Among the largest purple app accounts (ones with at least 3k followers).. here are the 100 most influential accounts (based on eigenvector centrality algorithm)

Among the largest purple app accounts (ones with at least 3k followers).. here are the 100 most influential accounts (based on eigenvector centrality algorithm) https://docs.google.com/spreadsheets/d/15Ndxoalc_Jjr7GX2h7weK-Z59K90JtcxmIiw67q1kR4

I like running, poker and oysters | Founder, Social Forensics (SocialForensics.com) | Berkeley & UPenn | Learning to listen

I think you're overemphasizing nodes within the largest communities while undervaluing influential nodes in smaller communities. Plus decay.

For example, Naomi/afrochicks isn't on the list.

hi im naomi i’m an athlete and creator!

join /actrackclub for fitness and /gmatcha for vibes

 - portrait.so/naomi

Reality is it's tough to offset the benefit received by accounts that were on the auto-follow list

Limiting eigenvector centrality algorithm to the network of accounts that have a minimum of 3k followers, however, helps a bit on that front

@afrochicks shakes out at 205 across this dataset; you're 116

Correct, this is simply looking at following/followers relationships

Temporal DAGs are useful when looking at engagements (e.g. predicting user interactions).. less so for following/followers relationships

To me, influence = ability to propagate information/content across a network.. and this captures that well imo

Oh you're just analyzing links? Have you tried temporal DAGs?