|
To use the topic that provoked this as an example, I would equate "rule of thumb" in this context with "zeroth-order approximation." If I measure a 5% fuel economy improvement on my already low-drag, CVT, electric-assist (all of which are differences from the cars which generated the RoT) car from a change and I assume that means I improved drag 10%, I have no basis on which to guess that. Maybe my car actually needs a 20% improvement in drag to get a 5% fuel economy improvement at 55mph, in which case increased resolution of the 10% guess doesn't improve its accuracy--it's still just a guess. I have absolutely no way of knowing on my car specifically without measuring something else to corroborate or disprove, and I could be way off as a result.
Put another way: If the RoT was a first-order approximation, we would have to know that drag improvement between 5%-14% on any car results in 5% fuel economy improvement, and we simply don't know that. The friction wasn't with the terminology; it's with the concept.
|