It seems insane to try to connect a twenty year shift in one global variable to one causal factor. This approach is why the social sciences continue to struggle to create understanding. It is all they can afford to do, though.
That said, 3) I think possibly best explains both: the increase in average number of friends due to influencer dynamics skewing the distribution, and the increase in polarization due to the tactics in social media.
However there seems to be no chance it is a durable or reproducible link, as it depends on the novelty of polarization techniques which wear over time and become known and integrated in education, reducing their effectiveness
That said, 3) I think possibly best explains both: the increase in average number of friends due to influencer dynamics skewing the distribution, and the increase in polarization due to the tactics in social media.
However there seems to be no chance it is a durable or reproducible link, as it depends on the novelty of polarization techniques which wear over time and become known and integrated in education, reducing their effectiveness