Description

As the problems with algorithms have become common and apparent, legal scholars have begun to consider how to govern algorithms. Among a variety of legal solutions, algorithmic transparency is, practically and theoretically, a well-known regulatory principle, which has attracted significant attention in recent years. However, as ex ante regulation, algorithmic transparency has inherent limitations. Although algorithmic transparency can alleviate some “black box” problems, this article demonstrates that algorithmic transparency is often neither feasible nor necessary.

This article further argues that the role of algorithmic transparency in algorithm governance should be a subsidiary one and should not be applied in every case. Based on a critical analysis of algorithmic transparency, this article attempts to reconstruct the algorithm governance theory and further argues that compare to essentialism-driven ex ante regulation, pragmatism-driven ex post regulation is a more appropriate regulatory strategy.