https://www.deso.com/

An immense amount of engineering work has been put in to making it "scale" through the use of low-level use-case-specific optimizations, and reams of documentation has been written, carefully ignoring the fact that social data is truly immense, and that however it's optimized and compressed, it will ultimately scale to PETABYTES of new content PER DAY. The documentation handwaves at "sharding" to reach "1 billion users", but there's no protocol for discovering nodes that hold the data you're using.

Nostr shards from day 1, which has created a lot of problems for us. But at least we're starting with the right assumptions. But we desperately need to lean into the pain of locating relevant content so we can properly solve it over the long term.