Sorry Joseph, I concur that with Ned that we're best off leaving this alone (risk of breaking doctests with the changing repr, risk of fooling an overly specific type test, etc).  The code is not incorrect and does correspond with its docstring which claims 1 for true and 2 for false.

The sentiment is appreciated, but in general, the time to get APIs perfect is when it is initially released.  Afterwards, we really only want to make implementation improvements or API extensions until the API is flat-out broken or has major usability problems (even then, it always causes consternation to change the API).