[goffster]

https://www.bridgeba...H9%7Cpc%7CD8%7C

Win spade ace, club to king, dump diamond for a more normalized result.

Instead it makes a perhaps good matchpoint play of the club to the 8, enabling the possibility of perhaps 3 independent finesses.

[steve2005]

from the bidding pretty good odds North has Q♣
Gib does double dummy so takes the finesse

[smerriman]

If the subject was a literal question, then yes, robots play differently based on the scoring - their whole algorithm is based on choosing the play that leads to the best average score for the deal (where the score is calculated in IMPs or MPs, as a crosstable against its other possible choices).

Of course, it's not great at MPs due to the fact it doesn't know when it's "ahead", so will always go for a 51% overtrick even if it were on track for a great score anyway.

steve2005, on 2024-August-19, 18:44, said:

from the bidding pretty good odds North has Q♣
Gib does double dummy so takes the finesse

I would say the opposite; a vacant spaces argument outweighs the HCP, so South is favorite to hold the queen. Even if it weren't, a basic robot which only looks at the values of cards double dummy will play the K every time, because it will know when to drop Qx from North, and Qxx+ is definitely against the odds.

However, this came from a paid tournament, so this was an advanced robot which was playing single dummy, thus knows it has to make decisions on what layouts to play for in advance. The older version of GIB is torn between the two options and varies between them almost 50/50 each time I run it. Does seem a very close call, given playing the K will in some cases leave you stuck without the ability to finesse in trumps or clubs later, which may or may not be enough to make up for the finesse being the slight underdog. (At IMPs, it plays the K every time, of course.)