That this is true is a bit of a tautology. Assuming the graph is monotonically decreasing sounds unlikely to me, though; I think there's likely to be multiple peaks - like the proverbial "7-year itch" for marriages.

Another graph that would be relevant would be the likelihood of people who have been exposed to facebook through their friends, media saturation and advertising, yet still haven't joined, to be tempted to join later. In other words - facebook has already saturated the internet world's radar, and people who haven't joined already are completely aware of its existance, function, and benefits, and still not joining. Those will be very expensive users to get, so IMO retention should be their major focus. I stayed through the "6-month updates-full-of-farmville problem", because they fixed it, they barely kept me through the "1-and-1/2-year privacy panic" after backing off on some portion of their stridency (and profitability per user), and they lost me completely, along with a number of other people I know, after the "2-year bored-to-shit-with-people-I-went-to-HS-with-telling-me-where-they-went-to-lunch-and-what-they-think-of-Obama lull" which I'm not sure they can do anything about.

If a large cohort of similarly aged users hit some peak that facebook is unable or unwilling to iron out at the same time, and there's no supply of cheap "Wow, there's a place on the internet where I can keep track of all of my college buddies!" users to replace them, there could be a precipitous fall.