News for hackers. Stuff that matters.
Re:A letter about Google AMP by bsimpson on Tuesday January 09, 2018 @02:16PM (#16108873)
As I understand it, a round-trip request on a mobile network can be very expensive [0]. One of the ways that AMP enables sites to load instantly-upon-click is by starting those requests in the background. (Then, when the user clicks, it's just a matter of bringing the IFRAME into view - no network involved.)

Loading a traditional site can have side effects (e.g. reporting a view to an advertiser's analytics). By using the AMP validator, Google Search knows it's safe to preload and prerender an article, without triggering side-effects like analytics.

So it's more than just loading faster than some threshold; it needs to be safe to cache and prerender the content. The AMP validator is what asserts these are both safe.

(Note, I don't work on AMP and don't represent Google - this is my personal understanding.)

[0] Ilya Grigorik's time-to-glass talk https://www.youtube.com/watch?v=Il4swGfTOSM

Re:A letter about Google AMP by tscs37 on Tuesday January 09, 2018 @02:22PM (#16108927)
I have a very low datacap, I don't want a website to preload anything, I'd rather have a slow load.
Re:A letter about Google AMP by lern_too_spel on Tuesday January 09, 2018 @02:56PM (#16109349)
Preload is treated as a browser hint. Presumably, your browser knows you're on a capped network.

https://medium.com/reloading/preload-prefetch-and-priorities...

Re:A letter about Google AMP by tscs37 on Tuesday January 09, 2018 @03:05PM (#16109462)
That would be preloading assets, not the preloading of entire AMP sites as the GP describes.
Re:A letter about Google AMP by wmf on Tuesday January 09, 2018 @02:22PM (#16108940)
Best practices can reduce page load time from ~10s down to ~1s and then preloading can reduce it to maybe ~20ms. Google's "preloading or nothing" attitude and the refusal to discuss tradeoffs between market power and speed really isn't helpful.