The open solution to a faster mobile web would have been so easy: Just penalize large and slow web pages without defining a dedicated mobile specification. That's it. This wasn't done in the past, slow pages outperformed fast ones on the SERPs because of some weird Google voodoo ranking, heck sometimes even desktop sites outperformed responsive ones on smartphones. If they had just tweaked these odd ranking rules in way that speed and size got more impact on the overall ranking there wouldn't have been any reason for AMP—the market would have regulated itself.
I'm wondering who at Google is responsible for AMP. Who created AMP's random specs (no external CSS but external fonts files, preference for four selected font providers, no JS but their JS, probable ranking preference of Google cached AMP sites, etc.). Why did they decide on the spec themselves and not as a part of an industry group? Again why didn't they just tweaked their ranking algorithm and btw, they could have also made Android's Chrome faster, it's still significantly slower than iOS' Safari. I'd be happy if this person could comment on the abuse of power (Sundar Pichai?).
It's obvious they have a different view on this. You can see this first-hand on their pagespeed tool. Pagespeed ranks your pages mostly according to random features, irregardless of size and performance. Actual test you can perform: 1kb web page with no compression ranks lower than 1mb web page after compression: "because you should enable server-side compression".
Their mobile assessment tool is similarly a joke.
The sad thing is that I'm likely getting a lower ranking on my website with has 5kb vanilla uncompressed js (gasp, not even async!) compared to the glittered rating of a 5mb homepage which loads 5mb+ more excluding webfonts from external CDNs.
You can taste that this has spread into google monoculture by the performance of their own web services.
I gave up on Google Maps a long time ago. On my desktop or laptop, it loads in chunks. Like, the map loads, then the search bar loads, then the navigation (zoom tools, etc) loads. When I click and drag, many times I end up selecting the page instead of moving the map because it hasn't finished loading. I use Bing Maps on desktop and Apple Maps on mobile, only using Google Maps when I need to verify an address is the right place or find a company's hours, since their data is better than their competitors.
But man their performance is worse. Way worse.
A bunch of people, but Malte Ubl is one of the main people and has the best tech talks on why they picked the techniques and hacks they did.
I'd recommend watching that for how it works than asking @amphtml questions.
Node people: I did a node specific presentation at LNUG you might like:
I use AMP on CertSimple but appreciate that AMP favors 'speed achieved using specific techniques' rather than speed per se. There's the RealPolitik element where we as site owners have to do what Google say.
Why on god's name anybody would appreciate that?
A million times yes. As to why this didn't happen, I feel it would have been too restricted to "merely solving the problem", and not active enough on pushing Google's agenda to tighten its control on the web.
All other things being equal, option A being "efficient & neutral" (just penalize slow pages) is, as a business, inferior to option B "more convoluted but with potentially profitable side-effects" (AMP).
EDIT I should have better read the linked article before posting this, what I'm writing here is close to what the article "Lock In" section says.
They are a business, doing exactly what they are supposed to be doing.
It's up to us to find a way to kill google and take back our internet.
Edit, i love the downvotes for this. Google is legally required to act in their shareholders best interests. You all should understand fiduciary duty. They are building long term value for a massive (and growing) customer segment at the expense of a relatively small base of idealist tech users. Its morally reprehensible but still the correct decision in today's business climate.
What you're giving is an explanation, not an excuse. Nobody is incredulous at Google's motivations (money). We're just not satisfied with it.
Before you can solve a problem, you must identify it clearly. That's what we're doing here. Saying "yeah but it's only logical" isn't the point. Look:
> It's up to us to find a way to kill google and take back our internet.
You mean, like discussing why AMP is bad and why we shouldn't use it?
- we saw it happen once;
- we said it would happened again;
- we said how;
- we said why;
- and it happened again.
And apparently nobody gives a damn. People are still trying to get short term gains by adapting to Google's will and wrecking our industry, efficiently sawing the branch we are all seating on.
