GIB Takes Pointless Finesse

[eagles123]

[hv=http://www.bridgebase.com/tools/handviewer.html?bbo=y&lin=pn|eagles123,~~M13109ko,~~M2607n8g,~~M70v3m2e|st%7C%7Cmd%7C4S2789THD8JC23568A%2CS456AH37QKD25QACQ%2CS3QH69TD3479TKC47%2C%7Crh%7C%7Cah%7CBoard%202%7Csv%7Cn%7Cmb%7C1H%7Can%7CMajor%20suit%20opening%20--%205%2B%20%21H%3B%2011-21%20HCP%3B%2012-22%20total%20points%20%7Cmb%7C1S%7Can%7COne-level%20overcall%20--%205%2B%20%21S%3B%208-17%20HCP%3B%209%7Cmb%7C2S%7Can%7CCue%3A%20limit%20raise%20or%20better%20--%203%2B%20%21H%3B%2011%2B%20total%20points%20%7Cmb%7Cp%7Cmb%7C4H%7Can%7C5%2B%20%21H%3B%2013%2B%20HCP%3B%2014-19%20total%20points%20%7Cmb%7Cp%7Cmb%7C4N%7Can%7CBlackwood%20%28H%29%20--%203%2B%20%21H%3B%2018%2B%20total%20points%20%7Cmb%7Cp%7Cmb%7C5D%7Can%7COne%20or%20four%20key%20cards%20--%205%2B%20%21H%3B%2013%2B%20HCP%3B%2014-19%20total%20points%20%7Cmb%7Cp%7Cmb%7C6H%7Can%7C3%2B%20%21H%3B%2018%2B%20total%20points%20%7Cmb%7Cp%7Cmb%7Cp%7Cmb%7Cp%7Cpc%7CCA%7Cpc%7CCQ%7Cpc%7CC4%7Cpc%7CCK%7Cpc%7CD8%7Cpc%7CDQ%7Cpc%7CDK%7Cpc%7CD6%7Cpc%7CDT%7Cpc%7CHJ%7Cpc%7CDJ%7Cpc%7CD2%7Cpc%7CH5%7Cpc%7CC2%7Cpc%7CHK%7Cpc%7CH6%7Cpc%7CH7%7Cpc%7CH9%7Cpc%7CHA%7Cpc%7CC3%7Cpc%7CH8%7Cpc%7CC5%7Cpc%7CHQ%7Cpc%7CHT%7Cpc%7CS6%7Cpc%7CS3%7Cpc%7CSK%7Cpc%7CS2%7Cpc%7CH4%7Cpc%7CC6%7Cpc%7CH3%7Cpc%7CD4%7Cpc%7CH2%7Cpc%7CC8%7Cpc%7CS4%7Cpc%7CD9%7Cpc%7CCJ%7Cpc%7CS7%7Cpc%7CS5%7Cpc%7CC7%7Cpc%7CC9%7Cpc%7CS8%7Cpc%7CD5%7Cpc%7CD7%7Cpc%7CCT%7Cpc%7CS9%7Cpc%7CDA%7Cpc%7CD3%7Cpc%7CSJ%7Cpc%7CST%7Cpc%7CSA%7Cpc%7CSQ%7C]300|400[/hv]

[1eyedjack]

Yes it does this a lot. Blind faith in opponents' bidding within described constraints.

 

It might as well take a practice finesse that it perceives to be 100% odds-on to succeed, as you cannot have the 8+ HCP "promised" for your 1S overcall unless you hold the Diamond King.

[iandayre]

I am sure that Jack is correct. And I have heard that it is difficult, if not impossible, to change GIB's card play parameters. But it seems to me that GIB can never be taken seriously as a bridge-playing program if it makes insane errors like this. Is it really impossible to program GIB to take its certain tricks when those amount to ALL the remaining tricks? Why should it ever take the slightest unnecessary chance?

 

I imagine it is more than just GIB choosing between "equal" plays. Did all GIB declarers in this contract play the same way? This implies that the programming makes it somehow more attractive to take the unnecessary finesse, as if to prove GIB can really count points accurately.

[johnu]

GIB recovered nicely by not taking the practice spade finesse later in the hand. :)

[Bbradley62]

I imagine it is more than just GIB choosing between "equal" plays. Did all GIB declarers in this contract play the same way? This implies that the programming makes it somehow more attractive to take the unnecessary finesse, as if to prove GIB can really count points accurately.

This hand was played in MBC. Have settings been changed such that GIB now uses the same seed number for a particular board at all tables? At one time the situation was that GIB would behave consistently given consistent circumstances in tournaments but not necessarily in MBC.

 

All GIB-Easts who declared 6♥ in uncontested auctions made their contracts. All 3 whose South overcalled 1♠ took the unnecessary finesse. At two tables, South bid Michaels instead of 1♠; at one of them South didn't lead ♣A and declarer made his contract, but and at the table where ♣A was led, East again took the unnecessary finesse. If the seed numbers are still random, this may be enough empirical evidence to suggest that GIB "prefers" taking the unnecessary finesse, otherwise not enough to conclude either way.

[kuhchung]

Sorry, I'm late to this thread. Is this a basic bot?

 

I can't recall a premium bot making this kind of play, but obviously my memory may be faulty.

 

I remember screwing around with "Just Play Bridge" and GIB was declaring 3NT with a suit combo of KQJx opposite AT8. It ran the 8, losing. It was literally the only way the bot could have lost a trick in the suit. I laughed a lot.