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

Marcus Rohrmoser writes
<https://lists.w3.org/Archives/Public/public-swicg/2023Apr/0059.html>:

> Designing for huge nodes' needs doesn't align with the 'fedi' nor the
> 'verse' if conflicts with small nodes' needs. And complexity is against
> small nodes.


Huge nodes already exist and it is inevitable that there will be more of
them. While it is essential that Fediverse protocols are easily implemented
for small nodes, it is also essential that the needs of large and even
"huge" nodes be addressed. If the full breadth of scaling needs are not
supported well, the Fediverse itself will not scale. The key truth in
Marcus' statement is that it is important that complexities introduced to
serve the needs of large nodes must not make it difficult to implement or
deploy a small node. Similarly, simplicities introduced in order to enable
deployment of small, simple nodes, must not be allowed to conflict with the
needs of larger nodes. While addressing a broad range of scales is
challenging, I believe that it is possible through the careful definition
of protocols, options, and extensions.

According to https://fediverse.observer/list there are now at least 14
Fediverse nodes with more than 100K users. The largest, Mastodon.social,
has over 1 million users. On the other hand, Facebook claims
<https://investor.fb.com/investor-news/press-release-details/2023/Meta-Reports-Fourth-Quarter-and-Full-Year-2022-Results/default.aspx#:~:text=Facebook%20daily%20active%20users,2%25%20year%2Dover%2Dyear.>
that during December 2022, they had 2.0 billion average daily users and
2.96  billion monthly active users. During the same month, Twitter claimed
<https://www.statista.com/statistics/303681/twitter-users-worldwide/#:~:text=368%20million%20monthly%20active%20users%20worldwide.>
368 million active users. So, scaling to the size of Twitter would take
over 300 nodes the size of Mastodon.social, or vastly more smaller nodes.
To scale to the size of Facebook would require at least 3,000 nodes with as
many users as Mastodon.social has today -- or, an even more vast number of
smaller nodes.

I suggest that the current protocols are neither optimal for supporting
thousands of million-member nodes, nor for supporting millions of smaller
nodes. Those who dream of a SocialWeb reliant on a Fediverse composed of
small nodes would be better served by seeking to ensure that small nodes
provide distinct value rather than resisting the protocol enhancements and
innovations that would address the needs of those building larger nodes.

Crippling our protocols to limit their utility to only small nodes will be
no more effective than are the many efforts to provide security via
obscurity. In any case, if the openly defined SocialWeb protocols do not
expand to handle the full breadth of requirements, we will soon find that
those who address those broader needs via more closed processes, such as
BlueSky, Meta, etc., will provide the preferred foundation for the future
SocialWeb.

If we want to ensure that federation works well for small SocialWeb nodes,
we'll need to also ensure that it works well for large ones.

bob wyman