About half a year ago, I started pondering a question: how were my Facebook friendships changing over time?  Or, to be specific: who was unfriending me?

I decided to start taking monthly snapshots of my Facebook friend list, a task made easy by Facebook’s data-export feature.  I imagine a full history could be obtained using the API, but I haven’t explored that.

Since the first snapshot in December, I’ve added 19 friends and lost 10, for a net gain of 9. Many of those I lost have been mundane: a guy I talked to once during my first year at Stanford, a girl who was trying to get me to be her roommate, and several people whose names I didn’t recognize (yikes).  There were a few surprises, but I won’t out them here.

Change in Facebook friends over time

Other than the jump from December to January, when I moved to Colorado, my total number of friends has been pretty stable.  I seem to lose about 0.6% of my friends per month, but I’ve been gaining new friends at about the same rate.

This brings to mind several questions:

• If I were to stop adding Facebook friends, what would my friend count eventually settle at?
• How has my true friend count (as opposed to my Facebook “friend” count) been changing over time?
• What distributions best model the addition and subtraction of Facebook friends?
• How many of my remaining Facebook friends would I care about unfriending?

To try and address that final point, I went through and classified all of my 346 Facebook friends into two groups: people who would hurt my feelings if they unfriended me, and people who would not hurt my feelings.  Turns out that 140 of them fall into the “oh well” category, which leaves about 200 people that I’d really care about losing.

Interestingly, that’s very similar to Dunbar’s number.  Perhaps that’s the natural lower bound on Facebook friend decay.