It clearly depends on the nature of the event. In what limited modeling I have done of optimal TT pacing strategies, one's entire MP/duration curve comes into play, not just FT (1hr power). There are realistic course and condition (wind) scenarios where the optimal pacing strategy calls for, say, a 5 min segment at a pretty high power relative to FT (e.g., FT+50w). But, even though one can theoretically ride the segment at a certain power and still finish the event within the NP constraiint, it's not an option unless the specific power/duration combo is on the cyclist's MP/duration curve. And, while it seems pretty clear that most of the MP/duration curve derives directly from FT, that connection gets more fuzzy as you shorten the duration, especially <10mins. I can even model scenarios where a rider beats another rider who has a higher FT, both of whom are riding optimal pacing strategies on the same course under the same conditions, if the rider with the lower FT has a little different MP/duration curve at the low end (e.g., <10mins).