If you want, in essentially every programming context where `-x` is meaningful, you can write `0-x` instead. It’s arguably less confusing.
The point he’s trying to emphasize is that in his proposed number system, every exactly represented number has an exact multiplicative inverse, which can be computed by flipping bits, analogous to computing the additive inverse with a floating point number or signed integer.
And you don't need to make a point that every number has a multiplicative inverse. That's already established by simply pointing out that these are the projective reals and that 1/0 = ±∞. All non-zero reals already had multiplicative inverses just as they had additive inverses.
Edit: he clearly wanted to emphasize that this is a unary operator. Using 1/x as the symbol instead of /x or ÷x would weaken the desired association.
For instance, when I read an equation like this:
"Information = 1/uncertainty" is just plain clearer.