The problem with hypothetical risks is, when do they become rational and when do they pass into the boundary of the irrational?

It's not useful to continue because [the software] clearly is not stable.