I think there's opportunity in packaging this as a system installed & independently updated dynamically linked library much like WebView is updated on Android KitKat onwards. Eg. a single "system" Chromium component that's installed centrally & used by multiple programs much like Microsoft Runtime DLLs on Windows.
That's the issue.
With this, my / will reach terabytes.
There's two things why not: cheap isn't free, and "elegancy principle": Don't build shitty hacks, build software that's easy to maintain, as minimal as possible, and will be used for decades.
The same reason security libraries are distributed independently so that they can be automatically updated without the program using them being re-compiled & re-released. On the Android platform, Chromium WebView updates far more frequently than the apps that depend on it & includes security patches on update.
RAM in not as plentiful, nor are level-3, level-2 and level-1 cache. having a second copy or almost-copy of a library can increase utilization of any of them.
The performance improvement has basically come from the massive market that web is and consequently Google, FF and MS optimizing their JS and rendering engines (plus Moore's law)
I find that this phrase irks me quite a lot these days. Feels like it's at every corner of software development waiting to pounce and ensure thoroughly mediocre results.
There's something to be said about, "don't let the perfect be the enemy of the good" but at the same time, "good enough", especially when it comes to UI speed/responsiveness and UX quality, feels like a massive cop out. Isn't there a middle ground somewhere that doesn't call for such huge concessions?
We've already solved fragmentation in the native UI space, with GUI toolkits like GTK and Qt. If someone extended those to mobile devices, then we'd have something a lot less terrible than HTML5/CSS/JS, which performed a lot better.