I fear if these federated systems do grow popular enough

If an instance did grow "too big to moderate", it would surely be defederated from, yeah? I'm struggling to think of a situation where responsible admins from well-moderated instances would willingly subject their users to spammers from an instance (no matter how big) that can't control itself.

email spam is the original spam

While it's true that there was occasional commercial misuse of email in the ARPANET days (when commercial use was against the rules of the military-funded research network), it wasn't called "spam" then.

Until the mid-1990s, "spamming" typically meant sending repetitious messages rather than inappropriate commercial messages. It wasn't about what you said, but about how many times you said it. The transition from one meaning to the other mostly happened on Usenet, as commercial abusers took advantage of typically-lax moderation policies to repeatedly post unsolicited advertisements. Major commercial email spam was a branch off of Usenet spam operations.

• 1985: "spamming" on MUDs meant sending junk messages to disrupt a roleplaying session, originally from a player doing this with the text of the Monty Python "Spam" sketch.
• 1991: When a Usenet modbot had a bug that caused it to repeatedly post the same message, a Usenet admin who was also a MUD player referred to this as "spamming" Usenet. The term caught on to mean "excessive multiple posting", regardless of content; most early Usenet spams were religious proselytization or political kookery.
• 1994: Lawyers Canter & Siegel post the first major commercial Usenet spam. They go on to write a book promoting Usenet and email spamming as a business tactic. At this point, "spamming" starts to be used to refer to inappropriate commercial posting, regardless of volume.

I look at that as as proof it wasn't written by GPT.