The reason to use AMP is SEO. If you want literally any benefit other than SEO, just improve the performance on your own mobile site. Then you won't have to jump through weird hoops and come up with entirely new "magic carpet" user flows, because you'll actually control your own content on your own site.
The talk attached to this Medium piece was given at "TechSEO Boost", but the article itself dances around ever using the term. So we're left to wonder why AirBnb is very worried about being able to link users to their native app, while they apparently don't care at all about whether users are using their web site. It's because of SEO.
> the Accelerated Mobile Pages (AMP) project, which has some powerful backers including Google
> It would be fair to say that Google has been a very vocal supporter of the initiative from day one.
These are really weird ways of describing a Google-owned, Google-pushed initiative.
In a large enough organization, improving site performance is hard, as you have to win political battles against your marketing / bizdev teams, denying them the analytics packages they want in the name of performance.
"We could do that, but then we'd lose the special AMP placement on Google SERPs" is a much more compelling argument, and preferable to starting a proxy war between the CMO and the CTO.
Google has done a hell of a job flexing their Search muscles to encourage the behavior they want, here. All the while somehow open-washing the project so that SEO-types make the ridiculous statements you quoted, implying that AMP isn't a Google owned-and-run project.
On an engineering level, I respect AMP. From an anti-competitive / social engineering perspective, it's absolutely terrifying.
Why is it absolutely terrifying or anticompetitive? Microsoft implements its own AMP cache and has decided to use similar presentation for its AMP results. Baidu and Yahoo! Japan, both dominant in their own markets, also implement AMP caches.
The AMP spec, as written, allows anyone to host an AMP cache and serve AMP pages. Google’s is the biggest and the one you reach through Google search so people tend to assume it’s the only one, but Cloudflare runs one as well, along with the ones mentioned in the parent comment.
I've done a previous comment on HN asking someone for a link to the spec for the AMP cache, but there is none. How can you claim it's a open technology (pointing at Google, not Itaxpica or any others on HN) when there is no way (except reverse-engineer) to figure out how the cache works?
AMP has a whole suite of technical limitations that are unrelated to the problem being solved. For example they limit pages to 50k (pre-gzip) of unlined CSS rather than allowing stylesheets to be linked. This is not how sites are generally built, and at least for me, meant rebuilding large parts of my build process.
Google engineers seem to have a hard time differentiating between a platform and a framework. PWAs and suffer from the exact same problem. They’re great if you build a greenfield product to their exact specs, but difficult to use otherwise.
> AMP has a whole suite of technical limitations that are unrelated to the problem being solved. For example they limit pages to 50k (pre-gzip) of unlined CSS rather than allowing stylesheets to be linked.
How is limiting size and server round-trips of CSS not related to the problem being solved by AMP?
Because becoming essentially Chrome only is the antithesis of their core business model of getting as many eyeballs as possible. Not particularly good for the web either.
From a technical standpoint it wouldn't surprise me if people start proposing binary formats that beat text/plain + gzip in size and speed.
The "SEO" benefit comes at a big price. You give up the reddest part of your page's heatmap to a 3rd party with often conflicting objectives.
There's many downsides, but, for example...Google carousel results inject JavaScript that allows the end customer to left or right swipe right to a competitor...if you choose the AMP path.
I'm not clear on why Google doesn't give control of the top banner to publishers.
My point is that you wouldn't know it from the article. It makes it sound like you should implement AMP for user experience and fast loading, despite then going on to detail the ways in which it compromises user experience.
When a big company like Airbnb jumps on a technology bandwagon people notice, and write-ups like this influence the choices other organisations make. They could at least just be upfront about the reasons why they're doing it.
The reason Airbnb is not likely to be upfront about why they're doing it, is it's very unlikely that Airbnb wants to admit they're probably being paid for it or getting some other form of special consideration for writing this.
The search benefit isn't even a long-term one: Either AMP will go away for some reason at some point, or, if the status quo is maintained, everyone will be forced to use AMP by Google's search monopoly, rendering it a null benefit to your rankings as everyone else transitions to it.
I think Airbnb should throw away everything they've done to date and hire all the Hacker News commenters to run their strategy instead.
> it's very unlikely that Airbnb wants to admit they're probably being paid for it or getting some other form of special consideration for writing this.
