I think Core Web Vitals were incorporated into their approach to prepare for the retirement of AMP. Pushing publishers towards industry-wide best practices is a good thing. Forcing AMP as a solution really wasn't.
It is their job to turn their power in to money, and AMP is just a power play. It is what tech companies do - if we don't like it, something about the fundamentals has to change.
My only complaint is the smirky, chipper PR front they package with it. Microsoft, at least, wasn't too insecure to show some fang without the Candyland faux-earnest horse shit when they were king.
It looks like it's no one's job, though. And the incentives for those third parties are even worse aligned with user interests, in many cases, than Google's incentives.
And I say this as someone that thinks that Google is the scariest company out there, right now.
It's not their job, but it's certainly in their interest (since it gives their users a better experience), and it's not illegal, so it's understandable that they do.
Id argue its their job to index the internet, not dictate how people build their webpages to determine the order they are displayed. They want their crawler to be faster and to capture market share/screen time.
When Microsoft was the dominant OS, they enabled ever-harder-to-ignore auto updates because without them, deployments of their OS could be used to harm other people's systems through remote attack exploits.
Google is also at a scale where they can improve the quality of everyone's web experience with their scale. It' not so much "their job" as "their obligation."
I completely disagree. Its my website, my servers, my bandwidth I pay for. Google has zero obligation to dictate how others go about creating their own product. They can enforce web standards sure, but AMP is horrific, and Im sure they next-gen AMP will also be counter productive to third parties.
By that reasoning: it's their search engine, their crawlers, their business to route people to the websites likeliest to be useful to those users.
In that sense, their approach is in some way more equitable than Microsoft's: they're not forcing change upon your system by way of mandatory updates, they're simply saying that if you don't play the same game they play, they're unwilling to do business with you.
If you're free to maintain your server to your standards, why should they not be free to maintain their search service to their standards?
Is google search there to provide the most accurate information or the fastest load information regardless of content? "The primary goal of Google is to provide users with the most relevant, highest quality results based on user search queries, i.e. their wants and needs when performing a search online."
These are not mutually exclusive. And no, "most accurate" isn't really guaranteed; their goal is basically a closed feedback loop: they have signal on whether people liked the result they yield and that up-signals that result for searches in similar context in the future.
Speed is valuable because it allows users to more quickly digest whether the result is relevant to them.
That is where I also disagree. Giving a higher rank to a AMP page that people like vs an page the is more accurate information seems like an issue to me. People are larger not smart enough to discern an accurate answer (look at how bad society wants to regulate FB, Twitter, etc for posting information that does not pass 'fact checks'). At the end of the day it doesnt matter, these decisions are made by what drives advertising dollars not relevancy or accuracy. Today faster browsing and crawling equates to faster advertisement display.
I don't think a search engine is the tool you seek; you're looking for an expert-opinionated resource for sifting data, not an automated system to retrieve some data on arbitrary topics.
Microsoft enabled harder to ignore auto updates because users didn't want to update. If a user doesn't want to update, they shouldn't have to, as it's their computer. It's not their OS being used to harm users, it's users making their own decisions and accepting the associated consequences. Google should not be trying to force people to obey web standards, they should try to let users make their own decisions.
> If a user doesn't want to update, they shouldn't have to, as it's their computer.
And then they plug that computer into a global multi-user network and their machine is botnetted and used to harm other users. In that context, people are no longer making simple decisions and accepting the consequences; a tragedy of the commons is instead created.
Your thinking works when computers are isolated from each other. When they're not, it's in the same category as "states require annual vehicle inspections." Because when you're sharing the road with other drivers, you owe it to them that your vehicle is unlikely to undergo catastrophic rapid disassembly.
Right, but when I search "cookie recipe", there are millions of pages that match in the index. Yes, popularity is a good starting point, but even that has its limits. All things equal, I'd much rather have a page that loads in 1s than one that loads in 10.
I'd also much rather have a page that brings me to a cookie recipe when I search "cookie recipe", but Google ranks pages with tons extraneous text, images and ads above your bare bones recipe site. This has given rise to an immense amount of blog spam on Google, where you have to scroll through paragraphs and paragraphs of inane content created by an army of underpaid writers just to get to the relevant information you searched for.
They don't dictate anything - I am perfectly free to try to set up "the last page on the internet" which will be penalized in rank compared to anything else.
They would indeed prefer to be utterly ignored by designers over being gamed to but SEO remains something optimized for like clickbait titles and headlines.
It sounds like AMP will stick around, serving as the easiest way to guarantee a great Core Vitals score (because it's so locked down), but that you're welcome to find other ways to get a great score yourself. Which is how it should have always been.
I'm not super happy with the Web Vitals, either. They seem to be pushing the Web toward deploying a lot of otherwise-unnecessary JavaScript cleverness that ultimately presents itself to me in the form of a generally more obnoxious experience. Older and non-Chrome browsers may have difficulty rendering it. Total bandwidth consumption can go up, because it's A-OK to load a bunch of enormous resources later, just as long as it doesn't happen on initial page load, or take longer than 4 seconds to happen, as measured by the company with the fastest Internet connection in the world.
Long story short, I don't see it as best practices that serve Internet users as a whole. They seem to be more closely tailored to the interests of Google. And, by extension, the subset of netizens from which they can generate the most revenue.
I just looked into the search console for the first time in forever, and I'm not sure how Core Web Vitals would be pushing for unnecessary JavaScript. A website consisting of 100% static content seems to be fully okay according to their metrics - but funnily enough, some months ago they claimed that 50% of the pages were served too slowly on computers, while 100% of pages were OK on mobile. Same static server. Now it's the other way around, they randomly say that a random selection of pages is too slow on mobile, and it changes every day. Maybe it's just their own connection that sucks? It also doesn't help that they keep using acronyms for the broken stuff that aren't explained everywhere on the page itself.
Whilst there is an element of lab measurement involved, they do use field measurement, so metrics are collated from users rather than their own connection. This means that your data could just as easily be skewed by a browser/OS update that rolls out to a ton of devices at once, as much as anything at your end.
Do agree that the proliferation of acronyms doesn't help with wrapping your head around it all!
It's how the combination of the three encourages you to do things if you want to have a site with rich content. You're supposed to paint the page ASAP, so you don't want to defer loading any large content, but then you're not supposed to have the page layout shift around at all as you dynamically load all that content later, so you've got to do clever things with placeholders and swapping out content and whatnot. You've got up to 4 seconds to load all that stuff, which is enough to load an enormous amount of data over a fast internet connection, so much so that the same amount of content might take minutes to load over a slower connection. Fortunately, they've chosen methods for measuring that metric that are heavily biased toward measuring the experience of people who have 24/7 access to broadband.
So, yeah, Google may want to encourage a nice Web experience, but they don't want to back this with metrics that might discourage people from sending too much business in AdSense's direction, or fail to favor Chrome over alternative browsers.