So yeah, we need to give the explanation. Again. And again. Until people hear it.
We are responsible of the situation.
Not Google. Before Google there was others like MS, and after they will be others. It's like a disease. You don't blame diseases, it's very useless, you explain, and you act.
Yes, your site will not make as much money. But for God sake, it's the same rational for green energy, criminal prevention and the like. You pay NOW because you understand that it's necessary on the long run.
MS had their cases in court where they had to pay a lot of money for violating laws. Maybe it's time for Google to go through the same phase.
Especially when its a closed system that Google insists serving to your customers from Google's private servers. No technologist or business person would/should ever stand for that. Yet we are forced to.
The discussion should really be, how do we dislodge Google from its current entrenched position.
There was such a thing as ethical and societal responsibility for businesses.
Thinking making profit by any means as long as they are nominally legal is acceptable for a company is just something some people believe. No natural law that says it has to be so -- we could easily (and have had) believe the opposite.
Since when? Standard oil? Or going back further, the Dutch East India Company?
Which goes back all the way to ancient Babylon in some form or another. There are countless instances in history of shady businessmen being taken to trial or punished by an angry mob.
(And the "Dutch East India Company" was accountable to its society -- and even nationalized at some point. It just exploited OTHER societies and countries).
Violating anti-trust laws?
There's a reason no US attorneys bring anti trust lawsuits against Google and that's because Google would very likely be able to show their choices benefitted consumers (eg. say saved 10 mins and 50MB data a day across 200 million consumers without charge)
Edit: Again with downvotes. Have I made an inaccurate statement?
Sometimes what you state just isn't popular for the crowd who reads it and/or moderates it. Doesn't mean you are right or wrong. Just take a deep breath and give it a rest.
EDIT: The guidelines say: "Please resist commenting about being downvoted. It never does any good, and it makes boring reading." https://news.ycombinator.com/newsguidelines.html hence, nope, I won't change that unless someone who works for HN directs me to.
We can both agree that one of us needs to change.
The guidelines are fine, the part you should change is automatically downvoting anything. Just jumping on a bandwagon makes for pretty boring reading too.
1) Not all violators will necessarily be convicted.
2) Even for those eventually convicted, there's always a period when they are violating, but not yet convicted.
As is being guilty and getting away with it. People do that all the time too.
And courthouses and judges don't have some monopoly into assessment of guilt. Just into its official identification and punishment.
(If the legal system had the monopoly, or some moral monopoly, of only it determining guilt, then how would we, as a society, judge the effectiveness or not of the legal system itself?)
yes, we (the united states where google is head quartered) as a society, have agreed to at least pretend to allow the legal system to define guilt and innocence. But this isn't a philosophical debate. This is a discussion of what can be done to fix the AMP dilemma.
This is an extremely vague requirement. Potentially killing their brand name among developers and potentially initiating a slow long decline into irrelevancy certainly counts as not acting in their shareholders' best interests. When google first started, their "don't be evil" mantra was one of the main reasons they succeeded, so not doing unethical shit is in their shareholders' best interests.
He's right in that they're a company, but not that the solution to this problem is to "kill Google."
Much easier and truer to the problem is to "kill" this expectation people have that the magic of good-natured leadership will create companies that dutifully remain bound to arbitrary constraints as "don't be evil" or "not doing unethical shit". It's just blissful daydreaming.
Much better would be to collectively gravitate around community alternatives. You have most of the tools in front of you (open-source, remember?), but you decide that Google's 'already won' and think of yourself as reliant upon them (thus complaining when they don't do things your way).
We do. You should understand it. http://www.nytimes.com/roomfordebate/2015/04/16/what-are-cor...
The interpretation of "fiduciary duty" / "they are a business" that you're implying is an urban legend, easily debunked by the fact that Google has no fiduciary duty to open a gas station in rural North Dakota as soon as they realize from Google Maps data that one would be profitable.
