Demonstrating Decisions: Waymo’s World

This chapter details the work undertaken by Waymo—Google/Alphabet’s autonomous vehicle division—to demonstrate the safety of their autonomous vehicles. Through analysis of two documents, the chapter considers how Waymo has publicly reported so-called contact events—the company’s euphemism for incidents and crashes involving Waymo vehicles. Beginning in 2020, Waymo published a report detailing 47 contact events involving Waymo vehicles undergoing testing on public roads. Twenty-nine were considered ‘simulated events’, incidents that would have happened if a human operator had not assumed control. Generated instead within in-house simulation software, Waymo’s revealing of such work constituted an important milestone in the post-Uber ATG crash autonomous world, suggesting that the incident had forced a change in Waymo’s public documentation of the crash worthiness of their own vehicles. A subsequent report in 2023 documented a ‘conflict typology’ comprising of 16 such types of possible interactions between the Waymo vehicle and other road users. The chapter argues that Waymo, through such classification work, is effectively able to apportion blame away from their own vehicle. As examples of ‘deferred’ decision-making—shifting the meaning of such incidents into the future and away from Waymo—algorithmic ‘doubt’ is used as an eminently useful, reputational resource.