Russell's Principia Mathematica (PM) is indeed better

engineered than its predecessor by Frege because it has orders

on propositions. Because of orders on proportions, PM does

not allow the [Gödel 1931] proposition I'mUnprovable.

Furthermore, adding the proposition I'mUnprovable would

make PM inconsistent.

    The Gödel number of a proposition in PM is itself
    "incomplete" because it *doesn't* include the order of the
    proposition.  Allowing its Gödel number to represent a
    proposition is indeed a kind of "code injection" attack,
    which if allowed would make PM inconsistent.