I remember trying to explain why it wasn't a stupid idea to people way back when it launched on HN, and had kind of assumed that it'd died years ago.
----
Hypothesis: If you can find a way to provide value while you build up the network, the product would do better.
Second hypothesis: If you can find a way to acquire valuable users with a low-effort funnel and 'leech' users with a paid funnel that would also make the product perform better.
Taking these as assumptions for a moment, this would imply a natural advantage for federated networks if they provide interoperability between different services that have value on their own without network effects.