Perhaps the mobile loading time tolerances are larger, meaning that if the timings between desktop and mobile are largely the same, they could have different results.
> Maybe it's just their own connection that sucks?
Your Core Web Vitals report in Search Console is based on Chrome User Experience Report data. Meaning that this is data from your real users, not Google running simulated tests of your pages on their own servers. I.e. when someone loads your page from Chrome, Chrome reports back how long it actually took the page to load for that user (it doesn't happen with all users, they have to meet various opt-in criteria [1]). So, if you see that 50% of pages are served too slowly on computers, it means that 50% of your real users actually experienced slow page loads (as measured by the Web Vitals metrics). Perhaps your static site isn't as efficient as you think, or your server is slow, or the devices/connection of your users is much worse than you assumed. That's the power of this data; it shows you that in the real-world the experience isn't as great as you're assuming and encourages you to investigate further.
(For the record) The landing page of the Core Web Vitals report does indicate where the data is coming from. Next to "Source: Chrome UX report" you see a question mark. If you hover over that question mark then click the "Learn more" link it takes you to this page: https://support.google.com/webmasters/answer/9205520?ref_top...
Disclosure: Googler working on https://web.dev. I'm not on the Web Vitals team but interact with them.
The existence of first-contentful-paint, as well as the page speed index, would suggest they care to prioritize for these factors as well and not simply defer to after load.
Both of those metrics account for the visual completion of the page relative to its final appearance -- deferred resources would slow that.
It seems that Google's Pagespeed Insights, which is the basis for Core Web Vitals, by default considers content paint performance to be (a very arbitrary) 4x slower on mobile than on desktop.
Is showing full screen images on desktop an industry-wide best practice? How about the faux navigation bars meant to resemble browser chrome?
This "Google was just forcing publishers to fix their pages" meme desperately needs to die. Just consider all the extra standards crap they were pushing to introduce to perfect the deception. This was, as always, about owning the data.
Except... they didn't own the data? AMP was conceived because Google was worried about everything moving into Facebook's walled garden. That's what it was competing against.
Google doesn't need to own the data because Google is the world's gateway to the open web. They don't care who owns the data as long as they can crawl it.
To me AMP was just Google trying to turn the web itself in its own walled garden (after many failed social media attempts).
And, as I read in the article, it looks like behind this move there is some current "Antitrust Pressure" plus publishers quite pissed off about losing both control and revenue themselves (as much as 39% less conversions, they says.).
Clearly AMP was way more in Google master plans than a poor web performance palliative.
[Edit]
This was the article linked to this story when I commented (now changed to some Google Dev docs):
I never saw any convincing argument that AMP helped pushed a walled garden, at least not in any common accepted sense of that term.
It was open source, it was used by many others including competitors, it was optional and it didn't block access from anyone. Having an AMP version in no way "locked" you to any garden, AMP versions aren't even meant to be the canonical page anyways.
It may have had a lot of issues, but "walled garden" would not be one of them.
The carousel setup it enabled was certainly a (soft) walled garden. It hijacked the top portion of a publisher's page, the back button, and swipe actions, resulting in more time spent on Google.
Exactly. Walles garden would be requiring publishers to send articles directly to Google if they want to enable instant loading (like Apple News). Instead, publishers publish their articles publicly in a way that Google's competitors can and do consume.
That doesn't work too well with Firefox for Android, because Mozilla disabled most addons (there's like only 11 addons that are approved). I think it is possible to enable them in Firefox for Android Nightly[1] by creating an "add-on collection", which is kind of a pain and the first extension that I tried doesn't appear to work correctly (Redirector) and the browser just crashed after disabling the addon.
Another way would be to stick with Fennec v68 from fdroid, but that is not such a great idea, but at least the extensions do work.
Oh nice, I might try that out. The new FF Android has been such a bad experience for me from the start, and has a long way to go to catch up with the old version, imho.
I could not stick with the new Firefox for Android for more than 1 day... they didn't even implement something as simple as keyboard shortcuts (I understand that they have their priorities, but there are so many reasons why Fennix is worst then the previous engine Fennec, not sure why it was released in Alpha state as a finished product (did Google pay them to do this, IE: if you cripple your browser, I'll buy you a yacht)).
Kiwi Browser is the way to go since Mozilla decided that it knows better than it's users which addons should they be able to use or not. It's basically Chromium with all it's addons available.
On the one hand, the addon has relieved my AMP issues; I no longer have any issues using AMP links that get sent to me. On the other hand, this solves the problem for me but not others, and I'm less likely to notice the problem and reply all or leave a comment with the non-AMP link.
Also, it's available from the same developer as an addon for Chrome desktop; I suppose if you were just annoyed by the experience and don't care about de-Googling it works there too.
There's only so much one can do. At work I had to even implement AMP at the request of an upper level, requested by marketing or the seo team. Everyone hated it but I can totally understand it: it made total business sense to appear in the Google rolodex search result
The thing that frustrated me most from an e-commerce perspective, is that Google were pushing AMP shopping experiences and claiming improved user experience, but provided no data to support this.
Probably doesn’t matter since the war on urls is mostly a myth perpetuated by outrage-bait blogspam.
By default (i.e. until you click the bar) only showing the part of the URL that the site doesn’t control is really good for security and avoiding phishing. Now apple.paypal.secure.wendys.scamsite.info/payus/wwwcitibank just shows scamsite.info. This is a good thing! And doubly because URL paths and fragments have been made largely irrelevant to end users as app routes.
If you're going to rant about the death of the URL you should complain to HN as well that only shows the domain next to posts.
If we're at the point where we're de-emphasizing something because it's almost always visual noise and you don't need to see it often why is "hide it until the bar is selected" worse?
I work at an Alphabet company, so at work I use Chrome. Lots of internal tools are really buggy on anything but Chrome, so I have little choice. There's lots of copy-paste of URLs, including fragments, and it's really jarring to have the text suddenly change right after a mouse-up. It's surprising and unpleasant. And of course the browser tries to modify the selection between mouse-up and copy to make sure I get the whole thing, so now the software is not doing what I told it to.
It seems harmless to show the full URL and not bait-and-switch the text upon interaction. I don't see it as the browser's job to protect users from their incompetence.
Chrome already visually de-emphasizes the URL path, clearly they don't think it's enough.
If you care at all about the health of the web, you should support browsers in protecting users from their own incompetence. Users don't blame themselves for phishing & fraud, when it runs rampant on a platform they simply switch and the end result is that native walled gardens that do protect users win over the web. This is exactly what happened with news on the mobile web; mobile news websites got so slow that Apple & Facebook started gaining a lot of ground with completely closed & proprietary news platforms (Apple News, Facebook Instant), until Google released AMP which both helped & pressured news orgs to get their shit together on the web.
You started with protection and ended with performance, but I don't see where you connect them. Could you clarify?
