Fact check:
The XMPP core protocol is split into three RFCs, one describing the fundamental building blocks of the protocol (RFC 6120, "XMPP core") and the second describing how to compose these for instant messaging (RFC 6121, "XMPP IM"). Finally, there is an additional short RFC which describes the address syntax and rules (RFC 6122).
These documents are stable, and need to be updated rarely. Other things such as audio/video calls, file transfers, push notifications and such, are defined by the XMPP Standards Foundation (XSF), as I imagine you know, in documents known as "XEPs". These have enabled the evolution of the protocol over the years, and are why XMPP didn't get stuck frozen in time in the early 2000s.
People like to point at the "5 million" XEPs (actually, 468). However the truth is that these are not required reading. They are not all applicable to IM clients (e.g. XMPP is used in IoT and other places), and many are simply proposals/experiments that didn't take off (XEPs have a lifecycle, including states such as "proposed" and "stable"). The XSF is a place of sharing and collaboration, and these proposals are a natural side-effect of that process.
The actual required reading is published annually in the "compliance suites" which describes what implementations are expected to be supporting in any given year. This promotes interoperability, encourages regular reviews of the ecosystem, and helps developers to discuss and plan their roadmaps. The 2022 document is here: https://xmpp.org/extensions/xep-0459.html
An alternative resource is the protocol overview on modernxmpp.org: https://docs.modernxmpp.org/client/protocol/
Why is all this relevant? Because, although I don't claim XMPP to be perfect (what software/protocol is?), I think this has been a pretty successful model that has stood the test of time and continues to deliver sensible community-driven evolution. The Matrix model is extremely similar (they have "MSCs" instead of "XEPs", and they regularly release versioned documents, though not strictly on an annual basis), IRC folk are trying something with IRCv3. Open protocols and ecosystems are hard work, but they are worth it.