[pescetom]

It seemed recently that for some reason the priority of GiB work was to enable claims rather than fix bugs, even at the expense of disabling an addition to GiB that the original developer retained essential.
So why does a rented Advanced 2/1 GiB not manage to claim here , if not at trick 10 at least at trick 11?

[smerriman]

pescetom, on 2025-October-15, 15:01, said:

It seemed recently that for some reason the priority of GiB work was to enable claims rather than fix bugs, even at the expense of disabling an addition to GiB that the original developer retained essential.

This seems very misguided.. The claim algorithm was specifically introduced to fix bugs.. e.g. the well known one where GIB has a 100% line of success, but goes down by taking a losing finesse it thought was guaranteed for other reasons. Having an independent algorithm that can look at a layout and determine if it can guarantee a certain number of tricks without having to resort to flawed Monte Carlo simulations seems a good starting point to fix this.

I don't know how that algorithm works, or why it doesn't catch your case (and that's definitely worth asking), and still plan to test the impact of GIBson myself, but if Lorand says the current implementation resolves enough bugs that it outperforms the version with GIBson, I believe him, even if it's somewhat surprising that it was so ineffective and results in some cases where it goes wrong now that it didn't use to..

[pescetom]

I can see a conservative algorithm failing to claim at trick 10 (when it is possible to voluntarily lose a trick, although even a Taliban TD would allow a claim of "all mine" here), but not once we have all the remaining trumps.