I see these kinds of features sort of like padding on football players. It enables (encourages?) carelessness. Maybe the jury is still out on the net outcome, but I'm so far away from web development and the overall ecosystem as to have no opinion.
The common theme is that in both cases, the web was losing ground to closed, native platforms. If web browsers don't protect users, they stand to lose users to platforms that do protect users in much the same way that slow websites lost users to Apple News & Facebook Instant.
Of course it encourages carelessness, that's what users want: they want to care less about the tech they use.They simply want the benefits of tech, not the maintenance or burden of care. It's why many prefer an iPhone that makes all their software choices for them. It's why many trust an email provider to sort & filter their email instead of doing it themselves.
I mean it's literally one click away. Let me tell you how relevant /reply?id=25150055&goto=threads%3Fid%3DSpivak%2325150055 is to me on this page. If it was id=25151055 I would be worried.
What does the "war against URLs" even mean? Sounds like it could be one of two things:
1. Chrome only showing the domain in the address bar
No, I doubt it'll make any difference here. Google's reasons for doing that are anti-fraud, not anything to do with AMP. And (I'll give up shouting this into the void one day, I swear) Apple did this with Safari years ago and no-one cared. It's outrage for outrage's sake.
2. The whole "web packaging" format issue
Google proposed a standard that let you "fake" the URL of a page (albeit with cryptographic signing to ensure it actually came from the right host). This might be dead in the water, yes. Mozilla and Apple already came out against it, and one the primary use cases was AMP. I doubt Google will un-implement it any time too soon, but I think it'll end up being a weird edge thing that very few people care about or use.
> (albeit with cryptographic signing to ensure it actually came from the right host). This might be dead in the water, yes.
Which is a pity, because this could have been extended into a way to allow browsers to pin a specific version of a web app and warn the user if the server sent something different the next time they visited.
Having a TOFU security policy for web apps would have made them roughly comparable to native apps for certain threat models, especially considering that native apps often auto-update without giving the user a meaningful chance to check whether the update is backdoored.
For both web and native apps, though, it would be nice if there were independently run Binary Transparency logs published so that apps would only download an update whose source code had been publicly available (and reproducibly buildable) for some reasonable amount of time to allow an audit to occur.
There's not many issues that will gather enough reactions for Apple to ever respond. It would need to be something like removing the eggplant emoji to have anything sizeable from their perspective.
A decent amount of people cared about the move to hide most of the URL, just not enough for it to be a wide societal protestation wave.
The problem here is that google has an interest in hiding the URL other than anti-fraud. Apple isn't a major player in the internet market and they have a history of "dumbing down" features, or at least reducing their apparent complexity, to make the experience friendlier to the average user. So when Safari hid the full URL, that didn't raise many eyebrows.
Many of Google's moves in recent years, most notably AMP, appear to be part of a larger plan to make the internet require Google to function. Messing around with the URL visible in the address bar seems to fit in with this plan to reduce the average consumer's awareness of where exactly they are going on the internet.
I hate AMP as a user. It breaks my browsing experience in strange and sometimes unexpected ways. (like certain scroll features can stop working, sometimes weird features like search in page can also break. Sometimes when you try to highlight some text, it will highlight a whole section. And the solution is always --> click the button to get to the actual page and then it is fine)
No to mention it’s practically a dark pattern that they push on you. I hate how they’ll send me to AMP links and then the UX to get the original URL I wanted and not some abomination of a google amp URL is terrible and they purposefully make it as subtle as possible that you’re not on the actual domain or site that you wanted.
The funny part is that when I tried the latest version on Chrome on Android 11 back in September, whenever I clicked on an AMP link, it would think Google's AMP domain was a phishing site.
I have been using DDG full time for a month or so now but they really need to work on making their search results more readable. I might use a browser plug-in to customize the styling myself when I get some free time.
IIRC, DDG lets you configure fonts, sites and colors for results, somewhere in the settings. Works well for me, especially in conjunction with their "cloud save" feature.
In my experience Google is now about as bad as their competition was when they broke through: lots of irrelevant cruft on every single search, and not just spam but completely unrelated pages that doesn't contain a single keyword from my search.
DuckDuckGo isn't fabulous either but it is faster to retry in Google from DDG than the other way around so it is my default search engine now.
I wonder: why is it soooo hard - for both DDG and Google to just respect + or "" or the verbatim option in Googles case?
Because an empty set is far more valuable to me than a rich set of irrelevant results?
(Lately it seems I've been assigned to an experiment that has slightly better results and also shows me the context they think are relevant in the results page. That helps immensely, but of course I have no way to get that to stick :-/ )
PS: If anyone wants a billion dollar idea, recreate Google from before the DoubleClick acquisition: contextual ads, relevant results. Because that spot is empty, and given Moores law, the fact that the original PageRank patent has expired and more it should cost a tenth today vs then. Also Google seems to be collapsing under their own weight so there's less chance they manage to catch up.
PPS: I'd pay 10USD a month and accept contextual ads on top of that to get back something like old Google.
Honestly, I don't think google-like search engines are the way to go anymore. They're being gamed. That's fine for things like Stack Overflow or finding the background of a person. But it's not good for finding a place to eat, a book to read, health advice, a specific game genre, or the best tom yum recipe. Google subsidiearies like YouTube and Google Play do a really poor job at this, and Google isn't much better.
I think what might work better is an unweighted filter. A google style search engine filters out all tom yum recipes below a certain quality or all health sites below a certain credibility level. And then it tosses you a random one. Google tries to make a one size fits all engine, so it's doing very poorly in many of these. But Bing/DDG isn't faring much better. Stack Overflow and wikis are basically just search engine plugins
I especially hate it for reddit AMP links, since 100% of the time I have to click on the original link anyway to read the full comments, so it's an extra middle step for no good reason.
Absolutely not. AMP is still around and already integrated deeply into thousands of high-traffic sites.
Google is probably just trying to reduce the likelihood of being broken up. AMP is widely adopted, so the damage of their preferential treatment has already been done.
How do we help it die then, it's the most unwelcome, undesirable crap I've seen in years. Google used all its power to force it down everyone's throats, but it sucks so badly and nobody wants it.
Not publishers, who recognize it for what it is (a very, very thinly veiled attempt at stealing users, traffic and analytics, and Google building its own walled garden).
Not users, for whom navigation is broken, links and scrolling behave weirdly or break completely, and the address bar can't be trusted any more.
A lot of news aggregators and RSS feeds will default to the AMP URL. People post and send AMP URLs all the time. You can see it more and more on reddit lately.
Really unpopular opinion: Tech break ups are a crack pipe dream.
Parallel divisions will just result in the most recognizable names winning out. There is a simple and wrong idea that they can just break off Youtube, Android, and the search engine. Except
that has as much logical coherence as saying Walmart should not be allowed to own their own pharmacies because they have too much dominance in groceries. None at all.
