First, a rhetorical introduction.
Success can breed hubris, as we all know. Granted, one of our goals is to increase our certainty, yet no amount of success reduces uncertainty.
Engineers know this and get some comfort from their lab efforts, namely testing. Managers? Well, we have to guard against their paths to perdition (of several types, but an example is not knowing about near-zero).
Second, the message.
Fortunately, people have thought about the technical aspects of these things for some time, yet some of the more important views are considered as unnecessary and as being outside of practical importance. But, their consideration keeps us honest.
Any engineering view that would take computational output (we'll elaborate more on this) as equivalent to some natural event (observation and measurement thereof, okay?) has a very problematic basis, and we are talking much more than just quasi-empirical issues here. That this might come into acceptance is a natural result of advanced computing, its successes, and misunderstanding of what the successes (more luck than not) mean.
We'll be addressing these things more regularly, hopefully with real-world events to point to.
3 days ago