[toast1]

I've noticed several cases recently where the GIBs have taken different actions, in identical situations, at both tables of a Robot Challenge.
(Robot Challenge is me playing with 3 robots at one table, with scores compared vs. a virtual table of 4 robots.)
For example, the auction is the same at both tables, but GIB chooses a different opening lead at my table than at the other 'virtual' table.
Or, the auction goes 1D-(2C)-to GIB, and at my table it makes a negative double, and at the other table it passes.

I don't remember this happening before -- indeed, I thought that it should not, based on how GIB works.
Could it have something to do with the rollout of the new GIBBO?

[lorserker]

the hands at the 'virtual' table are played in batch upfront.
the reason is that the robots at the other table were still the old ones and you had the new ones at your table.
this will resolve soon as the deal pool will refresh.

[dB451]

Well, I HAVE noticed it happening before - long before the development of GIBBO. So, we have yet to hear a reconciliation of that observation with their statement that the robots - whether GIBBO or earlier - are completely deterministic. I didn't save the evidence (although some may still be visible), but it was VERY clearly the case. Can anyone identify a specific case prior to GIBBO?