Conversation

As a little side project this week I’ve put together a simple playground for learning about input pointer latency in web browsers and experimenting with the idea of using path prediction, as demonstrated in this video. Play here: rsms.me/pointer-latenc

0:24

13.4K views

423

Rasmus Andersson

@rsms

Apr 24, 2020

Responsiveness and interface latency is something I’ve been passionate about for decades. The fact that you’re always working on old input data in a web browser—in any windowed app, really—is unfortunate when you want to create something that feels snappy and responsive.

Rasmus Andersson

@rsms

Apr 24, 2020

On pointer events in web browsers, I’ve put together this poster that illustrates what happens in Chromium on macOS: figma.com/c/file/8239558

Ryhan

@ryhanhassan

Apr 24, 2020

Neat! Seeing the prediction visualization reminded me of this drawing tools with predicted paths

Quote Tweet

Andy Matuschak

@andy_matuschak

· Jan 21, 2018

(it's very interesting how the "feel" of this pen with realtime predictions is totally different from this pen with 200ms predictions, which is totally different from this pen with 2500ms predictions… it's so fertile making discrete systems continuous!) twitter.com/andy_matuschak

Show this thread

Replying to

and

Ah, thanks for the shout. You can play here: andymatuschak.org/scrying-pen/ Fun memory: in ~2013, when I was at Apple, I made all scroll views predict future touches to get a tighter response. We pulled it before shipping—not a big enough diff. There's an API for this now for drawing.

6:04 PM · Apr 24, 2020Twitter Web App

Replying to

and

iOS does touch prediction I know. Did the scroll code—which sounds very interesting—get ditched because of that perhaps?

Andy Matuschak

@andy_matuschak

Apr 24, 2020

No, the system-level temporal touch prediction is something you have to op into with a special API, or at least it was as of a couple years ago. Was added for drawing (i.e. Pencil). The scroll view predictive stuff simply didn't "feel" enough better to justify the complexity.

Show replies