They only have angry old confused vested interests trying to protect their fiefdoms and no plan. It is a replay of the "Repeal and Replace!" bullcrap where after going on propagandizing for years prove they never had a goddamn plan and never tried. Mark my words - these techlash morons will be the dog who caught the car if they actually got what they think they wanted. Yet again like with the link tax where they complain about free advertising and then are outraged when nobody decides to link to them.
At this point I just wish all of those morons going on about antitrust would shut the fuck up because they know nothing but useless demagoguery and making things worse.
Google's solution to this is to basically build the distributed web so that browsers can show the correct origin and URL when they hit Google's AMP proxy.
I think you've got this backwards. This is "democratized CDNs" not "AMP for Everything." Web Bundles are a replacement for AMP that's not Google specific.
With Web Bundles the only thing that a cache can do is serve you an immutable asset that is signed by the original publisher exactly as if you had connected to the origin server.
Apple sort of does this on mobile Safari and its a frustrating / unexpected experience.
If you are on a website in mobile safari, and you hit the share button, and select Copy. What gets copied into the clipboard is actually the Canonical url rather than the actual one you are at.
So for /amp/ pages, it ends up copying the non-amp url.
imo, it's a bad experience because sometimes canonicals and non canonicals aren't exactly the same. For example, if I'm looking at a specific filter or sort, then it might make sense for the canonical to be the non-sorted version. But if I'm sharing a specific url, I want it to share exactly what I'm looking at.
The article and some of the comments here seem to be over-interpreting this change.
This is NOT ending AMP, and I don't know where anyone's getting that idea.
It will continue to rank by "page experience signals", and presumably AMP pages will continue to be the standard to beat.
In other words, if you can go ahead and replicate AMP's loading speed on your own (maybe self-hosted AMP?) then awesome.
I do hope sites do that, because sites loading instantly rather than taking 5 seconds is a huge win.
But in practice non-AMP news sites are still filled with bloat that makes them take 5 seconds to load.
AMP has massively increased the performance of many news publishers forced to switch to AMP. We'll see if publishers are now motivated to try to make their native sites equally fast. I'm not holding my breath, though. I assume they'll stick with AMP because they don't want to re-architect their sites, or don't have the technical bandwidth to do so.
But this way Google appears more neutral with "objective" page performance measures, rather than explicitly favoring any single standard.
Yeah I hate AMP on principle but unless I’m really mindful about it my brain unconsciously clicks the AMP links because the experience is so much better
It's not, though. If the fastest pages still get promoted to the top, and AMP results in the fastest pages, then the incentive remains just as large as before.
And the NYT quote doesn't suggest they'll move away at all. It just says it's "important" for Google not to make AMP a "requirement".
> If the fastest pages still get promoted to the top, and AMP results in the fastest pages, then the incentive remains just as large as before.
The first two components of your sentence are conditionals. Even IF they're both true, the very fact that they are uncertain, whereas before they were certain, makes the incentive MUCH MUCH smaller than before.
Any uncertainty at all will make internal discussions with technical leadership on maintaining AMP infrastructure weaker in companies like NYT. If they don't drop it immediately, there will at least be a ongoing internal dialogue about potentially doing so, brought up any time bugs need fixing in, or resources need reassigning to, AMP infra.
Yeah, AMP sites I get load really fast. But that's not really useful when they come without all of the original page functionality or don't even show all of the content. That's my gripe with AMP pages.
> This is NOT ending AMP, and I don't know where anyone's getting that idea.
Where the idea might come from would be that many (particularly large) websites would have implemented AMP reluctantly, and currently incur extra maintenance cost for AMP as part of a deliberative ongoing cost-benefit analysis. This removes one of the major obvious benefits in that analysis (AMP pages may still benefit but it's harder to qualify), so there may be a stronger technical argument within companies to drop support.
> It will continue to rank by "page experience signals", and presumably AMP pages will continue to be the standard to beat.
This is a tricky prospect. A concrete "we will treat your AMP page preferentially" is a lot easier for tech managers to grok than "AMP stands a good chance of scoring well on metrics we prioritise, so may be preferred, if you believe you can't achieve these metrics without AMP".
On the flip side, it's completely conceivable that Google could continue to treat AMP pages preferentially while claiming it's due to proxy metrics, since their algorithms are not public.
Isn't that the point, though? I'm more than happy for Google to use loading speed as a signal in their search rankings. The issue was that sites were penalized for not using AMP even if they had equivalent or better performance.
If AMP becomes a lower bound for performance and non-AMP sites slim down their pages to compete, great. If sites that don't compete on performance die off as their search ranking decreases… also great!
Or if everyone continues to use AMP because they are not able to match the performance, that’s also great? :)
The entire AMP design was driven by the need to make same origin serving of the content possible in a safe manner. That is where the performance gains come from, not from having lighter markup. And that is still just as true now as it was five years ago.
AMP pages are bloated compared to browsing with NoScript and ads blocked. It isn't hard to deliberately beat AMP performance with enough motivation to make it happen. There needs to be an upheaval of ad vendors to make self hosted images without JS viable.
Same origin loading also allowed Google to take over the top of your page, hijack swipe actions (to send your visitor to a competitor), and so on. Some publishers may value moving away from AMP to prevent that sort of thing in the future, despite perhaps a small cost in performance.
The entire AMP design was driven by a desire to put Google in charge of publishing, heading off Facebook. Putting 1MB of render-blocking JavaScript in the critical path was never good for users - just ask the Chrome developers! - but it was a strategic play which fortunately failed both because it wasn’t as fast as claimed and most publishers didn’t want to cede control of their UX.
Why does the same origin matter? There's no substantive difference between my page being served from Google's edge network vs. from Cloudflare's or Amazon's.
At first I thought 5 seconds was a long amount of time, and then I loaded my company's website and noticed it really does take 5 seconds to load. It's a self-hosted Wordpress site on the GB business internet in our office.
How can I possibly compete with AMP? I notice that even Reddit.com takes about 3 seconds to load. cnn.com, 2 seconds
Making WordPress fast requires some pretty serious caching effort, in my experience. Your internet connection might not be quite up to snuff for server hosting either, although you can test that with ping and checking the time to first byte.
Also, Reddit isn't exactly fast—it takes six seconds to load for me. Wikipedia and HN are better examples.
If you want the very lowest bound possible on latency for your internet connection, try loading https://cloudflare.com/cdn-cgi/trace. That should load in milliseconds. From there, you can do a little more profiling work on your site.
There are plugins that help automate the easy stuff and provide lots of bang for the buck. I took a day a while back and got my local gaming league’s Wordpress site down from something like 6.5 seconds to 0.99 seconds by switching hosts, removing heavy plugins, compressing images, and lots of caching. Totally doable if you are mostly static content (which should be all news sites).
