@evan if the protocol you made is worth it's money (which I assume it is, but I didn't work out the details yet), its implementations should allow fetching state on demand (of the users' social circle).
If this is true and the implementations are reasonable, then I see no risk to the infrastructure whatsoever. Of course, people like me who want to preserve *everything* will have to buy more and more disk space over time, but it's pretty manageable.
@evan (sorry for so much text)!
To conclude: I also voted 100M+ in hopes that we'll some day get ActivityPub servers running on handheld devices in an edge computing setting. Because otherwise, only people who can afford to pay 15-50 buck a month for what essentially (for them) is a "twitter account" will have access as the big public instances close their doors under the load of the bills.
These are my hopes. But my fear is that big instances will start monetising (which we have already seen on that fork of masto which has implemented ad functionality).
@evan your second point is of large interest. I too believe that activity to ActivityPub and SMTP should be a human right, but it incurs a ton of costs. If you think about it, even SMTP isn't granted for free anywhere, but rather in exchange for the metadata of peoples' inboxes (how many governments issue E-Mail accounts to their citizens?).
Same will be with ActivityPub, except even more pronounced. I can maybe pay for 1-2 more active people on my instance, but it will increase the pressure on HDD linearly, and then I'd have to run tiered servers, one with persistence (mine! mwa-ha-ha!) and one without persistence (for the users beyond UID#5).
So hoping that "someone else will pay for the people" is a very dangerous hope, I feel. Governments aren't eager to, and companies will steal peoples' souls *snaps fingers* again. :(