Using your logic if slavery were to become legal again, businesses should rush to use the free labor because "fiduciary duty".
This is why they teach business ethics.
Yes, companies do this today. Business ethics counts for nothing if it's not mandated. Compare Walmart's revenue to Whole Foods and you'll see why.
Anyway, its wrong to say business ethics account for nothing. Companies can and do self-regulate. e.g. http://www.hbs.edu/research/pdf/07-023.pdf
That was entirely clear from my comment, and you deliberately chose to ignore that. I will not apologize or make corrections.
But your claim seems to be that an entire technological sector spanning multiple countries employs slave labor? Its hard to take you seriously at this point.
My claim is that many companies use (what amounts to) slave labor and that there are financial incentives and economic structures designed around this. It's hard to compete without it, since your competitors are going to be using it. As one example, take entry level programmers on an H1B visa making less than their American counterparts, living with three other H1B roommates in a two-bedroom apartment, sharing one car, working 80 hours a week, knowing that if they get fired they get deported. Or Chinese workers unable to quit but working such horrid conditions that they kill themselves to get out of having to work. Or children in Africa making clothes to sell at Walmart in Illinois.
But again, that's just exactly what I said originally. I literally cannot make my opinion any more clear, so if you're still confused the only thing I can conclude is that you're deliberately misunderstanding in order to provoke an argument, something that is far too common on HN. I don't care if you take me seriously. I care as much about your opinion as you seem to care about mine.
Best interest != monetary gain.
Generally, Google has huge resources to spend at making their search and their advertising platform outperform others, and they haven't degenerated enough to allow a swift moving company to disrupt their market.
Oh, and if you try to fork AMP you fail the official validators and suddenly your site isn't AMP anymore. What is happening in this case is that open source is doing nothing to stop AMP from being proprietary because it's worthless unless you have the "blessed" version.
Take for example the most obvious change, to make the original URL visible in the banner. There's tens maybe hundreds of GitHub issues on this and it's still languishing. In the rare case anyone official replies it's some bs about how it will affect the user experience or "interfere with Google's cache" in some roundabout and unspecified way. They leave a few of these issues open so that whenever you start another they can close it as a duplicate. Otherwise their repo would be overrun with them.
Open source doesn't matter if someone holds the keys to the kingdom and "validates" your sources. A dark pattern indeed.
AMP in its current form is certainly evil. The only way that could be changed is if Google opens up the ability to cache the page by a different provider with different validators.
I smell antitrust
Well, thats basically how Android works.
And with the upcoming policy changes (in CDD and GCM) things will get even worse.
With AMP, you can modify the source and ... ... run your own search engine???
Android is software in the traditional sense. AMP is a protocol, even though it's phrased as software. The principles of free software apply to software, and protect your freedom when applied to software. They don't work right when applied to protocols.
It's open source, not anarchist source.
You see the other half of AMP is completely proprietary and runs on Google's servers. It doesn't matter that you can edit AMP because the server (Google's end) will refuse to run your edited version.
But if it's just a provider open sourcing the implementation of their communication protocol/operating system running on their devices, it's good for the community that they can look into it and learn; but just as well, it doesn't imply that any of the changes you're able to make actually do anything.
They are not even listening to the Linux team with their thousands of highly qualified developers and are essentially running their own fork of the kernel where everything has to be done The Google Way (TM).
What about the owners of the phones? I pay money — good money, quite a bit of it — for my phone. I own it. It should do exactly what I want, not what Google, Apple, Samsung, Sprint, Verizon or T-Mobile want.
If you're stuck on an AMP page in your mobile browser, you can click on the browser's "Request desktop site" option to load the full page.
PS: Not intending to be sarcastic.
PS2: I work for google, but not on something amp related.
Which leads to the question, why is google doing this? They, you, could easily promote AMP pages while not masking the real URL! The answer is simple, profits over what's best for users.
