Are you really not able to think of any dead social networks that deleted all of their content? Hell: Google alone has managed to delete at least two!

Storage costs have historically dropped so fast that this just isn't an issue I'd worry about. I ran a mail provider with ~2m accounts around 2000. We had less aggregate storage than my laptop has on a single M.2 SSD now. With redundancy, the cost per GB for us at that time was around 10,000 times higher than it'd have been today. Our total processing power was lower than my laptop. Our bandwidth use was a tiny fraction of my home internet connection.

In other words: If it's an issue today, it soon won't be. Costs are low enough per user today to be viable, and they'll only drop.

> You do not have to generally worry about your entire account and content being gone because some mod gave up.

I do have to worry about what happens when I have to start again because engagement is cratering and I can't just migrate elsewhere or run my own, though (e.g. I get more engagement on Mastodon than on Twitter despite 100x as many followers on one of my Twitter accounts). Since I can (and did) choose to run my own instance, I don't need to worry about that again.

> The idea that a bunch of enthusiasts can replicate this, is misplaced.

This was the kind of argument used against open source 20-30 years ago. It's was just as ridiculous an argument then. This is magnitudes easier than what was achieved with open source, and it's also flawed because just as detractors of open source you're presuming that it will only be enthusiasts, and that e.g. nobody will start and operate commercial services for those who prefer that (some commercial hosters already exists for Mastodon for example).

Maybe?

> and this should be considered fact.

I see absolutely nothing about ActivityPub that is inherently hard to scale. Adding more points of presence & smart fan-out seems like it could keep scaling indefinitely, from what I can see.

> Costs will forever go up

Let's simplify follower-ship and assume instead a simpler model of bi-directional friending. If you ahve M users each of which have N total friends, the naive fear is this: that costs will keep growing terrifyingly. But I expect it's actually more a sigmoid curve. As you grow into millions and especially many-millions of users, more and more you'll have your people following the same person - which reduces traffic - and more and more of your people followed by multiple people on another server - which reduces traffic. The actual growth curve here is more sigmoidal, with a big variability depending on the number of active instances fediverse instances out there.

> You do not have to generally worry about your entire account and content being gone because some mod gave up.

To get listed on joinmastadon you have to promise to give at least 3 months heads-up, during which time users have to be able to activate an account transfer to other systems. There was a case where mastodon.au nearly pulled the plug with a short notice, but someone else stepped up to take over the system.

The potential for this to be a huge problem is absolutely super real, but in practice, I have almost no real concern over this and think it should be broadly disregarded. The server I was on shut down, but the mod gave us almost a full year of continued service, and another year of read-only service during which we can transfer. Trying to drum up fear & doubt over the instability of this system seems irresponsible & premature, given how well things have gone.

> The idea that a bunch of enthusiasts can replicate this, is misplaced.

I agree that there are huge advantages & I think we absolutely have taken much for granted. Right now you're looking at this through shit-colored glasses though, and I think it's naive to wish so hard for this all to fail. It's convenient & easy to say it'll never work.

But there's so many strategies where we can start to turn the scale into something useful. If servers publish WebBundles of user's feeds, users can p2p distribute the signed http content among themselves, perhaps via WebTorrent. Rather than live in fear of our users and our scaling, we can rely on our users to help us tackle scaling. Yes we can, man. It's totally doable. Get off the ledge.