Wow, I feel naive now for defending AMP providing better user experience. As it turns out, Google is evil and I simply did not want to believe it.
Unless, of course, the team leadership was on it and this was an insider job.
Whether front line engineers knew isn't that important. AMP Leadership definitely did.
did they though? couldn't this be a small (artificial delay) library injected server side by some higher ups at Google?
> Google ad server employees met with AMP employees to strategize about using AMP to impede header bidding, addressing in particular how much pressure publishers and advertisers would tolerate.
I think it’s easier to believe that the AMP team was simply naive, or willfully blind to the implications, or they didn’t care (probably all 3; teams are big enough that there’s room for all).
I think part of it was that the AMP team members were active here, and seemed genuine and competent. I'm somewhat curious how much they knew. This delay thing wasn't the only terrible part of AMP. The initial banner with the [x] button that sent you back to Google was terrible. As is the still-present hijacking of swipes and back button behavior on other people's content.
I did an Algolia comment search and can see my AMP rants getting more traction over a 4 year period.
Edit: I think it also had defenders because it forced sites like news organizations to re-think their 20MB+ ad and tracker laden story pages. So it wasn't all bad.
This was mainly it. Modern basically-just-text-content sites are hilariously bloated and slow, and this seemed like a genuine attempt to fix it. I remember I defended it for those reasons. It didn't pan out that way, though.
Also, I definitely remember AMP critics, or at least skeptics, getting plenty of upvotes from the very beginning, I'm not sure it was that hard to criticize.
Webdevs moaned because it restricted what they could do, which is partially why I liked it. They had to build a page that was fast, rather than fancy.
However I do see the downsides, as google owns the stack, limits your choice.
Also is doesn't help for dynamic search. The main situation I'm thinking of is I'm on moble searching for something thats hosted by a local news site. the AMP page generally loads in <2 seconds. The original page will take 5 megs of data, and ages to load.
100% totally decentralized.
https://github.com/ampproject/amphtml/blob/main/docs/spec/am...
Perhaps I'm wrong, but my interpretation of this ticket is that Google's AMP validator currently does not validate self-hosted JS, which means no special handling ("acceleration") on their search results.
https://github.com/ampproject/meta-tsc/blob/master/README.md
I assumed that the non-amp websites were slower due to the JS bloat, I still despise what the Web pages has become, however it turns out Google wasn't a honest broker here and instead of providing a tool that helps them and the users, they provided a tool that helps them and harms the users and bullies the website owners. That's evil and drains the last bits of trust I had left for Google.
In Australia, AMP pages load slower, not faster.
AMP breaks the iPhone Safari user interface, by not actually scrolling the whole page.
The URLs don't represent the actual site to which the user will navigate.
Logins break.
App-integrated links don't work properly.
And on and on.
AMP is a cancer on the web enforced by a monopoly for their own shady purposes.
This is factually incorrect: as the allegations noted, there are many cases where AMP slows page load time – it required a megabyte of render-blocking JavaScript with no fallback. I could tell which sites used AMP because they loaded slower and failed entirely if you had a less than perfect network connection — this was quite noticeable on the subway here in DC. I heard plenty of normal people complaining about it because it was a daily occurrence.
In addition to scrolling, the other thing which people noticed was that AMP broke the URL for sharing. This had several outcomes: one was that desktop users often got less usable mobile pages but the other was that it made spoofing easier because people trusted the google.com domain. This was used by fake news sites in the 2016 election cycle and at least the Russian government used it to spearphish investigative journalists.
They knew all of those risks in advance — and people accurately predicted them in 2015 – and the stated motivation was hard to reconcile with how much less work it would have taken to see the performance benefits by simply incorporating speed into search rankings at a high weight; this suggests that theories about the goal of controlling third-party ad markets were correct.
Luckily it wasn't a big problem, since Google also intentionally gimped the search results to an old mobile version when using Fx, so one didn't get amp links.
I guess this is my reminder to mention once again that using a Chrome useragent on FF mobile gets you better search results. It's one of the extensions you can download on FF for Android right now, in fact.
It clearly isn't only techies that dislike it so much that they actually pay to get rid of AMP.
Anyway, the AMP experience was a fresh breath in the world of 10-20 seconds webpage loading times. You see the lightning icon, tap it and the content instantly comes. Swipe right and left to go through the other results and everything is displayed instantly, top notch experience - especially when Googling for stuff from the news websites.
I agree. It's one of the imperfections of the AMP. I like AMP for quickly skimming between the results and then load the original page if I find something lengthy ro read.
It also strips out all the crap the site's UX/UI people use to try and keep you on the site.
I think most non-techies have no idea what AMP is to begin with.
It's just, once again, "my phone is doing some annoying and unexpected shit I didn't ask for".
Anyone whos usually logged in somewhere is impacted negatively
To re-cap:
1) Sharing AMP links is weird, feels wrong and Non-AMP links are hard to get
2) Breaks Safari's reader mode
3) Gives sub-par experience for websites that provide better experience when logged in since the AMP version is always the non-logged-in version.
None of those are deal-breaker for me because AMP pages come as a search results and fast loading results are very valuable for me when skimming to find the right one. Once I find it, I can switch to the full version(but Google makes it harder and harder. maybe this should be 4.).
- Left and right swipe events were hijacked, and sent the user to somebody else's website with a related story!
- The back button would go to Google, even if you had right or left swiped as described above...that is, non-google pages were not in the history.
Then, not just carousel pages, these applied to all AMP pages:
- A very tall banner at the top pushed your content down. It had an [x] button that would send you back to the google search rather than the expected behavior of removing the banner.
- Finding the real url was 2 clicks deep
And then just the overall weird smugness about calling it an open standard. Even though the "open standard" required including a google-owned javascript file via a <script> tag, and the validator would fail you if you didn't. You weren't allowed to host it yourself.