Both of those, and searching for a solution to a problem which doesn't exist. Nobody can tell reliably whether a cadence variation of 10% affects biomechanical efficiency, so there is no call for a gear system with jumps smaller than that. CVT probably does make it easier to automate gear changing, for people too stupid to work out for themselves when they're slogging/spinning out.
Ah, but it's the thought of smooth gear changes while accelerating hard that gets me dreaming of CVTs.
Ah, but it's the thought of smooth gear changes while accelerating hard that gets me dreaming of CVTs.