Running a report on https://www.digitalocean.com via https://pagespeed.web.dev, I get back a result for desktop with a multiple-second-long Total Blocking Time and a recommendation to Minimize main-thread work, attributed to around 30s of time taken with the majority of that being classified as “Other”.

Compared with running Lighthouse myself against the site though, I get back a 0ms Total Blocking Time and while the Minimize main-thread work recommendation is listed (in grey, not even a warning), it is attributed to only a couple of seconds of time taken.

Why are the Total Blocking Time, and the Minimize main-thread work time, so drastically different between PageSpeed Insights and local Lighthouse? This doesn’t feel explainable purely by device throttling?