Context Graph: It’s time to bring context back to the web
Nick Nguyen
1.2K49

No. Stop. Don’t fiddle with the UI without addressing the core user-facing issue with Firefox, and that’s the real or perceived slowness compared to Chrome and other Blink-based browsers.

Look at http://david.li/paint/. On Linux, it’s borderline unusable, but it’s very snappy and fluid on Chromium. And that’s on a reasonably powered laptop with a discrete 3D card.

Look at https://jsbench.github.io/#50a1aa92666d9703aabf56ed2436c64d. Chrome is beating Firefox by significant margins when it comes to DOM manipulation, and I suspect that’s due to it deferring commits to the underlying tree for as long as possible. I suspect the process-per-tab model also helps.

Those are _technical_ issues that should be addressed appropriately, but there is also the area of _perceived_ performance. Things like asynchronous panning go a long way toward bridging that gap. I am not sure what can be done to improve this — perhaps, on new navigation, starting to render a cached version of the page while the request is in flight in anticipation of a “not modified” response?

As part of wider UX, how about getting serious with adding pages to the user’s home screen on Android? With Chrome, developers can specify a manifest that tells the browser to not bother with rendering toolbars or anything else that because the app will take care of it. Firefox should try doing this too — logically, the content can also be run in its own process, separate from the main browsing window.

Implementing new Web API features is secondary. The Payments API can let you add a quick and easy way to donate to Firefox — the user is going to pay anyway, might as well add a checkbox ‘Support the Open Web with a small donation to Mozilla’ which adds on a donation to each payment processed with the new browser-based UI. I’d keep that box checked.

I love your idea. It would make a great add-on to Firefox. I’d use it. But it will be totally useless in a browser which is slower and less widely supported because it focuses on blue-sky experiments rather than what users actually want in a browser — speed and features that help them go about their daily routine, like buying clothes or ordering a taxi. And then, maybe then, when you sort e10s and Quantum, the blue-sky experiments in content discovery and context finding will go from frustrating users (“They’re spending money on _this_ but Facebook still takes forever to load!”) to delighting them (“Hey, this is cool, thanks for recommending this article to me!”).