Jennifer Forestal wrote an interesting article in the most recent issue of the American Political Science Review titled, "The Architecture of Political Spaces: Trolls, Digital Media, and Deweyan Democracy".  In it, she highlights the importance of how website architectures can either invite or prevent different types of political activities by creating different types of spaces.

So if the challenge is to create online spaces that encourage democratic discourse, then the problem is trolling. As an example, Forestal describes how, in August 2014, Jezebel - a site dedicated to "what contemporary women want to talk about" - had its comment section overwhelmed with images of violent pornography posted by anonymous users.

In Forestal's assessment, this was made possible by the open architectural design (through code) of the site's commenting platform, Kinja, which allowed anyone to create an account and comment anonymously as often as they wished.  Also, by granting each commenter the power to be moderator of any ensuing discussions, each discussion thread was effectively treated as a unique and exclusive event run by the thread's founder, rather than conceiving of all of the threads on the site together as part of a single collective, or community, enterprise.  Thus, trolls were able to proliferate.

Ultimately, Kinja was changed so that comments were divided into "pending" vs. "approved" categories - expanding its boundaries of approved commenters to include users of the website with "a long history of demonstrated good work".

Here's the thing... this redesign of Kinja's architecture kept the conversation "open", however it added a layer of editorial control.  Some might say it created an elite class of gatekeepers - done intentionally and by design.  Think of it as having created a Superdelegate system - a mechanism for the party establishment to check the impulses of the rank-and-file.

Without placing any value-judgments, the big idea here is something that myself and others have written about extensively - that software architectures absolutely structure people's behavior in cyberspace.

Forestal recommends that, in order to avoid trolls, online democratic communities need to be designed as "small" spaces, with stable members, and yet still be "flexible" enough to be responsive to change so as to prevent gated communities that discourage new experiences and encounters.  But this is not the most revealing conclusion; in fact, it's been the holy grail that designers have been pursuing for years but no one has quite figured out how to pull off.  The main contribution of Forestal's piece, then, is to frame this familiar challenge in a better theoretical context.