[cavoirom]

> And every blue-stodon instance would have to scrape every single PDS. So it's a much much worse O(n²) issue, isn't it?

They could rely on some relays to build their AppViews. The flexibility is already there.

[kuschku]

I feel like you're missing the point of a decentralized protocol if at every step you suggest to use centralized services instead.

[cavoirom]

You can self host every component in the stack. You can create a new one for your use case.

[lmorchard]

And then once you've self-hosted ever component in the stack, you have one (1) Mastodon instance equivalent :)

[cavoirom]

Yes and no, it could scale to the whole ATProto user base.

[lmorchard]

Scaling is not useful if that's not your goal

[demux]

A mastodon instance is not useful if that's not your goal lol