Because doing so would communicate with the publisher's web server before the user has clicked on any results, deanonymizing the user to the publishers of all the results on the page.
This is the entire point of AMP, yet 90% or more of the commenters on AMP articles don't understand it, which makes the comments completely useless.
Is there a source for this claim? I’ve never read anything about AMP that claims the intent is privacy.
Furthermore, I don’t see why AMP specifically is needed for what you describe — it seems to me that Google could do the exact same thing with normal webpages.
> I’ve never read anything about AMP that claims the intent is privacy.
Why else do you think it loads from Google's cache?
> it seems to me that Google could do the exact same thing with normal webpages.
Nope. The publisher needs to opt in to having their content served by Google; and the ad metrics, analytics, and login need to be delayed until the user clicks on the result. That can't happen without cooperation from the page. Hence AMP. I don't even do web development, and the design of AMP is mind-numbingly obvious even to me.
This is a smart play by Google now that they're under antitrust scrutiny. Instead of abuse their market position by forcing AMP down publisher's throats, they can do much the same thing but with an appearance of neutrality.
I think most of us here know that 99% of publishers will not be able to match AMP in performance, so the incentives haven't changed.
I like the change though because this will force everyone to pay attention to website performance - something that's often an afterthought. We used to be able to use the internet over a dial up modem and it was OK, not great, but OK. Now that would be nearly impossible.
I've noticed that without the benefit of blockers, pages load quickly and then progressively degrade as intrusive assets roll in. I hope this change punishes that behavior as well.
I continually get emails from Google, "Search Console has identified that your site is affected by 1 Mobile Usability issues:
site: Content wider than screen"
The thing is, the software I have on http://doomlaser.com is only for desktops. And I just noticed that a direct search for one of my apps, Cursorcerer, is now ranked #2 behind an entry for it on MacUpdate. A little frustrating. I guess I could redesign for mobile, but I would really rather not. Over 90% of my traffic is from desktops for a reason. Whatever happened to getting "the full Internet, not a mobile Internet" on your phone?
>I do hope sites do that, because sites loading instantly rather than taking 5 seconds is a huge win.
>But in practice non-AMP news sites are still filled with bloat that makes them take 5 seconds to load.
I wish people would stop saying "AMP means pages load instantly" as though it's an absolute. My experience with AMP pages has been the opposite; increased loading time, or pages that often don't even load at all. Come to find out in a HN thread from last year[0], AMP pages add an 8-second blank page delay if the user disable's Googles third-party js.
Going to be an unpopular opinion here but as an end user I am disappointed to hear this because AMP was a great improvement to the performance of my mobile browsing experience. I understand that AMP isn't required for a great performing mobile site but before AMP no one seemed to take the steps to deliver one without it.
I have to, begrudgingly, agree to that point. Most major news sites are barely usable without multiple content filters (ad blockers, etc.). AMP really pushed them into making things load quicker and with less weight.
The point of this change is that it now allows more high-performance sites into Google Search, by allowing both AMP and non-AMP sites that are high-performance. It's good news.
Nothing here makes AMP less able to deliver high-performance pages.
I have the same experience: unbelievably broken on iOS.
It just has to be better on android because somehow their own engineers have to use it day to day and ship it in good conscience at the same time, right?
If a site is able to reach the same performance without AMP (which is certainly possible), then I don't see what the issue is. At the end of the day we want fast pages. AMP with the incentives was a good way to force sites to be faster, but if they want to achieve it their own way, I don't see an issue either. As long as the overall bar is not lowered.
AMP was one of the things that pushed me off Google search and off Google Chrome on Android. The experience was mostly worse; especially reddit AMP is really awful, even given how awful non-AMP reddit is.
The few things that bothered me the worst were, blank white pages while loading, to earn the 'one contentful paint'; fonts loading late; and it's difficult to share links, because they had the garbage urls because google was proxying. The fake address bar was the icing on the cake.
To me, completely breaking down when I use the very normal browser feature "pinch to zoom" on iOS has always seemed wild. I have no idea how they get away with this.
For me the biggest problem with AMP was that it was not delivering comments on a given article. Often the mainstream journalists have an agenda and if you see that comments are disabled under an article, it is a strong indicator that it is a propaganda piece. If comments were enabled, however, you could see many counter opinions to a given article that would give you pointers where to look for the true picture of what happened. AMP makes for lower quality news and just because of that I avoid it, but it also cuts Ad revenue from the outlets that use it.
There are plenty of major sites where AMP is flat out worse, even beside the inherent clumsiness of AMP caches. Reddit and The Guardian spring to mind.
It should have been a generic "If your site takes too long to load or lags while scrolling we will heavily derank your page and visibly show you this so you can improve"
Page speed is the one metric you want SEO nuts optimizing.
I know that AMP isn't going away, but man I won't miss it. Android and AMP just never seem to get along. The number of times I've clicked a link to watch some video that won't load because of AMP is too high.
Just the other day, I saw almost everything I searched on Google being AMP, and sharing was broken. I had to find and click the little share icon by scrolling up and having it appear. Terrible for most users who won't discover it!
Great news, I'm working on a personal project and this weekend I was scheduled to implement the AMP version. I'll put the time into the "normal" site optimisation rather
I honestly see a successful AMP as the end of the web as we know it--there's no way for us to know if the content isn't censored by Google. I'm certain their plan is to sell their hosting service and show providers that they could use Google services. What's the logical next step?
> I honestly see a successful AMP as the end of the web as we know it--there's no way for us to know if the content isn't censored by Google.
Is there any way to tell.that the content of an HTML page isn't censored by the hosting provider used by the page owner, or by the browser vendor?
Doesn't seem AMP adds anything to that. Sure, you can't know, in theory, that the host (which may not be Google, Microsoft and others run AMP hosts) has censored it, but what does that actually change vs. HTML?
It would be nice if Google gave an option to disable AMP. I don’t have any philosophical objection to it, but too often it causes rendering glitches and other annoyances on my iPhone. It also makes it annoying to copy the link if I want to save it or forward it etc. The performance increase is from AMP is pretty marginal and not worth all the hassle for me.
Android users can use kiwi browser (chromium) to disable amp plus some other cool stuff like extensions. secondly, almost all websites provide share button to copy absolute url of that webpage.
I hope this might be the first step at least to Samsung Internet offering an extension to block all AMP sites, so I can at least stop seeing AMP sites ever, ever again.
quick tip: if you use the system share button to copy the url, you get the real one.
You can also take advantage of this with an app like Opener to navigate to the real page in the cases where the amp site is in such a broken state that you cannot get to the fake browser bar.
When I was struck in a developing country in March-April during pandemic, I was glad for AMP to minimise data and save costs. I am saddend that hn does not recognise the different (poor) parts of the world need quick/bloat free websites.
Yes, ideally all webdevs would make the non-AMP page bloat free. But as you know many a world pages are horrible and load tons of trackers.
