When games are replayed, the seed remains the same. Usually this is what you want, but sometimes when testing different strategies abnormal luck or unluck with a particular seed can manifest itself and warp the conclusions of some testing. One fix is to manually reenter in all of the cards instead of queuing up a game, but I propose an easier solution: A random seed checkbox. It could be unchecked by default when queuing up an old game.
This would make replaying games much more useful. Currently I can't tell how much my strategy is due to luck. If I could play the same game with a different random number seed, then I could get some idea what role luck plays.