> Most people don't expect upper bounds, though, when talking about estimations.
What do you mean? If all I know about a number X is that it is an estimate for Y, I would expect X to be an upper bound for Y 50% of the time and a lower bound 50% of the time.