[1eyedjack]

Sometimes when I play in Instant tourneys the historical bidding and play to date may be identical at my table to that which took place at other tables, and yet there is a divergence of what happens next.

As far as I can tell there are two possible explanations for this effect:

1) The GIB version at my table is a later version than that which was in force when the original tourney took place

and/or

2) The random number seed used for sampling possible actions by GIB may be different at my table than that which was used at other tables.

There is nothing that can be done about cause 1 above (the differences in GIB versions). Or rather, the problem/effect has been fully aired, and anything practical is being done, or has been done, or maybe will be done, or has been dismissed as the lesser evil.

But I wonder whether, in some instances, use of the same random number seed (cause 2) would generate the same action by GIBs despite being of differing versions and, if so, whether steps are currently taken (or, if not, might be taken) to record the seed used in historical tourneys and re-use them in the Instants, to try to reduce (while maybe not eliminating) the incidence of the effect of identical history of a hand generating diverging action. Each new update to GIB does not, after all, cause it to change every sequence. Quite a few sequences remain unchanged with GIB updates, which leaves just the random number seed as the explanation for divergence in those cases.

Maybe you are already using the same random number seed and cause 1 is the only explanation for divergences, in which case ignore this suggestion (although it would be good to know).

[barmar]

We haven't changed the random number seed in a long time. Explanation 1 is most likely the reason.