Big e-mail providers are primarily concerned with spam–which is defined by the volume and unsolicited nature of the messages, rather than the opinions they express. Sometimes their attempts at stopping it impose collateral damage, and their response to that collateral damage can be arbitrary and capricious–but that's a different issue from what we are talking about with Mastodon coordinated de-federation, which is much more intentional than collateral.

The terms of service of those big providers say that they can ban people for "hate speech", but in practice they rarely do that, and on the rather rare occasions they do, it is usually a particularly egregious case of it.

By contrast, the big Mastodon instances seem to be very keen on banning "hate speech" – and defining that term in a much broader way than most other platforms do. See https://joinmastodon.org/covenant point 1

> That's how it is on Reddit too no? You visit and post a comment on subreddit A, somehow mods in subreddit B will find out and pre-emptively ban you.

Almost no one runs a personal instance. Most people have accounts on multi user instances. This is unlike reddit, where accounts are global and independent of subreddits. On reddit, mods from subreddit B can preemptively ban you after they learn that subreddit A doesn't like you. But they can't ban entire subreddit, say C, so that every user that's subscribing to C can no longer interact with anyone subscribed to B. This is the scenario that plagues Mastodon - not individual blocks, but defederating whole instances.

> That's how it is on Reddit too no? You visit and post a comment on subreddit A, somehow mods in subreddit B will find out and pre-emptively ban you.

I've been avoiding Reddit recently, but when I used to use it – I never agreed with that kind of behaviour, and any subreddit which does it is one I don't want to be part of.

However, that said, that's community-level not instance-level, and so I'm not sure what that has to do with federation.