Surely we need "negative NaN" as much as "negative zero".

Sssshhh ... you'll give 'em ideas.
The standard that I find really absurd is NaN ** 0 being defined as 1. The logic seems to be that since *all* numbers raised to the power 0 are 1, then NaN ** 0 should also be 1. (Those who create these standards are apparently unaware of the fact that a NaN is not a number.)

Cheers,
Rob