Beyond relevance is significance.

Twitter produces data which is almost impossible to interpret from the content after the fact, simply because there is too much of it, and its "granuality" is too small. It is, however, possible (indeed unavoidable) to infer significance from this data: e.g. waxing and waning group concerns (which can then in turn be analysed using # words, or by types of sources, etc.) It is the stats themselves which are the object of inquiry, as such, they are "self signifying". Types of content can be treated as indicators which can be counted, the results of which are simply another (meta?) level of information which consciousness can use in its sense making quest.

See also - Against Interpretation 2.0