%7Cmb%7CP%7Cmb%7CP%7Cmb%7C7C%7Can%7C6%2B%20%21C%3B%2015-%20HCP%3B%2017%2B%20total%20points%7Cmb%7CP%7Cmb%7CP%7Cmb%7CD%7Can%7C1%2B%20%21D%3B%2018%2B%20HCP%3B%2019-20%20total%20points%7Cmb%7CP%7Cmb%7CP%7Cmb%7CP%7Cpc%7CD3%7Cpc%7CDT%7Cpc%7CDA%7Cpc%7CD2%7Cpc%7CD5%7Cpc%7CDK%7Cpc%7CD7%7Cpc%7CD8%7Cpc%7CCA%7Cpc%7CC3%7Cpc%7CC2%7Cpc%7CC7%7Cpc%7CC6%7Cpc%7CC9%7Cpc%7CCK%7Cpc%7CC8%7Cpc%7CCQ%7Cpc%7CCT%7Cpc%7CH4%7Cpc%7CH2%7Cpc%7CH8%7Cpc%7CHT%7Cpc%7CHA%7Cpc%7CH6%7Cpc%7CS6%7Cpc%7CS5%7Cpc%7CSK%7Cpc%7CS2%7Cpc%7CCJ%7Cpc%7CS4%7Cpc%7CH3%7Cpc%7CH7%7Cpc%7CC5%7Cpc%7CD9%7Cpc%7CH9%7Cpc%7CD6%7Cpc%7CC4%7Cpc%7CDQ%7Cpc%7CS9%7Cpc%7CS7%7Cpc%7CS3%7Cpc%7CS8%7Cpc%7CSJ%7Cpc%7CD4%7Cpc%7CSA%7Cpc%7CHJ%7Cpc%7CSQ%7Cpc%7CST%7Cpc%7CDJ%7Cpc%7CH5%7Cpc%7CHQ%7Cpc%7CHK%7C]400|300[/hv] Nice double by North, I guess.

https://webutil.bridgebase.com/v2/tview.php?t=ARDCHALLENGE:149e6ccb.296c.11ed.897d.0cc47a39aeb4-1661977969&u=smerriman&v3b=web&v3v=6.3.8 smerriman 6.5 - 9.5 xbabarx

Huh, never noticed that before. Looks like a typo in the description, since it happily opens 2M with 4 diamonds.

Nothing to do with clubs. Same as I mentioned in another thread recently - GIB never bids freak hands properly because it can't reconcile the difference in HCP and total points, with it failing both the rules for 1♥ and 4♥ based on different criteria. The same happens to later heart bids, with none matching.

You're probably going to have to provide a few more details, including what system you're playing. If your partner isn't a passed hand, 2 of a new suit is forcing, so you can't pass. What you bid instead - a new suit, rebidding your original suit, or 2NT - varies depending on agreements, with all having multiple meanings. Likewise, 1NT is forcing for some (if you opened a major) and not forcing for others. If it's forcing you may have to bid a 3 card minor if you have no other suit to bid, while if it isn't forcing then pass is probably fine.

There's heaps of threads on GIB not giving preference correctly. In this case, it thinks you've shown longer clubs than diamonds, even though that's nonsense. The function that does this is hidden in the code rather than the database, but my suspicion is that every time you bid clubs, it increments by 1 the number of clubs and diamonds you hold, which sounds reasonable. The problem is that the since the first bid of clubs is strong at the 3 level, it starts clubs at 5, and diamonds only started at 3 for the opening bid. So each time you increment, it's keeping clubs one higher than diamonds.

paulg's post in the Water Cooler I hardly ever read that area of the forum given it's for non-bridge topics, and given he was a prominent forum member, I feel like he deserves a thread of his own. I only knew nige1 through his forum posts, so very little compared to many others, but of my memories: [hv=https://www.bridgebase.com/tools/handviewer.html?d=s&a=7SPPP&s=sakqjt98765432&n=hakqjt98765432&w=dakqjt98765432]400|300|I rank: 1. His unique use of the handviewer diagrams, always ranking all available options rather than just giving a single answer 2. The fact that he taught me that 'pusillanimous' was an actual word 3. In forum challenge events where a round is meant to take 2-3 weeks, he would always have all challenges done within a couple of days 4. His clever squeezes and endplays commentary where he would always minimize unimportant spots to emphasize the technique. [/hv]

Well, your post is about *you* using Gerber, and the robots responding to it, rather than GIB using it. As far as I'm aware it's specifically told it can never bid Gerber itself over NT openings (though understands humans can and will respond to it) - it can only bid it itself in three specific situations: - after 1m - 2N - after 1x - 1y - 1N - after 2C - 2N I have seen a couple of these pop up, though they're very rare.

