Field experiments can guide wise use of platform power if we re-design the relationship between democracy and behavioral science.

Designers and internet researchers have become powerful policymakers governing human affairs. As social platforms and intelligent agents become routine in the daily life of billions of people, the public has come to expect these systems to address deep-seated social ills.

Tech companies are currently expected to manage social problems including terrorism, discrimination, suicide, self-harm, eating disorders, hate speech, child pornography, misogyny, copyright violation, and political polarization, to name a few. Advocacy organizations have even opened lobbying wings in San Francisco, hoping to influence company policies.

Over the years, we’ve argued that we have an obligation to test the risks and benefits of social interventions online, but there’s a catch: behavioral experiments tend to be designed for top-down control rather than a democratic society.

This paper by J. Nathan Matias and Merry Mou that reports on the last two years of CAT Lab’s work to redesign large-scale experiment software for democracy.