I posted similar comments before seeing yours.

Yes M-B is probably pseudo-science (and of course this poll is unscientific). However, I suspect that some moderate (better-than-chance) degree of validity has kept people using it all these years.

The MMPI supposedly was validated by assessing personality types independently by other methods and then empirically finding corresponding patterns of answers. I don't think that's viable for the M-B, however, as it's shorter, has been more exposed to the public and probably easy to game.