lostinlight, again — that's exactly the wrong thing. No one wants to follow random people. NO ONE. Most want to find people they already know. Finding someone with common interests is a very much secondary activity on social media.

Mastodon did have a thing where you connected your Twitter account and it matched your Twitter follows against its database of Twitter<->Mastodon connections. But this was a single point of failure: it ran on a single server, and it used a single app ID, that stopped working when Eugen deleted his Twitter account. They weren't able to get it working again.

Now, my idea is to avoid single points of failure at all costs, and most preferably extract contact lists out of centralized services without their consent — no way in hell they're giving one. So, if necessary, I'll have to resort to things like importing GDPR export archives, or even parsing their websites. APIs would work too, but only where API access doesn't require a manual pre-approval (for example VK).