Why is there is clear hate - AMP can be used for good and bad.
Because what Google is doing now is what they needed years ago: punish bloated sites based on real experiences. AMP was an attempt to block Facebook news and use search dominance to give them control over how publishers use the web.
If it had just been about performance they wouldn’t have needed so much extra work and they wouldn’t have ignored the performance and reliability drawbacks to putting so much JavaScript into the critical path for page loads. Those 5+ second AMP mobile page views were completely avoidable if your goal was performance rather than control.
uMatrix + uBlock Origin solve this much better than AMP does... with the advantage that it works for exactly the websites you want with exactly the resources you choose - no more, no less. uBlock even allows you to block images if they're larger than a certain size.
This. uMatrix is a godsend for people who want to control loading 3rd party assets. It usually requires some configuration for new sites but once configured my phone runs like a dream.
(ran, since firefox disabled the plugin on my phone)
I hope they don't use Lighthouse and PageSpeed blindly - as the score does not correlate significantly with perceived load time and performance, and is also easy to game!
Have you tried running Lighthouse and PageSpeed on a site that serves up Google Ads via AdSense? My site can be optimized like crazy but still score low because of Google's javascript ad loading code.
Like spending two weeks optimizing site load - from 2 seconds to 200ms - and then adding Google and Facebook analytics/trackers and you are back to 2+ seconds load time again.
279 comments
[ 2.8 ms ] story [ 275 ms ] threadMy only complaint is the smirky, chipper PR front they package with it. Microsoft, at least, wasn't too insecure to show some fang without the Candyland faux-earnest horse shit when they were king.
And I say this as someone that thinks that Google is the scariest company out there, right now.
It's absolutely Google’s job to decide what quality signals to incorporate into their search rankings.
Google is also at a scale where they can improve the quality of everyone's web experience with their scale. It' not so much "their job" as "their obligation."
In that sense, their approach is in some way more equitable than Microsoft's: they're not forcing change upon your system by way of mandatory updates, they're simply saying that if you don't play the same game they play, they're unwilling to do business with you.
If you're free to maintain your server to your standards, why should they not be free to maintain their search service to their standards?
Speed is valuable because it allows users to more quickly digest whether the result is relevant to them.
And then they plug that computer into a global multi-user network and their machine is botnetted and used to harm other users. In that context, people are no longer making simple decisions and accepting the consequences; a tragedy of the commons is instead created.
Your thinking works when computers are isolated from each other. When they're not, it's in the same category as "states require annual vehicle inspections." Because when you're sharing the road with other drivers, you owe it to them that your vehicle is unlikely to undergo catastrophic rapid disassembly.
https://joyfoodsunshine.com/the-most-amazing-chocolate-chip-...
The second recipe result is Betty Crocker and as you’d expect — recipe at top, steps with photos after.
I personally find more of the former than the latter when looking for specific recipes (Red wine chocolate cake was my most recent search)
They would indeed prefer to be utterly ignored by designers over being gamed to but SEO remains something optimized for like clickbait titles and headlines.
Long story short, I don't see it as best practices that serve Internet users as a whole. They seem to be more closely tailored to the interests of Google. And, by extension, the subset of netizens from which they can generate the most revenue.
Do agree that the proliferation of acronyms doesn't help with wrapping your head around it all!
So, yeah, Google may want to encourage a nice Web experience, but they don't want to back this with metrics that might discourage people from sending too much business in AdSense's direction, or fail to favor Chrome over alternative browsers.
Your Core Web Vitals report in Search Console is based on Chrome User Experience Report data. Meaning that this is data from your real users, not Google running simulated tests of your pages on their own servers. I.e. when someone loads your page from Chrome, Chrome reports back how long it actually took the page to load for that user (it doesn't happen with all users, they have to meet various opt-in criteria [1]). So, if you see that 50% of pages are served too slowly on computers, it means that 50% of your real users actually experienced slow page loads (as measured by the Web Vitals metrics). Perhaps your static site isn't as efficient as you think, or your server is slow, or the devices/connection of your users is much worse than you assumed. That's the power of this data; it shows you that in the real-world the experience isn't as great as you're assuming and encourages you to investigate further.
(For the record) The landing page of the Core Web Vitals report does indicate where the data is coming from. Next to "Source: Chrome UX report" you see a question mark. If you hover over that question mark then click the "Learn more" link it takes you to this page: https://support.google.com/webmasters/answer/9205520?ref_top...
Disclosure: Googler working on https://web.dev. I'm not on the Web Vitals team but interact with them.
[1] https://developers.google.com/web/tools/chrome-user-experien...
At least they link to a lot of docs and advices.
Also funnily enough they complain about Google Analytics.
Both of those metrics account for the visual completion of the page relative to its final appearance -- deferred resources would slow that.
It seems that Google's Pagespeed Insights, which is the basis for Core Web Vitals, by default considers content paint performance to be (a very arbitrary) 4x slower on mobile than on desktop.
https://developers.google.com/web/tools/chrome-devtools/spee...
That to me seems like a pretty naked attempt to get everyone to "just use AMP for mobile."
This "Google was just forcing publishers to fix their pages" meme desperately needs to die. Just consider all the extra standards crap they were pushing to introduce to perfect the deception. This was, as always, about owning the data.
Google doesn't need to own the data because Google is the world's gateway to the open web. They don't care who owns the data as long as they can crawl it.
And, as I read in the article, it looks like behind this move there is some current "Antitrust Pressure" plus publishers quite pissed off about losing both control and revenue themselves (as much as 39% less conversions, they says.).
Clearly AMP was way more in Google master plans than a poor web performance palliative.
[Edit]
This was the article linked to this story when I commented (now changed to some Google Dev docs):
https://themarkup.org/google-the-giant/2020/11/19/as-antitru...
It was open source, it was used by many others including competitors, it was optional and it didn't block access from anyone. Having an AMP version in no way "locked" you to any garden, AMP versions aren't even meant to be the canonical page anyways.
It may have had a lot of issues, but "walled garden" would not be one of them.
And it was as optional as publishers were almost forced to jump in it to stay relevant in News SEO.
This has silenced my AMP anger.
Edit: and switching from Google Search to DuckDuckGo. The addon only needs to take care of amp links i get sent via messenger/mail/etc
Another way would be to stick with Fennec v68 from fdroid, but that is not such a great idea, but at least the extensions do work.
1. https://blog.mozilla.org/addons/2020/09/29/expanded-extensio...
Also, it's available from the same developer as an addon for Chrome desktop; I suppose if you were just annoyed by the experience and don't care about de-Googling it works there too.
By default (i.e. until you click the bar) only showing the part of the URL that the site doesn’t control is really good for security and avoiding phishing. Now apple.paypal.secure.wendys.scamsite.info/payus/wwwcitibank just shows scamsite.info. This is a good thing! And doubly because URL paths and fragments have been made largely irrelevant to end users as app routes.
