It shouldn't, the algorithm as described works for arbitrary data. The whole point of the hashing is to transform the stream of arbitrary data to a stream of un

It shouldn't, the algorithm as described works for arbitrary data. The whole point of the hashing is to transform the stream of arbitrary data to a stream of uniformly distributed random integers, so that you can then apply the reasoning from that paragraph.