I've heard of duck-typing, and static-typing, but this is the first time I've heard of Schrödinger-typing...

Yeah, Appendix A Crockford's "Good Parts" (could be Appendix B, I can't remember. The book's so small I can't find it.

  > Boolean([])
  true
  > []==false
  true
  > [] == ![]
  true

How the hell does that work? Aren't all non-booleans implicitly cast?

edit: rephrased

It's because the == operator converts its arguments to be of the same type, and it does so in a rather unintuitive way. This makes it both inefficient and frequently baffling. You pretty much always want === instead, which does not allow comparisons between different types.