Commercializing Contrarian Ideas: Evidence from AI Contests
Job Market Paper
Entrepreneurs with contrarian approaches – even when they are right – often struggle to attract resources. At the same time, skepticism can dissuade potential competitors from pursuing the opportunity: the potential for contrarians to capture value after commercialization depends on whether they can persuade resource-holders without inducing a competitive response. To investigate this, I leverage hundreds of Artificial Intelligence contests as “moments of clarity” that can shift beliefs around a contrarian approach. Through a difference-in-discontinuities design, I show that when contrarian contestants win, they become disproportionately more likely to found a startup (a 7x higher effect than for winners using popular, state-of-the-art methods). The startups they found attract more venture capital at higher valuations than those founded by mainstream winners. And yet, the papers describing the underlying contrarian contribution receive fewer citations than those of mainstream winners for up to 10 years after the contest, especially when the race was close. Overall, the results support the notion that contrarians face significant commercialization barriers prior to proving themselves but can capture extraordinary value thereafter.