I would like to thank authors for their study.
However, reducing complex systems such as communities to simple metric is fraught with danger of missing the point. The authors used the Wikipedia user pages to determine political affiliation of the users and then assessed their interactions with each other. They came to some conclusions, which ring true to me both as a wikipedian and a scientist:
*People who declare allegiance to a political party are more active on Wikipedia than an average editor;
*People interact more on their personal pages with people with similar political convictions.

The article main conclusion, that there is no political preference in interaction where it matter most, on the articles talk pages, contradict my experience and I spotted a problem with author’s methodology. Firstly, it’s not the quantity, but quality of interaction, which matters. Yes, there may be the same amount of posts directed to the members of the same and adversary party. But in a confrontation you talk more to your opponent than to your supporter. I think further analysis of whether the posts directed to the opponent were positive, neutral or negative would shed more light on the matter.

More importantly, the authors did not analyse partisan interactions were it really matters - in the articles themselves. All changes made to the article are recorded on the “article history page”. An edit is often cancelled - “reverted” and wikipedians are notorious for the “edit wars”, i.e. multiple reverts of each other’s edits. A comparison of history page and talk page would most probably turn up discussions about such wars.

Interestingly, the “talk page” of the article itself is almost as interesting as the article itself. While digital anthropology allows you to study your tribes at home, the tribe people will follow you home. The talk page is full of another Wikipedia community phenomenon - wikipedians, who fall out with the main community, or “anti-wikipedians”.
One of them, who admit paid editing - an anathema to the Wikipedia volunteers -, says that he created a lot of fake Wikipedia users, so-called sock puppets with different political affiliations, therefore authors classification of political affiliation of the wikipedians using their homepages is invalid. However, the sample the authors used - 1400 - allows a significant margin of error and if some sock puppets were counted as genuine users, this doesn’t matter.

As doesn’t matter that anti-wikipedian claims "Wikipedia - a totalitarian state". While the authors’ conclusion that the “wikipedian” identity always takes precedence over their political affiliation may be a bit too optimistic, they are right in principle - Wikipedia is edited by different people of different social, political and gender identities - and therein lies its strength.