Heheh.
How do we know the simulation didn't segfault, stay down for 4 "real years," and then reboot again to leave us all right where we left off just 3 micro seconds ago? It may have rebooted while I was typing this message for all we know.
These conversations are best discussed in a hot tub with whiskey.
Our simulation got forked in real time so that different outcomes could be tested.
It supports the multiverse / parallel universe theory too :)
I've thought about that more than a few times. Save state, introduce new parameters, resume testing. Maybe each major leap in the laws of physics was actually a parameter change that we picked up on really quickly...
Try not to drop your smartphone in the hot tub when you're refreshing the comments page...