How Does Interpersonal Surveillance Affect Human Reviewing Behavior in the Presence of Algorithm-Generated Ratings? Evidence from Initial Coin Offerings (ICOs)

Abstract

While algorithm-generated information is increasingly prevalent, how it is used as a basis for human evaluation is not well-explored. Using the context of online professional ratings of initial coin offerings (ICOs) projects, this study examines how increased interpersonal surveillance and different experiences impact human expert reviewers' ratings relative to algorithm-generated ratings (AGRs). Leveraging an interface design change on an ICO rating platform, we find that increased interpersonal surveillance leads experts with advisor experiences to lower their ratings and become less likely to rate above AGRs, compared to experts without advisor experiences, though their ratings do not necessarily converge to AGR levels. While these effects hold for experts with high reputation concerns, those with low reputation concerns tend to converge to the AGR levels in their ratings, despite not significantly changing their general rating levels. These suggest that human experts with advisor experiences may strategically assign high ratings and overrate relative to AGRs to impress project teams. The heightened reputation concerns stemming from the increased interpersonal surveillance can prompt strategic adjustments in reviewing behavior and potentially curb overrating. Overall, increased interpersonal surveillance drives human reviewers to use AGRs as a benchmark, possibly correcting humans' potential rating biases.

Keywords: Algorithm-generated ratings, Online reviews, Interpersonal surveillance, Initial coin, Rating Bias