https://webutil.bridgebase.com/v2/tview.php?t=ARDCHALLENGE:40b7541c.2628.11ed.897d.0cc47a39aeb4-1661618984&u=smerriman&v3b=web&v3v=6.3.8 smerriman 8 - 8 Stephen Tu https://webutil.bridgebase.com/v2/tview.php?t=ARDCHALLENGE:4eab55ad.2682.11ed.897d.0cc47a39aeb4-1661657662&u=smerriman&v3b=web&v3v=6.3.8 smerriman 9 - 7 toast1 https://webutil.bridgebase.com/v2/tview.php?t=ARDCHALLENGE:017ddf0f.26b9.11ed.897d.0cc47a39aeb4-1661681155&u=smerriman&v3b=web&v3v=6.3.8 smerriman 9.5 - 6.5 natali_ I apologize to natali_ and everyone in the event for blatantly exploiting a bug in board 11 (nullve's poor influence). Punishments by consensus :(

As I think I mentioned in another thread, GIB's entire system is broken over 2♦ interference - most bids and continuations are undefined, and when there isn't a single bid that it understands, it passes.

I think you misread. The point is that the bot is *not* using simple arithmetic - it should be adding its 16 points to your 20 points, coming up with 36, and concluding with 4 missing that you may be off an ace. But yes, as I mentioned in an earlier thread, one GIB developer said he had a long term goal to make GIB always ask for aces if it were able to before leaping to slam.. but clearly it wasn't trivial as he was never able to do it.

The 2 would be led from both xxxx and Jxxx, so unfortunately there's not too much in terms of inferences there. East does show up with the Axxx of clubs though (edit - specifically, A963, which could be relevant when the T drops) which definitely complicates things. It felt like one of those hands where even if you cut off your own entries, the opponents are probably going to have to put you in somewhere given you have tricks in all 4 suits, but maybe it's not so easy after all..

None of the simulated hands have North holding 3 spades, so no issues there. Had it run it about 20 times before it came up with a conclusion of 3♠ rather than 3N or 4♣, but the time it did, 3 of the hands were: [hv=?n=sk5hkjt982da7ckjt&s=sat864dj32caq752&w=sj92haq753dk984c6&d=s&v=b&a=1sp2hp2sp2np3cp3hp3sp4sppp]400|300[/hv] extrapolates 3♠ raised to 4♠ for +620, 3NT passed for -100, 4♣ raised to 5♣ for -200 [hv=?n=sk5haj5432dat5ck4&s=sat864dj32caq752&w=sj97hq9dk874ct983&d=s&v=b&a=1sp2hp2sp2np3cp3hp3sp4sppp]400|300[/hv] extrapolates 3♠ raised to 4♠ for +620, 3NT passed for -100, 4♣ corrected to 4♠ for +620 [hv=?n=sk5hat8532dat7ck4&s=sat864dj32caq752&w=sj9h64dkq64cjt963&d=s&v=b&a=1sp2hp2sp2np3cp3hp3sp4sppp]400|300[/hv] same as case 2 [hv=?n=sqhkjt853daq5ck83&s=sat864dj32caq752&w=s953h762dt84cjt96&d=s&v=b&a=1sp2hp2sp2np3cp3hp3sp3nppp]400|300[/hv] In this case, it predicts North will correct 3♠ to 3NT, so both spade and NT bids are fine - but 4♣ will result in going down in a club game. There were also some hands where both 3NT and 4♠ made, but there were only 9 tricks available in NT, meaning a small win for 4♠, like: [hv=?n=skhaj7643dk9cjt43&s=sat864dj32caq752&w=sqj52hqt9852da4c9&d=s&v=b&a=1sp2hp2sp2np3cp3hp3sp4sppp]400|300[/hv] Plenty of hands where both options led to the same score (eg down 1) too. Of course, there were also hands where 3♠ is worse, and in almost all simulations these come up often enough to make it choose 3NT.. the point is, no matter what the odds, there's always a small percentage of the time a simulation will come up with the wrong conclusion, even if the bidding system is perfect. And of course, many of the outcomes aren't what would really happen, but only with double dummy play, but you can't help that.

[hv=https://www.bridgebase.com/tools/handviewer.html?s=SJ942HADK3CQ87542&n=SQTHKQJ43DAQT4CKJ&a=1cp1hp1sp2dp2np3nppp&d=s]400|300[/hv] Lead is the diamond 2. Whatever number of NT you decide to play in (would you bid 4 as North?), this is MP and you just need to cash the seemingly inevitable 10 tricks. I somehow only managed 9.. is there an easy route to avoid mistakes? [edit] I should say, South's bidding was actually a robot, so don't be too harsh :) The play's the thing.

3♠ is based on a simulation too, and like I said, the result of this is 3N most of the time (4♣ did pop up too). So no bug, just another outlier.