Re: Federation by both Large and Small servers must be well supported.

Marcus,
I'm somewhat perplexed by your concerns about Websockets. There are many
examples of systems that have found WebSockets to be useful. For instance:
Nostr relies heavily on WebSockets and the folk who are implementing the
many Nostr clients and relays don't seem to be particularly concerned
either with scaling issues or with any "complexity" due to WebSockets use.
In fact, when compared to ActivityPub, Nostr is simple
<https://github.com/nostr-protocol/nips/blob/master/01.md> to the point of
being trivial. It isn't much more than a pure PubSub protocol (e.g. very
much like WebSub <https://www.w3.org/TR/websub/>). Nostr's usage seems to
be growing -- if their published usage statistics
<https://stats.nostr.band/> are to be trusted.

You asked that I elaborate on: "small nodes would be better served by
seeking to ensure that small nodes provide distinct value"
What I'm suggesting is that since large nodes already exist, and since the
presence of large nodes leads almost inevitably to concerns about "embrace
and capture," if we truly wish to preserve an important role for small
nodes, we should be trying to clearly articulate and demonstrate the
concrete end-user benefits that small nodes provide. If such benefits
cannot be demonstrated, there will be no motivation for users to prefer
small over large nodes. Thus, any federation that exists will, over time,
be seen to do little more than provide unnecessary, unproductive friction.
Over time, sources of friction are removed.

If the only argument for small nodes is that they prevent the largely
theoretical problems that arise with large nodes, then I fear that most
people's response will be similar to that of those who heard, during the
1930's or earlier, that continued use of fossil fuels would one day present
a challenge for our climate. The almost universal response was:
"Fascinating issue. We'll worry about it when it happens..."

You suggest that Meta, etc. "are not social today and won't be tomorrow. Be
it with our help or without."
Well, quite literally *billions* of people consider them to be "social"
applications today, whether or not they support anything that one might
consider to be a SocialWeb protocol. Arguing fine theoretical definitions
of what is or is not social is of limited utility. The reality is that
Meta/Facebook, and other large "social" apps, could choose to embrace and
capture the social web at any time. Just as AOL's unleashing of its hordes
of users onto the Internet had a massive impact back in the 1990's, if
Facebook or Twitter were to connect to the Fediverse today, the impact
would be dramatic and lasting. Some may argue that their own instances
would block users from mega-nodes, but the reality is that a probably
larger number of instance operators would find it in their users' interest
to allow connecting with those who primarily used Facebook or Twitter as
their primary interfaces.

bob wyman