Data generated by people isn't random, it has statistical regularities that can be used to narrow down the search space and since quantum computers are better at certain kinds of search problems it will be easier to find fragments of files based on just the hashes of the content.