ActivityPub spec is a marvelous bit of tech writing. It’s also a pretty amazing example of HTTP’s enduring resilience as a protocol - tons of relatively simple things exposed as objects, with a well written spec for each. I’m glad I read it, because it really reframed what we’re actually doing here (basically posting json to our webserver, and getting copies of each others shit dropped off to shared inboxes across our federated server list). It’s an elegant design.
All that said, it’s going to dramatically fall down if it gets anywhere close to twitters scale - unless it recentralizes very quickly. Basically a small number of super-servers will have to appear, because the infrastructure to run the protocol will require it. Huge amounts of storage, duplication everywhere, it’s going to get crazy quickly. I don’t think that’s unique to activitypub, tho.

@adamhjk

I don't see yet why the duplication inherent in Activitypub will become crazy or prohibitive.

Each single person will only be able to read so much, given human reading speed.

By today's standard, relatively small storage can store what a single person can read in a life's time. So servers that cater to single people are quite feasible, storage-wise.

If several people team up and use the same server, storage requirement per person drops.

Do I have any fault in that reasoning?

@dj3ei the flaw is that the local timeline will be anemic. Discovery drops. And the hyper connectors that drive viral conversations will degrade, since they’ll cause a flood. Imagine an account with millions of followers - server inboxes will help, but only a little. But yes, storing your own data is feasible if you’re capable of caring and feeding it.

@adamhjk

The average Fediverse user follows a limited number of others, probably < 200.

So the average person is also only followed by that number of others. (Most less. Very few many more. Those few will need special solutions, yes.)

The *average* load a person poses to the Fediverse network when exchanging text messages is quite manageable, both storage and traffic. Modest computing power easily handles a lifetime's text feed.

The game changes when videos are involved, yes: Links FTW!

@dj3ei 25k people on hachyderm. I had 15.5k twitter followers. I was small time, kelsey Hightower has 193.2k. The flood of outbound traffic on a single Kelsey Hightower publish has to be crazy town. Especially if he starts putting media in.

@adamhjk
How many of those 25k people on hachyderm.io have more than 200 followers? Better yet: Do you have a chart "number of followers" vs. "percentage of users that have at least that many followers"? I'd find that very interesting!

I wonder whether hachyderm is already too large and should be split up. I imagine its local instance timeline has become unfollowable a long time ago. In other respects, too, it may have escaped human scale. What do you think?

@dj3ei

@adamhjk

N=1 here at #hachyderm: I've regularly dipped into /local for the past week and it has offered plenty of value at 30k users. It's more a creek with a strong diurnal pulse than a raging torrent. Easily used for serendipitous social interactions and community "events".

Consider user's read-to-write ratio, & average non-reply posts per day. And, more intentional use of unlisted (vs global) replies helps keep public timelines at "human scale".

Thanks, @xian ! I stand corrected! Apparently, even 30k local users do not (as I was afraid) make the local timeline unusable.

@adamhjk