It would be interesting data to know what the typical distribution of hobbies would be like. I suspect most folks are like +Mark Ivey
's distribution: Try a lot of things, some things stick for month, some stick for years, some are long term but sporadic. Like I just picked up yo-yoing again after not doing it for about 10 years. Not surprisingly, it was a chance social interaction with a member of the existing community that sparked my re-engagement.
In my design process, I expect the funnel to drop most people out. But I want to be sure that there's enough there that they could keep advancing if they fell in love. So you end up designing on multiple levels: Unaware, aware, trial, repeat user, long term user, lapse user.