Even distributed protocols aren't a silver bullet. Embrace and extend can make short work of that. As can joining standards boards and playing political games. As can simply turning off federation.

Sometimes I think HTTP and SMTP are only limping along because there's a large enough enterprise base that doesn't want to see any changes at all.

Maybe that presents a defense (get your competing protocol entrenched in corporate IT), but not many technologies get birthed out to the consumer space from the enterprise space the way http and smtp did.

I think the problem is that historically, we only ever collectively solved the email server. It's unfortunate that we didn't solve the more general case of communication through a server hub. We shouldn't have email addresses, we should have connection addresses, that provide access to mail, chat, audio, video, whatever.

There's no such thing as a decentralized conmunications tool. Even email requires the centralization provided by ICANN.

You can send email without using DNS.

I'm not sure I'd call IP addresses centralised, if perhaps you were thinking of ICANN's running of IANA?

If you consider the internet inherently "centralised" then surely you'd still find CB radio or similar to be decentralised. Or do you also consider the allocation of wavebands to be "centralised"?

Depends on how far you want to push the definition of decentralized. The routing infrastructure of the Internet is still fairly centralized as is the allocation of IP addresses.

Making a messaging service successful requires huge amounts of time and effort. If it's so decentralized that it cannot even be acquired, how could it get built?

To be clear, I don't doubt that the technical side is very feasible. It's the user acquisition side that seems improbable except as a commercial venture.

I suppose that is its own form of decentralization.