>>36
"as you should know, there is no such thing as a (non-compactly supported) measure on the set of non-negative integers"
First off, I believe you forgot to include the adjective translation-invariant. There certainly is a non-compactly supported measure on the set of non-negative integers. Let mu({k}) = 2^(-k-1). Then mu({0,1,2,...}) = 1, making ({0,1,2,...},P({0,1,2,...}),mu) not only a measure space with a non-compactly supported measure, but also a probability space.
Regardless, your point that there isn't any uniform random distribution on the non-negative integers is well taken. I shouldn't have cut corners in that example, but it can easily be rectified: Let k be uniformly distributed in the integers modulo 2.