The discussion revolves around improving Lighthouse performance measurements for Discourse by excluding the “Chrome-Lighthouse” user-agent to reflect real user experience. rrit proposes enabling Lighthouse to measure the application view as seen by actual users, contrasting it with the current crawler-view measurement. codinghorror expresses concern over Android performance disparities and insists that Lighthouse tests should be run on iPhone devices to be meaningful. rrit counters that Google already collects real-user experience data from Android devices via the CrUX dataset, which feeds into Core Web Vitals. Falco suggests using a custom user agent to simulate the desired view, while rrit clarifies that the mobile/desktop view can be toggled via URL parameters (?mobile_view=1 or ?mobile_view=0), with ?mobile_view=auto resetting the setting.
When they run those lighthose performance tests on a recent iPhone model (even an iPhone 8 or later), I am open to this. Right now android performance is 3x (or more) slower than iPhone unfortunately.
Check your device at Speedometer 2.0 to see what I mean – this measures JavaScript performance in the browser:
Until they run lighthouse on an iPhone (any model) for context, I am absolutely opposed to any change here. When they do, I will happily accept the PR, with gusto.