What “civic signals” should be incorporated in algorithmic or technology design? Why? Make the case for a specific approach.
Originally written for Harvard Kennedy School’s Digital Platforms, Journalism & Information course, March 2020.
Foer’s “World Without Mind” frames the advent of technology monopolies as a successful project to mold humanity and human nature through the “reordering of production and consumption of knowledge” then imposed upon us. The industry’s once idealistic aspiration to create and experience a “collective social existence” is fraught with moral conundrums: in automating what users see, is Big Tech automating free will? In sorting news, are entire schools of thought and viewpoints getting left out? Can platforms encourage users to treat each other with respect online, and how?
The gatekeepers’ legacy of widespread social control is upheld through information ranking and the monetizing of targeted advertising at the mercy of its algorithmic core. In attempting to universalize products and services to relieve computing of the burden of human decision making, Big Tech went from offering users choice to paternalistic nudging towards its preferred choice. In Foer’s words, “technocracy is the new power order,” and it is a prospect we don’t fully understand nor are sure how to actionably contend with.
In a bid to reshape this fractured social existence mired by political divisiveness and fake news, rethinking what Eli Pariser calls our “digital civic fabric” with the help of positive “civic signals” offers an urgent first step. While expert third parties and civil society can leverage signal models to help better guide tech’s use and make up of their algorithms, this essay will also poke holes in Pariser’s framework. It will also examine other elements of a healthy online environment to consider as we embark on navigating the moral, social and cultural questions we face in demanding our platforms better serve us in a democratic context.
The fall of traditional media and the struggle of publishers to maintain their standing business model has been part and parcel of the rise of tech monopolies. Acting as the new gatekeepers, but missing the knowledge workers that defined old media and mediated how information was transferred to the public, tech has eroded the “boundaries between fact and falsehood.” This has meant seemingly giving up on journalistic values and integrity, making way for sensationalist headlines, subtly masked propaganda and the like. The hive mind mentality that has arisen, across both sides of the aisle, has helped spur political division, alienation and distrust, alongside false narratives and fabricated news stories. This is where Pariser’s category of “civic competence” comes in.
Civic competence emcompasses providing online communities the news literacy and local community information they need to gain awareness of civic topics and events. Platforms can empower individual users with tools and informedness about relevant and localized resources to enable stronger engagement within communities. One strategy to promote this goal would be for platforms to build in code to uprank this type of reliable information to help citizens become more actively and effectively engaged.
While this could arguably serve to fill the hole left behind by local news getting slashed and outpriced by media behemoths and social networks, a few challenges arise. First, it may be difficult to coordinate who draws the boundaries around this type of content. Who defines it should be deemed as important as the message disseminated, and both code and humans still struggle to root out bias. Secondly, picking one or two categories within the civic signal will allow for a narrower scope, and a successful observation of its impact over time. My suggestion to deploy and measure this algorithmic feature would be to set up a few primary “phases,” starting with a small sample of users in a small area, to then ramp up and tweak the algorithm’s design on a tri-monthly basis to widen its breadth.
Questions Pariser asks and which I will attempt to answer lie in defining “needed contextual information” and how broadly to define engaged citizenship. One method to outline contextual information would be to survey residents of a small town or area, or even choosing a few political organizations that have big online followings and activism to determine what information they feel is lacking or where they’d like further elaboration during online searches and social networking activities.
In order to better engage citizens, partnering with local library branches and city council representatives to determine where such activity could be bolstered, leveraging their offices’ existing feedback loops, and building bridges between that research and online practices could inform algorithmic design with constituent service in mind. These efforts should be quantifiable as possible in order to set benchmarks, make predictions and put forth future solutions to demonstrate positive or negative fluxes in response to such signals over time. There must also be a mechanism to ensure users see information from trusted local and national news outlets and organizations. A formal, clearly labelled, and free verification process to guarantee the top three local outlets are flagged as legitimate and prioritized by the algorithm should be put forth. If this program needs to be monetized, a federal tax on Big Tech could garner funding.
Two other civic signals that could be combined into one initiative are the closely aligned but distinct “outgroup humanization” and “connections between groups.” A major issue today is a lack of bipartisanship in sharing ideas online, and political polarization and misinformation about outgroups fueling fires in real life. A concerted effort to share news and op-eds from various groups in a bi-directional manner would build channels that are shared, rather than separate. Inclusiveness and belonging on social media is too heavily emphasized to the detriment of the users themselves, who end up sorting themselves into ingroups and interactions that perpetuate divisive groupthink. Code-driven exposure to opposing views across diverse communities can begin to foster trust, understanding and empathy — which can be transformed into face-to-face engagement to find common ground to unify divergent voices in the future.
The struggle with these signals lies in identifying what constitutes a “group,” namely whether one is understood as pertaining to race, class, gender, sexual orientation, socioeconomic status, religion, political affiliation or others. As such, choosing one or two subgroups to focus on in a pilot program will be the most strategic way to understand what sticks and what doesn’t in establishing new norms, what roadblocks need workarounds, and what best practices should be applied to other categories down the line.
Pariser’s lack of case studies crucially holds us back from analysing what demographics should be prioritized in building a healthy online society. As we undertake to put civic signals into design thinking, we should keep the following questions top of mind: Are certain groups more likely to need civic signals enforced than others for their safety and wellbeing? What kind of process could answer this nuanced and difficult existential examination?
Much like TikTok’s recent announcement that it has opened a “transparency center” around its operational and content moderation practices, we could create similar accountability systems for major social networks as they test out implementing and integrating civic signals. Inspired by tech’s foundational open source ethos, the function of such a center or forum could also enable the crowdsourcing of ideas to tweak methodology and benchmarking against realistic goals, and bring about suggestions if a signal fails to gain traction. In order to correct for code’s upending of human trial and error, our best path forward should be a turn to hypotheses, intuition, emotion and active collaboration amongst technologists, community leaders and everyday users to bring forth a true online “global village.”
