You can publish any old crap. Microsoft's crappy file sharing protocol has an RFC. At least one of the ludicrous "IPv6 is crap, we should just use IPv4 but with bigger numbers" proposals has an RFC. [This can't work, the numbers in IPv4 are in defined bit-level structures, "just" having bigger numbers is nonsense without a new protocol]
From the IETF's point of view all this does is use up a few kB of storage in the RFC Editor servers, and hey, maybe someone will find it useful. It usually makes cranks or corporate types go away and stop wasting everybody's time.
If you're thinking "Wait, so how do I know if RFCs matter and I should care?" I have two answers
1. The pragmatic answer. If you're reading about an RFC because everybody does this and you need to do it too, then I guess it mattered after all. You can decide you don't care about RFC 822 and you'll use email headers starting with an exclamation mark and they'll be in the form of a list of headings and then a separate list of values. But your method won't interoperate with anybody else's, so you'll be talking to yourself.
2. The textbook answer. The IETF marks its Standards Track documents with their Standards Track status, e.g. "Internet Standard" or "Proposed Standard" (there are some legacy "Draft Standard" documents too).
RFC stands for "Request For Comments". Some of them get turned into standards, but most are just the IETF equivalent of a forum thread. They're a way to start a discussion about a network engineering design.
https://tools.ietf.org/html/rfc6108#section-3.1