If you're going to rant about the death of the URL you should complain to HN as well that only shows the domain next to posts.
It seems harmless to show the full URL and not bait-and-switch the text upon interaction. I don't see it as the browser's job to protect users from their incompetence.
If you care at all about the health of the web, you should support browsers in protecting users from their own incompetence. Users don't blame themselves for phishing & fraud, when it runs rampant on a platform they simply switch and the end result is that native walled gardens that do protect users win over the web. This is exactly what happened with news on the mobile web; mobile news websites got so slow that Apple & Facebook started gaining a lot of ground with completely closed & proprietary news platforms (Apple News, Facebook Instant), until Google released AMP which both helped & pressured news orgs to get their shit together on the web.
I see these kinds of features sort of like padding on football players. It enables (encourages?) carelessness. Maybe the jury is still out on the net outcome, but I'm so far away from web development and the overall ecosystem as to have no opinion.
Of course it encourages carelessness, that's what users want: they want to care less about the tech they use.They simply want the benefits of tech, not the maintenance or burden of care. It's why many prefer an iPhone that makes all their software choices for them. It's why many trust an email provider to sort & filter their email instead of doing it themselves.
>"URL paths and fragments have been made largely irrelevant to end users as app routes"
No. Some SPA scenarios, for some end users != "largely irrelevant".
1. Chrome only showing the domain in the address bar
No, I doubt it'll make any difference here. Google's reasons for doing that are anti-fraud, not anything to do with AMP. And (I'll give up shouting this into the void one day, I swear) Apple did this with Safari years ago and no-one cared. It's outrage for outrage's sake.
2. The whole "web packaging" format issue
Google proposed a standard that let you "fake" the URL of a page (albeit with cryptographic signing to ensure it actually came from the right host). This might be dead in the water, yes. Mozilla and Apple already came out against it, and one the primary use cases was AMP. I doubt Google will un-implement it any time too soon, but I think it'll end up being a weird edge thing that very few people care about or use.
Which is a pity, because this could have been extended into a way to allow browsers to pin a specific version of a web app and warn the user if the server sent something different the next time they visited.
Having a TOFU security policy for web apps would have made them roughly comparable to native apps for certain threat models, especially considering that native apps often auto-update without giving the user a meaningful chance to check whether the update is backdoored.
For both web and native apps, though, it would be nice if there were independently run Binary Transparency logs published so that apps would only download an update whose source code had been publicly available (and reproducibly buildable) for some reasonable amount of time to allow an audit to occur.
There's not many issues that will gather enough reactions for Apple to ever respond. It would need to be something like removing the eggplant emoji to have anything sizeable from their perspective.
A decent amount of people cared about the move to hide most of the URL, just not enough for it to be a wide societal protestation wave.
The problem here is that google has an interest in hiding the URL other than anti-fraud. Apple isn't a major player in the internet market and they have a history of "dumbing down" features, or at least reducing their apparent complexity, to make the experience friendlier to the average user. So when Safari hid the full URL, that didn't raise many eyebrows.
Many of Google's moves in recent years, most notably AMP, appear to be part of a larger plan to make the internet require Google to function. Messing around with the URL visible in the address bar seems to fit in with this plan to reduce the average consumer's awareness of where exactly they are going on the internet.
It feels like a benevolent phishing scheme.
[1] https://amp.dev/documentation/components/amp-img/
That's why I find it rich that Google groups this under "page experience signals".
This is the ugly side of market dominance.
Google is entitled to have the opinion that AMP improves page experience.
But without competition, there is no loss of market share to give them the "this sucks" signal.
So the world has to wait for Google to degrade to the point that a new competitor starts to eat their market share.
What does the world have to wait for? The world is who chooses when a competitor starts to eat their market share.
DuckDuckGo isn't fabulous either but it is faster to retry in Google from DDG than the other way around so it is my default search engine now.
I wonder: why is it soooo hard - for both DDG and Google to just respect + or "" or the verbatim option in Googles case?
Because an empty set is far more valuable to me than a rich set of irrelevant results?
(Lately it seems I've been assigned to an experiment that has slightly better results and also shows me the context they think are relevant in the results page. That helps immensely, but of course I have no way to get that to stick :-/ )
PS: If anyone wants a billion dollar idea, recreate Google from before the DoubleClick acquisition: contextual ads, relevant results. Because that spot is empty, and given Moores law, the fact that the original PageRank patent has expired and more it should cost a tenth today vs then. Also Google seems to be collapsing under their own weight so there's less chance they manage to catch up.
PPS: I'd pay 10USD a month and accept contextual ads on top of that to get back something like old Google.
I think what might work better is an unweighted filter. A google style search engine filters out all tom yum recipes below a certain quality or all health sites below a certain credibility level. And then it tosses you a random one. Google tries to make a one size fits all engine, so it's doing very poorly in many of these. But Bing/DDG isn't faring much better. Stack Overflow and wikis are basically just search engine plugins
Google is probably just trying to reduce the likelihood of being broken up. AMP is widely adopted, so the damage of their preferential treatment has already been done.
Not publishers, who recognize it for what it is (a very, very thinly veiled attempt at stealing users, traffic and analytics, and Google building its own walled garden).
Not users, for whom navigation is broken, links and scrolling behave weirdly or break completely, and the address bar can't be trusted any more.
Competition. Replace it with something better using https://web.dev/vitals/
But isn't it also incredibly easy to switch off? At least I don't know any AMP-only website.
Parallel divisions will just result in the most recognizable names winning out. There is a simple and wrong idea that they can just break off Youtube, Android, and the search engine. Except that has as much logical coherence as saying Walmart should not be allowed to own their own pharmacies because they have too much dominance in groceries. None at all.
They only have angry old confused vested interests trying to protect their fiefdoms and no plan. It is a replay of the "Repeal and Replace!" bullcrap where after going on propagandizing for years prove they never had a goddamn plan and never tried. Mark my words - these techlash morons will be the dog who caught the car if they actually got what they think they wanted. Yet again like with the link tax where they complain about free advertising and then are outraged when nobody decides to link to them.
At this point I just wish all of those morons going on about antitrust would shut the fuck up because they know nothing but useless demagoguery and making things worse.
I wish apple would create a no-AMP option in Safari and just auto redirect to the normal page.
I hate AMP with a firey passion.
With Web Bundles the only thing that a cache can do is serve you an immutable asset that is signed by the original publisher exactly as if you had connected to the origin server.
If so I don’t like that either. They always feel a little bit broken compared to the real url.
If you are on a website in mobile safari, and you hit the share button, and select Copy. What gets copied into the clipboard is actually the Canonical url rather than the actual one you are at.
