Facets and Social Networks |
November 26th, 2024 |
bluesky, tech |
Narrow: pick a topic and make it your beat. People who care about that can follow you and pretty reliably see what they're looking for.
Broad: write about whatever you like. Some readers will be willing to scroll past posts on uninteresting topics, others will decide the combination is too noisy to be worth it to them.
Current social network technology strongly favors the former: if you write about just one area it's much easier for algorithms to figure out who to show your posts to. And so you see a lot of advice to build your personal brand about an area: write about cooking, or housing policy, or military history.
This is very much not for me. I want to write about whatever I want to write about, which is a lot of different things, and I've generally just accepted that this is a bad fit for Facebook and the other places people read my writing. But two exceptions:
- Julia and I have a group for kid stuff, because it seems to be especially polarizing.
- With RSS, I (and others) have single-topic feeds.
If this were just a me problem then it wouldn't be too bad, but this seems pretty big to me. In trying out Bluesky I'm running into a bunch of accounts that post about a range of things I find variously interesting. I can decide whether they're worth it overall, but this is not the right choice for the technology to be forcing me to make.
Instead, I've long wanted a social network built around the idea that each person's identity and interests have many facets, and tries to match specific posts with the people that would be interested in them.
I wonder if, at a time when advances in AI are making this kind of classification problem easier and there's more social networking competition than there has been for a while, someone might want to take this on? Perhaps Bluesky's custom feeds would be a good way to play with this?
(I don't think this can depend on people tagging their own posts, because people are generally lazy. But something based on classifying the post based on its content and who has liked it so far seems pretty promising.)
Comment via: facebook, lesswrong, mastodon, bluesky