Edit: I just did a search to find the CNN interview I mentioned, copied the URL to share here, and look it's a google URL. WHY? https://www.google.com/amp/s/amp.businessinsider.com/trump-v...
I hope this AMP experiment is soon over. As someone already said here, the web is confusing as it is. Google used to want to make things clearer, better. AMP is not. It's fixing symptoms instead of causes.
Well, that explains why they wanted Chrome to hide URLs.
Different teams inside a business can have coordinated long-term goals.
Regardless, their master plan clearly shows "Redesign HTTP", "Browser", and "Next generation web (Fish Pod)" as precursors that lead to "Private Internet", "Walled Garden", "need more cache", and "NEWSBANK".
I run a WordPress blog. From the documentations of Google and WordPress AMP plugins, I have always thought that the user will be directed to "originalurl/amp" on my domain when they click on the Google link on their phone. Now that I see it's a Google domain, this is really weird.
Perhaps to allow the content to be served from a CDN (over HTTPS), without requiring the site to CNAME over their domain to Google.
If webmasters are willing to CNAME over their domain to a caching proxy, then a less intrusive design is possible, such as the one recently announced by Cloudflare.
Money. Amp vs adblockers
Do you really want to go down the rabit hole of a browser displaying a different url than what you are viewing?
Of course you wouldn't mind, since, as you've said elsewhere that you work at Google.
"It is difficult to get a man to understand something, when his salary depends upon his not understanding it!" — Upton Sinclair
Such really simple syndication technology could even have a catch TLA. I can't really think of one, but I'm sure there is something.
That said, i think people disable JS in general for this reason.
Why doesn't Google drop the URL masking while keeping all the real benefits of AMP? The answer, profits over people. I say this as a long time fan of Google, but this AMP stuff is where the "don't be evil" tagline turned into a joke for me.
AMP itself allows for fast loading which is true, but to support opening pages in "zero loading time" you still need pre-loading since fast loading still means non-zero loading time. To get pre-loading working in HTML you need to 1) load the iframe of the page in the background, then 2) when the link is clicked, use history.pushState() in order to manipulate the URL to appear you have moved to another page. Unfortunately for (2) the Web History APIs need* both URLs to be on the same domain so Google has an excuse that they need to change the target URL to be under Google for this to work.
Also Google wants to present the scrolling UI that allows you to browse to other articles in the search results (potentially from other domains). This can't be done if you have completely switched to the original website.
See this link for more discussion https://github.com/ampproject/amphtml/issues/6210
Personally I find this misguided at best, an excuse at worst. A fully compliant minimalistic AMP website (or even just a simple site that doesn't use AMP but follow best practices) will load incredibly quickly even without pre-loading. It's disrupting a lot of security and UX best practices (by replacing the address bar with "google.com" domain) just to achieve a very minor improvement. Furthermore they control Chrome. It's not too hard for them to push forward a web preloading standard and start implementing in Chrome.
Edit: Also for the scrolling UI it seems like another way for Google to push more engagement to their site rather than the target URL. It kind of runs counter to the open web nature IMO because there are innovations that helps previewing links already (e.g. iPhone 6s/7 3D touch) and it's the kind of behavior better implemented in the browser itself.
Apple: These article webpages are annoying - let's add a client side button to make it readable.
Google: These article webpages are annoying - let's force everyone to grant us a royalty-free license to their whole website.
 The real problem might be that Google results are "getting bad" - a thing people bring up all the time but which can only be measured individually.
So people tap the first 3 results and don't get the info they were looking for. On top of that it's slow af so people are wasting time. Now Google serves up faster pages so they'll tap through 4 or 5 results.
So why not fix the real problem instead?
 Yeah it's optional but only to a point. First the spammiest blogs implement it and they get a free pass to partially skip the line. Then some bigger sites implement it and even things out. People here are rightfully afraid that within 12 months you can't even hit page 2 without AMP.
