Even if there are 1 million independent email servers, if Google, Microsoft, Amazon, and Apple send, receive, or pass along 98-99% of emails, there's still a fair bit of centralization in the network which should be acknowledged, though this is admittedly not the same as the protocol being inherently centralized.
How do you messure this? Every university in my country has thousands of mail accounts that are constantly used and having worked at an university IT department I have never heard of any numbers collection coming by to ask us how many mails our users sent to each other.
I wouldn't be surprised if that percentage was made up by the big ones to give you the feeling that self hosting mailservers is useless.
You're probably right, I hadn't considered the volume of mail being sent within institutions that might host their own mailservers, so maybe we should say more than like 50-60%.
Is it still a good example of federation though if a vast majority (99%?) of mails between such institutions or outside of orgs large enough to have a campus, tend to go through Microsoft/Google/Apple/Amazon?
