I've been starting to realize that I don't think I can sustain the free relay offchain.pub by myself long-term, even with funds from the pay-to-relay bitcoiner.social. Nostr relays will need to host files for applications like "github on nostr" and at a certain point I will need to experiment with non-free models for that to work.

This is based on my naive/late realization yesterday that you can trivially include base64 encoded data in a nostr event. That use case will accelerate the server capacity demand growth, probably by A LOT.

A couple things might give me time to delay ending the free model:

- aggressively expiring events off of the relay

- blocking base64 data, which is text like this: Qm9yZWQgYXBlIHNoaXQK

Both options really suck and I don't like the idea of doing either really. Frankly, I really doubt blocking it is possible without at least a high cost in terms of compute. I mean how many cpu cycles do you think it'll take to regex that? Nevermind the inevitable porn-on-nostr that is now going to be hosted by the relay itself, and the operator liability concerns that come along with hosting files. I do want that level of freedom for us all, but if that's the game we're playing I need to spin off another nym for that and do it on a whole new relay for my own opsec.

Anyway, I'm not doing anything imminently, yet. Just thinking out loud here.