A forthcoming paper in Proceedings of the Royal Society A entitled "Satellite conjunction analysis and the false confidence theorem" explores the practical consequences of treating a non-random event as though it were random, a practice that is currently common among statisticians, engineers, and applied scientists. When analyzing the uncertain paths of two objects predicted to pass near each other in orbit, this practice leads to a severe and persistent underestimate of collision risks. This is especially serious if it masks the chance of a cascade of collisions, each generating more space debris that increases the likelihood of further collisions, which could potentially render space unnavigable for generations. Further, this statistical problem is a general issue, not unique to satellite collision risk analysis. Fortunately, alternative approaches to inference are well-established, and we show how to identify methods that are free from false confidence.

Distribution of debris in orbit around the Earth. Image: NASA Orbital Debris Program Office

Further reading: