Essays on Bayesian Entrepreneurship: Evaluating and Commercializing Unconventional Ideas

Abstract

This dissertation investigates a fundamental challenge complicating the evaluation and commercialization of entrepreneurial opportunities: some ideas are valuable precisely because not everyone recognizes their worth. The first essay analyzes barriers against the commercialization of contrarian ideas. Researchers working with unpopular AI algorithms tend to commercialize their work only after a successful public evaluation. Those who clear this hurdle subsequently achieve better entrepreneurial outcomes. A regression-discontinuity analysis shows that this partly reflects status quo bias: for unpopular methods only, winning a contest serves as a certification, channeling disproportionate resources to the winner while equally strong near-misses remain sidelined. The second essay finds that greater judge disagreement in venture competitions predicts higher future success, especially for more distinctive startups. The third essay shows that skewness in idea value exacerbates asymmetric information in markets for ideas. Using data from auctions for digital businesses, I illustrate how this can explain why online marketplaces for ideas have struggled to emerge despite lowering transaction costs: informational frictions severely depress bids and prevent high-value digital startups from trading. The final essay, coauthored with Alfonso Gambardella and Scott Stern, introduces the archetype of Homo Entrepreneuricus: an entrepreneur who deliberately tests subjective beliefs through structured experimentation to navigate uncertainty.