So for /amp/ pages, it ends up copying the non-amp url.
imo, it's a bad experience because sometimes canonicals and non canonicals aren't exactly the same. For example, if I'm looking at a specific filter or sort, then it might make sense for the canonical to be the non-sorted version. But if I'm sharing a specific url, I want it to share exactly what I'm looking at.
This is NOT ending AMP, and I don't know where anyone's getting that idea.
It will continue to rank by "page experience signals", and presumably AMP pages will continue to be the standard to beat.
In other words, if you can go ahead and replicate AMP's loading speed on your own (maybe self-hosted AMP?) then awesome.
I do hope sites do that, because sites loading instantly rather than taking 5 seconds is a huge win.
But in practice non-AMP news sites are still filled with bloat that makes them take 5 seconds to load.
AMP has massively increased the performance of many news publishers forced to switch to AMP. We'll see if publishers are now motivated to try to make their native sites equally fast. I'm not holding my breath, though. I assume they'll stick with AMP because they don't want to re-architect their sites, or don't have the technical bandwidth to do so.
But this way Google appears more neutral with "objective" page performance measures, rather than explicitly favoring any single standard.
EDIT: the URL and title have since been changed by mods to a Google blog post. The original article/submission I was referring to was: https://themarkup.org/google-the-giant/2020/11/19/as-antitru...
The article has a quote from the New York Times that suggests they will move away.
And the NYT quote doesn't suggest they'll move away at all. It just says it's "important" for Google not to make AMP a "requirement".
The first two components of your sentence are conditionals. Even IF they're both true, the very fact that they are uncertain, whereas before they were certain, makes the incentive MUCH MUCH smaller than before.
Any uncertainty at all will make internal discussions with technical leadership on maintaining AMP infrastructure weaker in companies like NYT. If they don't drop it immediately, there will at least be a ongoing internal dialogue about potentially doing so, brought up any time bugs need fixing in, or resources need reassigning to, AMP infra.
Where the idea might come from would be that many (particularly large) websites would have implemented AMP reluctantly, and currently incur extra maintenance cost for AMP as part of a deliberative ongoing cost-benefit analysis. This removes one of the major obvious benefits in that analysis (AMP pages may still benefit but it's harder to qualify), so there may be a stronger technical argument within companies to drop support.
> It will continue to rank by "page experience signals", and presumably AMP pages will continue to be the standard to beat.
This is a tricky prospect. A concrete "we will treat your AMP page preferentially" is a lot easier for tech managers to grok than "AMP stands a good chance of scoring well on metrics we prioritise, so may be preferred, if you believe you can't achieve these metrics without AMP".
On the flip side, it's completely conceivable that Google could continue to treat AMP pages preferentially while claiming it's due to proxy metrics, since their algorithms are not public.
Plus, you loose whatever customizations you turned on (dark mode) in web site settings. It is particularly bad on reddit.
If AMP becomes a lower bound for performance and non-AMP sites slim down their pages to compete, great. If sites that don't compete on performance die off as their search ranking decreases… also great!
The entire AMP design was driven by the need to make same origin serving of the content possible in a safe manner. That is where the performance gains come from, not from having lighter markup. And that is still just as true now as it was five years ago.
How can I possibly compete with AMP? I notice that even Reddit.com takes about 3 seconds to load. cnn.com, 2 seconds
Also, Reddit isn't exactly fast—it takes six seconds to load for me. Wikipedia and HN are better examples.
If you want the very lowest bound possible on latency for your internet connection, try loading https://cloudflare.com/cdn-cgi/trace. That should load in milliseconds. From there, you can do a little more profiling work on your site.
2. Use a CDN and size images appropriately
3. Avoid JS as much as possible, use it only for enhancements
How are you going to achieve equivalent or better performance than instant? The whole point of AMP is that it can be safely prerendered.
This is the entire point of AMP, yet 90% or more of the commenters on AMP articles don't understand it, which makes the comments completely useless.
Furthermore, I don’t see why AMP specifically is needed for what you describe — it seems to me that Google could do the exact same thing with normal webpages.
Why else do you think it loads from Google's cache?
> it seems to me that Google could do the exact same thing with normal webpages.
Nope. The publisher needs to opt in to having their content served by Google; and the ad metrics, analytics, and login need to be delayed until the user clicks on the result. That can't happen without cooperation from the page. Hence AMP. I don't even do web development, and the design of AMP is mind-numbingly obvious even to me.
I think most of us here know that 99% of publishers will not be able to match AMP in performance, so the incentives haven't changed.
I like the change though because this will force everyone to pay attention to website performance - something that's often an afterthought. We used to be able to use the internet over a dial up modem and it was OK, not great, but OK. Now that would be nearly impossible.
The thing is, the software I have on http://doomlaser.com is only for desktops. And I just noticed that a direct search for one of my apps, Cursorcerer, is now ranked #2 behind an entry for it on MacUpdate. A little frustrating. I guess I could redesign for mobile, but I would really rather not. Over 90% of my traffic is from desktops for a reason. Whatever happened to getting "the full Internet, not a mobile Internet" on your phone?
>But in practice non-AMP news sites are still filled with bloat that makes them take 5 seconds to load.
I wish people would stop saying "AMP means pages load instantly" as though it's an absolute. My experience with AMP pages has been the opposite; increased loading time, or pages that often don't even load at all. Come to find out in a HN thread from last year[0], AMP pages add an 8-second blank page delay if the user disable's Googles third-party js.
[0]https://news.ycombinator.com/item?id=19919881
Nothing here makes AMP less able to deliver high-performance pages.
Maybe this is platform dependent? I'm using iOS.
It just has to be better on android because somehow their own engineers have to use it day to day and ship it in good conscience at the same time, right?
The few things that bothered me the worst were, blank white pages while loading, to earn the 'one contentful paint'; fonts loading late; and it's difficult to share links, because they had the garbage urls because google was proxying. The fake address bar was the icing on the cake.
Page speed is the one metric you want SEO nuts optimizing.
Is there any way to tell.that the content of an HTML page isn't censored by the hosting provider used by the page owner, or by the browser vendor?
Doesn't seem AMP adds anything to that. Sure, you can't know, in theory, that the host (which may not be Google, Microsoft and others run AMP hosts) has censored it, but what does that actually change vs. HTML?
I guess it's in Google's interest to force AMP.
You can also take advantage of this with an app like Opener to navigate to the real page in the cases where the amp site is in such a broken state that you cannot get to the fake browser bar.
Yes, ideally all webdevs would make the non-AMP page bloat free. But as you know many a world pages are horrible and load tons of trackers.
Why is there is clear hate - AMP can be used for good and bad.
If it had just been about performance they wouldn’t have needed so much extra work and they wouldn’t have ignored the performance and reliability drawbacks to putting so much JavaScript into the critical path for page loads. Those 5+ second AMP mobile page views were completely avoidable if your goal was performance rather than control.
(ran, since firefox disabled the plugin on my phone)