Bluesky's Hand-Picked Friend Lists Beat Algorithms, Study Finds
Your curated follow lists may be smarter than the algorithm — here's why.
Bluesky is a social network where, instead of an algorithm deciding who you see, users can build "starter packs" — hand-picked lists of accounts someone suggests you follow, usually to help newcomers find their people. A new study looked at 2,718 active packs containing roughly 144,000 accounts, sorting them into 17 topic areas like news, art, or science, and then checked whether the lists actually made sense.
They did. People grouped in the same pack posted about similar subjects, and their posts matched the pack's own description — meaning the curator wasn't just grabbing random popular accounts. Curators also tended to resemble the people they picked, at least in interests. And the social bonding was real: members of the same pack followed each other back far more often than they followed strangers, and they shared large overlapping chunks of the same audience. In plain terms, a good starter pack really does behave like a genuine clique, not a random pile of names.
There's a catch, though. Even inside a tight-knit pack, the average member's followers only followed a small fraction of the other people on that list. So joining one pack won't suddenly make you visible to everyone in that community — the lists create clusters, but they leave plenty of connections unmade.
The takeaway: humans are surprisingly good at grouping people who belong together, and platforms might get better results by blending human picks with machine recommendations instead of choosing one or the other. For you, that could mean follow suggestions that feel less random, faster discovery of real communities, and less time trapped in an algorithmic feed built to keep you scrolling.
- Starter packs (hand-picked lists of accounts to follow) are topically tight — members post about the same things as the list's description.
- Packing works socially too: 2,718 packs and 144,000 accounts showed members follow each other back and share overlapping audiences far more than random users do.
- The limit: the average member's followers only follow a small slice of the pack, so one list won't make you widely seen.
Why It Matters
Human-made follow lists could mean better recommendations and faster discovery of real communities — with less algorithmic noise.