For critical time dependent systems, I absolutely agree. I was unclear in that I was referring to far more mundane issues where slight skew can be problematic.

For critical time dependent systems, I absolutely agree. I was unclear in that I was referring to far more mundane issues where slight skew can be problematic. I've seen telemetry systems in which the aformentioned deltas would be computed by doing a timedelta() on the recorded timestamp from each machine. With even subsecond skew, you can get some really interesting artifacts; negative durations, that sort of thing.