Check the link rel canonical tag. Obviously not ideal but nowhere near "impossible"
Your average user (who doesn't necessarily know anything about software) doesn't really care about 1, and probably doesn't care about 2. They care about speed of the page, data consumption of the page, and maybe the look-and-feel of the page (bloated vs sane).
When I say HN user, i mean technically informed and most often skeptic people.
The distinction matters.
If I search for weather, lyrics, or other common sites, the top results are always slow, bloated, JS-filled nightmares. Yet the sites that are quick and small are never to be found.
Maybe they do and they need to weight speed higher. I don't know, but I know that I don't ever seem to get fast loading results.
This feels too much like creating a problem that you want to be able to step in and solve.
So you have to load the AMP page and then a potentially bloated normal page. That's AMP doing the exact opposite of it's intended (or at least stated) purpose. Noscript, flashblock and adblock have done far more for page speeds than than AMP ever will.
If I didn't know this was for mobile browsers I would think it was a gift to blind users who do not use GUIs.
amphtml links also provide an easy way around paywalls, most of the time.
I have no idea what is the true purpose of these amphtml links. It sounds like it's some advertising nonsense.
As a text-only browser user I see no ads and experience no page load delays, but I do see how overstuffed web pages have become.
In this regard, amphtml is a breath of fresh air.
If it was any good, it would be showing the desktop scaling/layout/whatever instead of the mobile when 'show desktop version' is on.
It's also not always about being on a responsive sites, a lot of places still do a m.website.com redirect when visiting the main site from a mobile device, but with no backwards redirect when visiting the mobile site (which is almost certainly non-responsive, otherwise there'd be essentially no need for the m.* site) as a desktop client - so you hit 'show desktop version', the m.* page refreshes and you just get the mobile site with funny scaling.
Request mobile site is really only applicable to the small amount of websites that serve a m.domain.com.
Now, they could have an option to "request desktop SIZE" that would result in sites that are responsively built rendering in "desktop size" mode.
The bar has been set. Can you find a way to both resolve personal frustration and provide a better experience? Maybe a browser plugin?
After which I was invited to meet Google AMP team and to express my concerns, you can read my Q&A here:
TLDR; A lot of concerns are getting addressed
1. Minor, but the bar at the top is now scrollable on all devices, including (finally) iOS:
https://www.alexkras.com/amp-toolbar-is-now-scrollable-on-sa..., it was not when I first wrote the article, so it's a good sign.
2. It is my understanding that the team is actively working on
a way to "fix" the link issue, and give an easy way to get to original article, although it remains to be seen how they will approach it.
3. You can opt out from AMP cache on the web site end, but it really defeats the purpose. Read more here: https://www.alexkras.com/i-had-lunch-with-google-amp-team/
4. Most importantly, looks like there is even internal pressure to give people an option to Turn Off AMP on the search engine side, if they don't like it. See this, for example: https://twitter.com/slightlylate/status/820344221450125312
@cramforce is THE tech lead on AMP and @slightlylate is also a big shot at Google on Chrome Team.
Personally, I have mixed feelings about AMP, on one side I really like the speed, on the other I hate how it breaks the Web as we know it.
The downside of AMP seems to me so comically larger than the upside that this should be much more a black-or-white issue than it currently is. Good that they are moving to fixing it, but very bad that this thing exists as it is right now in the first place.
I recognize that they are solving a real problem for users and I don't want to throw out a baby with the bath water.
At the same time I am VERY concerned with caching aspect of AMP. I am holding back on my judgment until they have some time to address some of the concerns, to see if I feel better about it.
But may be people on this thread are right, and a more aggressive stance is warranted...
I hate those bars that reappear when I scroll up, I scroll up to reread a paragraph that I realized I just skimmed or I didn't quite comprehend and instead I get a big blank space with a link to somewhere I do not want to go. They fundamentally break expected scrolling behaviour.
