One fingerprinting element is the browser's user-agent string. If we consider one browser, Firefox, we can see that it has different strings across different OSes, distros, forks (Tor, IceCat) and versions. Some of these populations would be very small (say, Guix SD users running the latest version of IceCat) and therefore individuals within those populations would be more easily identified (in combination with other data).
Has there been any attempt to coordinate user-agent strings among these communities? (Or resistance to such an attempt.)