So far it just does read-only behavior, but in comparison to the Reddit mobile web app, it does enough network traffic to replicate the same view: (1) fetch the front page, (2) fetch your list of subscriptions.
If there was a way to incorporate mod controls and refresh just for the right pane, this would be a GODSEND for monitoring active threads.
I think there is also an illusion of unresponsiveness at play here. With a traditional interface, the browser doesn't refresh until the main content is already ready, whereas with the new framework, the new page is eagerly displayed, before the content is retrieved. So you have a period of noticeable lack-of-content and this is perceived as slowness.
A bunch of fake paragraphs that just resolve to "F" are an aggravating user experience.
Whoa, that's some strong language!
> never what the layout suggests.
I suppose that the current fashion's major problem is that the silhouettes are too sharply defined. Probably, one way to fix it is to have multiple of them with some offset, so that the silhouette suggest the kind of media (text) that would eventually be displayed, without also suggesting how they would be formatted.
> aggravating user experience.
That's a legitimate criticism for the choice of placeholder, but not pertinent to the current topic. My claim was that the lack of actual text presented to the user makes the user more conscious of the delay.
Using server-side rendering might help get the initial page loaded faster.
The endpoints themselves also seem to be very slow. I clocked a page taking 800ms to just return the bare html (vs. about 90 ms for hacker news).