I think Shigolch brings up a very good point. Having access to recent rollouts gives away too much information about pending cube decisions.
That said, seeing a rollout shortly after the voting closes is very useful and I wouldn't want to have to wait two months (some games take that long) to see a rollout. Some of the best discussion has occurred after the rollouts were posted and I wouldn't want that to go away.
Is there a compromise? I think there is.
When we're on-roll, having access to a rollout of the box's latest move would tell us whether we had a double or not. Having access to a rollout of our last move would give us less information, and so on. I don't think it's necessary to go all the way through to the end of the game before the rollout information on earlier moves becomes useless to tell you what to do next. The question is: when does the information become sufficiently "cold" that it doesn't influence our pending decisions. When it reaches that level, posting the bot results is harmless.
One way to approach it is to simply wait N moves before posting the rollout. I'm not sure what the optimal value of N is, but N=4 (2 moves each) would probably work for most situations. That would mean waiting roughly a week for bot results - not too bad.
Another approach: have the moderator look at the bot results and the position and withhold the bot info until it's no longer useful to the pending decisions*. This would require that the moderator not participate in the play which would take a bit of the fun out of it for me, but I'm willing to go that route if other mods are, and especially if others are willing to volunteer to help out with the moderation duties .
There are some other good reasons to have the moderator recuse himself from the play - the bot can flag when a cube decision should be asked (we've missed a few of these) and also help with options for the voting. And then there's the conflict of interest thing...not that any of us would cheat for the team, but maybe it's better to not have that temptation.
I'm curious about what others think. If there are other proposed solutions, let's here them. I'm away for a couple of days, so don't be offended if I don't respond right away.
*for instance, posting relative equities (e.g. move A is the favorite, move B is a .03 error, etc) give away info that would be useful for making cube decisions.