I mean... if you're saying there are studies that say people are initially attracted to people who look like themselves, and there as well are studies that show that people who look the same but are otherwise unrelated ARE more genetically similar than people who don't (the doppelganger study) then there seems to be a transitive association there, no?

If you really want to press on thier argument, I feel like the idea that the longevity of relationships between couples is based on how much they look alike when they met is very suspect.

I add "when they met", because there is also a phenomenon where people look more like thier partners the longer they've been together, and it's in large part that expression mirroring over decades actually subtly synchronizes thier facial musculatures and wrinkle patterns. So, you gotta be careful with the correlation specifically with long term partners. In the LONG TERM they might look alike because they've been together for a long time, rather than what buddy was suggesting which was they were together a long time because they look alike.