In any difficult field of endeavor -- ballet, mountaineering, violin-playing, physics -- there are going to be far more people in the world who have devoted a relatively small number of hours to the field than there are who have spent a big chunk of time to it. On physics.SE, non-novice users are in short supply. Is there any way to tell whether the number of currently active, non-novice users on physics.SE is going down or going up? We have certainly lost some non-novice users such as Arnold Neumaier and Ted Bunn.
I suppose a quantifiable way of defining this would be the number of high-rep users. Although rep doesn't really correlate very tightly with competence on an individual basis, and high-rep users may get suspended for being naughty, I would think that the number of high-rep users would be worth looking at on a statistical basis. Of course it won't tell us anything about whether we're retaining new non-novice users, such as ZeroTheHero, who appears to be a non-novice but just showed up today.