Live data from Hacker News

How fast is AMP really?

timkadlec.com

151–160 of 167 posts

Re: How fast is AMP really?

#151
post #9

Earlier quoted context omitted.

Yes, except how do you propose to implement caching and stripping out the ad junk without AMP?

HN is seemingly incapable of rationally discussing AMP. AMP solves a very real problem, which is that our web stack is open to enormous abuse because it is sometimes too powerful for its own good. While we could make simple HTML that is as fast ( of course we can), the reality is that we don't, and many don't often for tragedy of the commons reasons. And the counter-argument that search rankings should just favor thi…

>In previous discussions I've noted that we need an HTML lite to counter AMP

this is essentially what the AMP team is building - there's a few proposed standards that have come from AMP, with respect to making pages cacheable and pre-renderable.

Re: How fast is AMP really?

#152
post #147

Earlier quoted context omitted.

If using uBlock Origin, you can try the example page with that cosmetic filter: ampbyexample.com##body:style(animation: none !important;) One thing for sure, the result is much more pleasant than having to wait 8 seconds if one does not want to un-block 3rd-party javascript from `ampproject.org`.

I suppose webmasters could add the inline css body{animation:none !important;} On mobile so I haven't had the chance to test

I should have included this in my first comment: this is a requirement from AMP to include the animation[1]:

    AMP HTML documents must contain the following boilerplate in their head tag. [...]
    body{-webkit-animation: ...
[1] https://www.ampproject.org/docs/fundamentals/spec/amp-boiler...

Re: How fast is AMP really?

#153
post #92

The first point in the article is the dead killer of AMP for me. My mobile provider gives me very little monthly data, I've stopped using AMP because preloading the AMP results is eating through my cap, with images disabled. Without preloading, AMP is slower than the non-AMP version of the same webpage with a simple script blocker. And usually causing more traffic too. I hope Google refrains from bullying website own…

No. AMP is much faster even without preloading. Here is an example of a random article on CNN: https://www.webpagetest.org/video/view.php?id=180402_4583919...

Re: How fast is AMP really?

#154

Heh. This reminds me of a recent pet peeve and observation of mine: progressive web apps are not faster. The Twitter Lite PWA, arguably the 'flagship' PWA ("developed in partnership with Google"), takes longer to display tweets for me than regular desktop Twitter.com, and the old 'static' mobile twitter site (which loads INSTANTLY[1] you can only get with user agent trickery) [1] Just did some testing. According to C…

In case of Twitter PWA, it loads twitts asynchronously, so they seem to load longer, but the whole PWA loads much faster in total: https://www.webpagetest.org/video/view.php?id=180402_08c5318...

Re: How fast is AMP really?

#155
post #38

Worth noting that it would be insane to preload content that didn't pass a strict filter akin to that of AMP. Each Google search would download dozens of MBs and produce a significant CPU workload, both resources being precious on mobile. Perhaps it was a communication issue: AMP should have been named/marketed "preloadable pages" rather than "fast pages".

AMP is not about preloading in the first place. It's about the fast CDN. Let's compare any NYT article in the AMP and standard HTML formats: https://www.webpagetest.org/video/view.php?id=180402_16d7083... - AMP loads much much faster.

Re: How fast is AMP really?

#157

I built a simple bash/javascript to get the AMP URL for a given URL, did reduce my data usage by tons. Look here: https://developers.google.com/apis-explorer/?hl=en_US#p/acce... curl -X POST -H "Content-Type: application/json" -d "{urls: ['$1'] }" https://content-acceleratedmobilepageurl.googleapis.com /v1/ampUrls:batchGet?key=YOUR_AMP_API_KEY For android app use: public static String GET(String url){ String result =…

You can try the AMP Desktop Extension, which uses this API: https://chrome.google.com/webstore/detail/amp-browser-extens...

Re: How fast is AMP really?

#158

Earlier quoted context omitted.

Try Brave browser. It's a modified version of Chromium with seemless ad blocking by the co-founder of Mozilla.

Can't edit my post since it's been 3 hours, but I'm very curious why I was downvoted. Was there some negative news around Brave that I missed?

Not aware of any negative news.

I presume it was because prior to my comment, no one had yet explained to the guy that you don't need to deal with such a proxy setup to block ads in browsers.

So, your comment sort of made it sound like ad blocking not being too great for him, was because of Firefox and that Brave would be better at it, whereas it really just was because of a missunderstanding of the guy.

Besides that, Brave is also controversial. As much as some people have screamed bloody murder, because Eich's free speech was supposedly taken away by journalists/SJWs/Mozilla or whoever, a good other percentage of people do think that his donation was wrong and that he should have reflected on it and distanced himself from this (past) opinion.

Furthermore, many people think him basing it on Chromium was wrong. No matter how pissed off he might've been at Camp Mozilla, and no matter how much journalists would've slandered him again, if he started supposedly trying to work with Mozilla again by starting a fork of Firefox, there's just no way that you can work for 17 years on anything and then a few months later go "Actually, this competing product is better".

Chromium being Google software and him trying to sell his Chromium-based browser as privacy-friendly does not help it either. There's millions of lines of code in there with tens of thousands of design decisions, all made by the biggest data broker on the planet. You're not patching all of those out in any reasonable time frame.

Re: How fast is AMP really?

#159

One detail which I feel must be mentioned: if you block `ampproject.org` by default -- directly or by blocking 3rd-party javascript by default, -- you will be "punished" with a 8-second delay before the page becomes visible. This is an entirely artificial delay, implemented through an inlined style CSS animation in AMP-based pages: animation:-amp-start 8s steps(1,end) 0s 1 normal both} @keyframes -amp-start{from{visi…

This kind of dishonest design is all too common now. If you pull back the curtain, you can find all sorts of other ways that the user is gaslighted.

One example is that the youtube mobile website will prevent you from playing a video in the background now. It didn't used to. This is to "encourage" you to use their Youtube RED app.

Re: How fast is AMP really?

#160
post #38

Worth noting that it would be insane to preload content that didn't pass a strict filter akin to that of AMP. Each Google search would download dozens of MBs and produce a significant CPU workload, both resources being precious on mobile. Perhaps it was a communication issue: AMP should have been named/marketed "preloadable pages" rather than "fast pages".

AMP is not about preloading in the first place. It's about the fast CDN. Let's compare any NYT article in the AMP and standard HTML formats: https://www.webpagetest.org/video/view.php?id=180402_16d7083... - AMP loads much much faster.

Not sure of what does webpagetest is showing evidence of. It could be anything... for one thing, AMP versions are much leaner so naturally they will load fast - even with no dedicated CDN at all.

Cloudflare and Fastly should be more than sufficient CDNs?

Admittedly, maybe the AMP CDN is an easier one to set up for the masses, but other than that I bet they all have about the same performance.

Post reply on HN