In my own search for knowledge around NaN, if found that the Go language specification doesn't give us more than "as defined by the IEEE-754 standard". Unfortunately, the standard itself is behind a paywall.
However, this GNU libc documentation helped me a lot in understanding the principle on comparing NaN:
In comparison operations, positive infinity is larger than all values except itself and NaN, and negative infinity is smaller than all values except itself and NaN. NaN is unordered: it is not equal to, greater than, or less than anything, including itself. x == x is false if the value of x is NaN. You can use this to test whether a value is NaN or not, but the recommended way to test for NaN is with the isnan function (see Floating Point Classes). In addition, <, >, <=, and >= will raise an exception when applied to NaNs.
As such we can see math.IsNaN's implementation:
func IsNaN(f float64) (is bool) {
// IEEE 754 says that only NaNs satisfy f != f.
// ...
return f != f
}
I'm adding this answer, as the NaN got me in trouble in a sort.Interface implementation and I hope it might help others.