All this happened, more or less.

The first example is a network of books on politics, compiled by V. Krebs (unpublished, but see www.orgnet.com). In this network the vertices represent 105 recent books on American politics bought from the on-line bookseller Amazon.com, and edges join pairs of books that are frequently purchased by the same buyer. Books were divided (by the present author) according to their stated or apparent political alignment—liberal or conservative—except for a small number of books that were explicitly bipartisan or centrist, or had no clear affiliation.

For our second example, we consider a network of political commentary web sites, also called “weblogs” or “blogs,” compiled from on-line directories by Adamic and Glance [32], who also assigned a political alignment, conservative or liberal, to each blog based on content. ... the network divides cleanly into conservative and liberal communities and, remarkably, the optimal modularity found is for a division into just two communities. One community has 638 vertices of which 620 (97%) represent conservative blogs. The other has 587 vertices of which 548 (93%) represent liberal blogs. M. E. J. Newman. 2006. Modularity and community structure in networks.

As a result, some sites are turning to a new strategy to keep their services "sticky" and their users satisfied: They're not just encouraging them to network online, but to attend offline events and parties in the real world as well. Yelp now regularly hosts parties in big cities around the country.