Sometimes digital public spheres are built with the intention of democratizing what information gets prioritized by creating upvote/downvote systems that drive what information appears on the front page of a website. Such systems are generally viewed as fostering equality (everyone gets to vote on each piece of information if they want to) and openness or transparency (vote totals are usually displayed next to the items). It’s tempting to think of these platforms as a way to constantly poll the public about what is most important, and in turn to assume that the content on the front page is an accurate representation of what the entire community of users thinks is most important.

What might not often considered by users of these sites is the impact bad actors can have on these digital public spheres. From 2006-2010 Digg.com was a hugely popular social news site where users could vote links to news stories and other content up or down. It had over 30 million monthly visitors at its peak, before management and software problems led to its downfall. Most readers probably assumed that Digg’s voting system meant that the front page was an accurate representation of what millions of people thought were the most important links to follow for the day. Unfortunately that was not always the case. CMS alum Chris Peterson wrote his master’s thesis on the Digg Patriots, “a group of Digg users who coordinated to make the social news site more politically conservative than it would have been without their intervention,” which he describes as user-generated censorship.

Peterson explains their tactics: “By coordinating their activities they were able to quickly vote down left-leaning stories soon after they were posted, which caused the Digg algorithm to determine that the story was not worthy of the front page, even if it was voted up afterward They also ‘deduced that the Digg algorithm treated comment activity as an indicator of interest, pushing more active posts higher and sinking less active posts lower, so they developed a strong norm of not commenting on liberal posts while creating purposefully outrageous comments on conservative posts to bait liberal users into a frenzied discussion.’”

One of the problems this reveals is the possibly naive trust we place in the fairness of digital mediators. We know that the public sphere is a place where people vie for influence over one another. We assume that people use upvotes and downvotes to represent their own set of interests and to engage in influencing content. But we may also assume that because an “unbiased” computer program tallies the votes, the result is accurate, fair, and transparent. We don’t often consider whether a group or individual might be able to find ways to influence the digital mediators. Because it all happens in a black box, the typical end user only sees the content the users and algorithms push to the top. Auditing the algorithm or otherwise checking to see if it has been compromised by bad actors is basically impossible.

This is not to argue that such systems are bad, or even that they are inferior to other methods of establishing equality in a digital public sphere. But it does raise questions about how we negotiate trust with systems that are opaque to us, and what it means to create an expectation of equal representativeness in a digital public sphere.

I am left with several questions about these systems: how can we know how open, equal and representative any algorithmic system is (and any digital public sphere using algorithms to drive content ranking), and whether groups are succeeding in over-representing their interests? Does exposing an algorithm to public scrutiny simply make it easier to compromise, or is there a level of open source refinement that could actually make it less susceptible to the kinds of manipulations the Digg Patriots used for ideological ends? What does a set of best practices look like for people that have a genuine desire to create digital public spheres that are open, equal and representative?