what would be a better subset then? if you take the set of all computer-indexed written text, you would end up simply measuring the rise of the capacity of the unwashed masses to air their thoughts in permanent, public, usually informal form.