His tests don't seem to distinguish between H2 (recent visits by the friend) and H4 (recent interactions) very effectively. I could see H4 being a legitimate strategy but H2 is a bit more concerning.
Yeah, I think it's a mix of those two things - my list is composed of: a) people I interacted a lot recently b) people I interacted a lot in the past (and then almost not at all for a long time) c) people I haven't really interacted at all "c" must come from H2, "a" from H4 and that leaves "b" as either flaw in H4 or also a H2 - the latter would be a good news for me because people from that category are actually gir…
^^ false dilemma, no?