The follow on experiment was to eliminate the possibility that Irwin, the fake profile, was the cause why PYMK wasn’t triggering. So I asked Vicky to visit some of my friends that were not her friends to see whether that would trigger PYMK. She visited them each a different number of times, ranging from 1 visit up to 6 visits for each of my 6 friends. I then contacted each of my 6 friends to see whether Vicky showed up on any of their PYMK lists. I was surprised by the result:
- 2 out of 6 friends reported that they could not find the People You May Know section of their Facebook profile
- 3 out of 6 friends looked at least 5 pages back in the list and could not find Vicky in the PYMK list
- All 3 friends looked at least 2 days after Vicky looked at their Facebook page, so that rules out any time delay differences
So we couldn’t trigger the PYMK phenomenon from that either.
The last thing I tried to reproduce PYMK phenomenon was to go through my own real Facebook profile’s People You May Know list and try contacting the people to see if they had visited my profile. But it appears that the list is predominantly pulled from mutual friends that I have with the suggested Facebook user. So out of the first 100 suggestions Facebook presented, I found 16 suggested Facebook users that did not have any mutual friends with me. One of the 16 users got back to me and she said she did not visit my Facebook page prior to me contacting her, but she did visit me afterwards and said we had a mutual friend (despite the fact that I couldn’t see that on my PYMK list).
So to wrap up, we did 3 different experiments to try to trigger this phenomenon, and we could not. I can only speculate on the possibilities below to explain our findings:
- The number of times someone visited your Facebook profile at one time was part of the PYMK list generation algorithm, and now is not, or
- The anecdotal stories were merely coincidences where a Facebook user listed in the PYMK list happened to visit the author’s Facebook page at the same time
My conclusion based on the evidence is the latter, that the anecdotes can be explained by mere coincidence. In my Third Experiment that is confirmed when the Facebook user got back to me saying we had a mutual friend even though it wasn’t displayed on my end (probably due to privacy settings on that mutual friend’s profile).
If anything changes and I have found hard evidence to the contrary, I’ll update this post.