It's like saying
If 5+3 = 8, then 11+1=72
No, it's not like saying this, because this implication is simply false. The difficulty with the original "implication" is that it is self-referential in a way which cannot be represented by the classical logics. If you try
A = A -> B
then you're saying
(A -> (A -> B)) and ((A -> B) -> A)
This is not a tautology, since asserting A and ~B falsifies it. In fact, its verity is equivalent to that of "A and B".
Proof: substituting (~x or y) for (x -> y), we get
(~A or (~A or B)) and (~(~A or B) or A)
= (~A or B) and ((A and ~B) or A) by DeMorgan
= (~A or B) and ((A or A) and (~B or A)) by distribution
= (~A or B) and (A and (~B or A))
= (~A or B) and A by A -> x or X
= (~A and A) or (B and A) by distribution
= B and A cuz ~A and A is a false.
If you try nth-order logic, you're totally screwed; you simply can't represent the statement anymore.
-Erasmus