I've been looking into the causes of poor scroll performance on Chrome for Android, and on a few sites parse.ly's time-engaged script has come up as being the thing that prevents scrolling from starting quickly due to the touchstart listener you add on the window (and other code on the page running slowly - eg. during page load).

I'd love to work with you to find some solution to this issue. It looks like you just want to keep track of the timestamp when the user interacts with the page, right?

One simple solution might be to switch from using touchstart time to using touchend time instead (since touchend events don't necessarily block scrolling, although we'd need to fix some things in chromium to actually optimize this case). But I can understand you might not want to change the definition for your engagement metric.

Another option is to have a way of requesting a non-scroll-blocking touchstart event. I'm working on such an API in passive event listeners. It looks like it would work for you, what do you think?