That doesn't scale for public broadcast. You need the ability to distribute user data to a lot of services across a wide network.

Shouldnt the data repository be approx the same size as what would be returned by a batch XRPC endpoint? Don’t Crawling Indexers solve the distribution problem? The internet is nominally ip-addressed but CDNs in practice make it content addressed. Creating 2 sources of truth seems bad (xrpc, repo).