Robustness and the Event Horizon Telescope: the case of the first image of M87*

Abstract

We examine the justification for taking the Event Horizon Telescope's famous 2019 image to be a reliable representation of the region surrounding a black hole. We argue that it takes the form of a robustness argument, with the resulting image being robust across variation in a range of data-analysis pipelines. We clarify the sense of "robustness" operating here and show how it can account for the reliability of astrophysical inferences, even in cases -- like the EHT -- where these inferences are based on experiments that are (for all practical purposes) unique. This has consequences far beyond the 2019 image.