AMP is cancer and should be roundly decried. It's disgusting you can't even turn it off in google search.
They're breaking the web and breaking everyone's web pages and content makers are meekly supporting them instead of kicking up a stink.
15-years-ago-me is confused.
Edit: Ah, yes, your "make the header clickable" fixes this.
The majority of the articles, excluding those from large name corps, are simple loading pages. Using primarily the basic HTML and inline CSS. You do have the occasional blog page with some fancy touches, but the majority of the links that I click on while browsing HN has me questioning why, on sub par satellite internet, that HN and a majority of it's submitted posts work when Facebook, Google, Bing, etc.. does not.
I feel it can.
Everything just feels so bloated.
In many respects I like gardens, I like curated contents.
But I like the web more.
The web needs diversity. We need more search engines for instance.
On a serious note. Escaping Google on the internet is impossible. Can't be done without inconvenience.
It's bad enough that I've had to switch to using Bing on mobile, despite the worse results, and I'm actually genuinely fearful for the first time about the openness of the web.
Time to give DuckDuckGo another try I guess...
It seems there is no alternative for those who care about their rank and clicks.
I wrote a small chrome extension that always forwards my page to the equivalent AMP page (if one exists) and the experience of reading the news is so much better.
AMP pages off mobile are really really amazing. Compare Non-AMP vs AMP
Wouldn't that effectively mean that in a search query followed by an AMP site visit only Google has the opportunity to show targeted ads ?
edit: This is with an ad blocker on though.
Here's a much better example.
AMP - https://www.washingtonpost.com/amphtml/news/post-nation/wp/2...
No-AMP - https://www.washingtonpost.com/news/post-nation/wp/2017/01/1...
As far as I can tell, your method of viewing a nice clean page depends on your trick not becoming popular!
Startpage.com delivers the same results google.com does. I've never hit AMP pages, so I assume they neither they redirect to AMP pages, similar to duckduckgo's !g.
On the normal implementation, clicking links on mobile from Google search, this is mostly not a problem because you're unlikely to land on some random guy's WordPress from a serp. But if you're triggering a lookup of AMP wherever it's available, you end up on a lot of broken sites clicking links from reddit and HN and stack overflow sources.
I've found a better compromise is an extension that doesn't load amp by default, but just puts a button in your toolbar that you can request an amp version with.
This probably would speed up loading times as it would only have to display/load certain content.
I love 'reading mode' on firefox and safari (mobile). Unsure what other browsers support it but they all should.
Even with ads the original is an easier read on desktop....
Is it possible to set it as default on iOS/Android somehow? AMP really frustratingly breaks link sharing, and I'd like to totally avoid it.
!g search query
Of course, this also works if you type this query in at duckduckgo.com
Edit to add: I guess it is a matter of time before amp pages also appear on the encrypted subdomain; another incentive to switch to DDG now.
Many years ago, Google Search existed only on http:// and schools used filters to block searches that they didn't like. Then Google shipped https:// for search on www.google.com, upset the schools because they could no longer block just some searches, then moved encrypted search to encrypted.google.com so that the entire domain could be blocked: https://cloud.googleblog.com/2010/06/an-update-on-encrypted-...
>1. Clicking on an ad:
> - https://google.com : Google will take you to an HTTP redirection page where they'd append your search query to the referrer information.
> - https://encrypted.google.com : If the advertiser uses HTTP, Google will not let the advertiser know about your query. If the advertiser uses HTTPS, they will receive the referrer information normally (including your search query).
>2. Clicking on a normal search result:
> - https://google.com : If the website uses HTTP, Google will take you to an HTTP redirection page and will not append your search query to the referrer information. They'll only tell the website that you're coming from Google. If it uses HTTPS, it will receive referrer information normally.
> - https://encrypted.google.com : If the website you click in the results uses HTTP, it will have no idea where you're coming from or what your search query is. If it uses HTTPS, it will receive referrer information normally.
