So, I recommend an errata to RFC 2119: "These words MUST NOT appear in a
document in lower case."
I'm glad you said "I recommend" instead of "I have recommended", as the
latter would violate the recommended (oh dear) rule.
This RECOMMENDED rule would also imply that documents can no longer be
published during the month of May, as otherwise the date line would put the
document out of compliance with this erratum.
Also, it would no longer be possible for the references to list any document
that was published in the month of May. I suppose we could make it a rule
that the month of May be referred to as "the month between April and June".
Of course, you didn't say that the reserved words muSt nOt appear in mixed
case, so mAybe that will become a workaround.
Seems to me that precision of meaning overrides graceful use of the
language.
Many IETF specs lack the desirable degree of precision, but the use of the
"reserved words" in capitalized or uncapitalized form just is not that big a
contributor to the imprecision. Failure to properly specify all the actions
in all the possible state/event combinations is a much bigger source of
confusion than failure to capitalize "must" (I mean failure to capitalize
"MUST"). I don't know why so much attention is being given to something
that is NOT going to improve the quality of the specs by any appreciable
amount.