Paul M. Romer

Digital Authenticity and the Future of Truth

Tuesday, 26 August 2025
18:00 - 18:30 CEST

Details

tba


Abstract

These days, research results are communicated almost exclusively via digital files. Any such file – be it an email, a document, an image, a website, a text message, a video, an audio recording –is code. When you interact with it, you run code that someone else has written on your computer. This puts you at risk. One way to protect yourself is to run code only if it is endorsed by someone you trust. To make the best use of the content of any message, you also want to know whether someone you trust is willing to put his or her reputation on the line by endorsing the content of the message. For both reasons, it is important to be able to authenticate assertions about the messages you receive. Unfortunately, with digital media it is much easier to produce plausible, fraudulent messages.

As a scientist, you would like to have a way to attest that you are the author, or one of the authors, of messages you send. You want it to be easy for others to be able to verify your attestations. You also want to be able to verify any attestations that others make. Digital signatures made possible by asymmetric cryptography make it possible to achieve this goal of verifiable digital authenticity.

In this lecture, I will describe with some precision the assurance that a digital signature can provide; the practical hurdles that have, until now, prevented members of the community of science from adopting digital signatures; and the risk that in the absence of easy authentication, the equilibrium that supports scientific truth will collapse. To the extent that time allows, I will also describe the game theoretic model of the equilibrium we call science on which this analysis is premised. For an informal presentation of this model, please consult "The Knowledge Machine" by Michael Strevens. The key insight is that a system of science that converges to the truth is an equilibrium outcome that might, or might not, exist.

Related Laureates