You used to have to use that domain to run searches over HTTPS. Google's enabled that for all users now, so it's kind of obsolete -- but apparently it still behaves a little bit differently?
It also results in lower quality news appearing at the top of searches in cases where they have implemented AMP and the better sources haven't.
(with all five visible comments, and since the full comments are almost certainly what I was searching for, I end up having to click through to the full page anyway every time)
The # 1 reason why AMP bothers me is when I want to share a link with someone, I don't want to send an AMP link.
AMP doesn't work on firefox.
In all fairness, lyric sites were terrible. Slow, riddled with ads, and sometimes incorrect. While Google's lyrics are a great service to the user, they're at the top of search and displayed inline. There's no reason to go to lyric sites anymore.
My biggest fear is Structured Data and AMP. With Structure Data, you volunteer your most valuable data in a format Google can easily consume and adapt to its own needs - all so you can get better page rankings. When Google introduces its own service in the same market - just like lyrics - you're effectively cut off from your audience. And with AMP, you don't have to wait for Google to siphon your traffic - you're volunteering.
Edit: on top of that, the video that got taken down was an iphone video of people dancing to a song playing over a stereo - kudos to how accurate their detection is, but goddamn!
In other words, Google seems like a very powerful gatekeeper of content that ends up consuming the content providers. Smacks of an abuse of their position.
I couldn't imagine dealing with supporting two deployments.
give a try for a day or so. Once I did, I can't go back to keeping it on because the web seems so slow by comparison.
This wouldn't be that big a deal if Google didn't emphasize the rank of AMP pages. There aren't a lot of alternatives out there to search, and Google dominates the market in much of the world.
Let's assume this got much worse, and evidence came out that private political interests worked with Google to interfere with informed consent by reprioritizing what some called "fake news", when really what was being called fake news was actually anti-government activists pointing out collusion between tech companies and the government. Clearly, Google would be an agent to destroying democracy.
I imagine detecting that a visitor came from Google, and showing them an interstitial informing the user they have come from a state-sponsored search engine, and letting them know what that means and the alternatives to Google.
I think it's actually pretty trivial to build something like this with recipes for the popular CDNs and servers out there. Even a JS snippet would be totally fine.
Google is pursuing AMP not because it wants to promote a better experience on the web, but because Google wants to be the provider whose technology and own practices perform best on SRPs. They are acting like capitalist pigs, and we should coordinate a protest against Google to let them know they can't just walk all over us with no consequences.
January 20th seems like an ideal day.
Edit: I couldn't reproduce this with Chromium on Ubuntu 16.10. I might set up an Arch Linux box to see if that makes a difference.
Maybe you use some strange fonts?
Mind you, this just proves that the header is superfluous. Existing technology called 'the back button' exists to take the user back to where they came from, and it's up to Google to decide how their results page behaves when this happens.
I really think AMP can benefit a lot from PR stand point by letting techies opt out from it, but leaving it on for "regular" users. See https://twitter.com/slightlylate/status/820344221450125312
The problem with that is that the UX is inconsistent between AMP results and web results. AMP results pretend they're this fake tab that can be "closed" and you return to Google, while web results take you away to the URL like you'd expect, which means navigation is the responsibility of the browser, and not of Google Search.
Frankly, this all would make more sense in the viewport-wrapping context of Google Chrome or Google News or the Google App rather than a behavior of Google Search, but that's not the limit of their ambition.
Or just, y'know, dress up the results page a bit more, and be honest about it. Brand the AMP subsection 'Read Instantly with AMP' and have it clearly and visibly wrap the results. This gains them a bunch of mindshare.
Or change the [x] to a 'back arrow', like their own splash page shows .
Google is search, everything else is secondary. 20 people in a room testing this should have been all the red flags they needed to fix it before launch.
The first time I encountered an AMP page I figured clicking the x would load the actual page. I was wrong.