I was looking through some data and have found an issue with the values of DOMHighResTimeStamp - either from performance.now() or event.timeStamp. Specifically, I take the difference between two times as a reaction time for a specific task. For all 200+ people, with a range of browsers and systems, recorded times are in a reasonable range of around 4000. However for 4 individuals, recorded reaction times for all trials were integers 12+ digits long (e.g. 24158598587397500 for one of them) - obviously impossible.

3/4 of these people had the following user agent:

• Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_6) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/14.1.2 Safari/605.1.15

i.e. Safari 14.1.2, OSX 10.15.6. There are others with the this OS or same operating system whose results are fine, but none with this specific combination. Is anyone aware of what happened here, or how to mitigate this in future? Normal polyfills won't work as the function appears to be returning something and so is found by the browser.

Also, the fourth person was using:

• Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_6) AppleWebKit/605.1.15 (KHTML, like Gecko)

i.e. Apple Mail 605.1.15, exact same OS