It doesn't look like anyone from Airbnb wrote this. But you're right, that's a good strategy - instead of the accommodation and experiences thing, they should implement AMP and get paid by Google for it. That's how business works: https://i.imgur.com/Vd6gI7X.jpg?1h
Congratulations Brian and Gil on implementing this!
> The search benefit isn't even a long-term one: Either AMP will go away for some reason at some point, or, if the status quo is maintained, everyone will be forced to use AMP by Google's search monopoly, rendering it a null benefit to your rankings as everyone else transitions to it.
Maybe in the long term, but meanwhile they get the benefits for being early adopters. If AMP fails, they will simply roll back.
Unless you're the 1% of websites you probably should implement AMP if you want fast loading (from search or news aggregators at least) because nothing is going to beat cached content loaded inline.
It's a terrible idea. AMP is a shameful attack on the open WWW.
AMP is slower than hand-optimized HTML.
AMP pages are often hosted on a 3rd party domain, even sending the wrong referrers to other sites.
AMP is also extremely slow on desktop browsers that have certain ad-blockers installed. If 3rd party scripts don't load, the user gets a blank page for 8 seconds. (Try it if you don't believe me.)
* But it's faster than HTML that people use in practice.
* Sure, but do you really care that users are using your physical servers or just that the content is out there? It's a trade that I think a lot of businesses would be happy to make. Who wouldn't want essentially free hosting complete with ad revenue?
* Do you think ad funded sites care about people who have a worse user experience with an ad blocker.
I don't disagree that AMP is bad for the open web, but it's not bad for the people publishing with it.
If someone is going to implement AMP templates they have the ability to implement better HTML in the templates instead.
It does matter that users are on your physical servers. It affects referrers and copied/pasted links, which affect brand recognition and SEO. Using AMP also puts a giant back button on your site that takes users back to the search engine results rather than deeper into your site. Many people do not want that.
I don't think that airbnb is an ad-funded site. Even if it were, do ad-funded sites really want a 3rd party dictating how they monetize their sites? If so, I doubt that they are thinking through it all the way and considering the long-term effects.
Using 1.6/0.8 3G speeds (with 300ms RTT) as the baseline, and specifically targeting a 3s SpeedIndex embarrasses most non-AMP experiences by virtue of the AMP cache underlying the SERPs. These are the terms with which Google's marketing AMP specifically with its own tools for marketers. (To wit: https://testmysite.thinkwithgoogle.com/intl/en-us)
You can get damned close--and even exceed it if you're a minimalist--by throwing every bleeding-edge performance optimization under the sun at it, but nothing is going to stop marketers throwing Adobe's render-blocking VisitorAPI in the <head> to offset many of those optimizations, either. (Seriously, and you can't async it, either. Marketers get angry and Adobe doesn't support it.)
The advantage of AMP dissipates rapidly as one tests in 4G and low-end cable speeds, however. At that point, hitting a 1s-2s SpeedIndex isn't terribly difficult if you're using front-end performance best practices, even with a framework bootstrapping itself to the DOM.
> AMP is also extremely slow on desktop browsers that have certain ad-blockers installed. If 3rd party scripts don't load, the user gets a blank page for 8 seconds. (Try it if you don't believe me.)
AMP stands for Accelerated Mobile Pages. Being extremely slow on desktop browsers with certain ad-blockers seems like an edge case of the least importance.
But the focus of the project is on the mobile experience. Complaining about the lack of optimization on the desktop experience, with an ad-blocker to boot, for a mobile specific project is an edge case, in my opinion.
Step 1. AirBnB, VRBO and others list inventory for free. Step 2. Google starts offering rentals straight from search and forgets you have a site. Remember Facebook News anyone?
Exactly this! Google Flights is already a thing and the hotel search works great too. Once they have enough people directly searching from Google, they can either charge Airbnb or hotels to send traffic or build its own Airbnb backend and roll it as Google product.
Google is doing this with everything. In the end they will answer almost every query directly, except when you are looking for a site. Even then, they could try to identify what do you want from that site (if you use chrome or the site has google analytics) and show you the info you are lloking for.
I'm still shocked that anyone treats AMP with anything other than contempt. Is it really that difficult to make your site faster? Blaming your crappy html/js/css on marketing demands doesn't pass the smell test for me. IMO too many engineers pass the buck instead of owning up to bad decisions and correcting them.
> I'm still shocked that anyone treats AMP with anything other than contempt.
The more people treat it with contempt, the higher its value becomes for the others due to its only advantage, getting a SEO boost from Google.
Also, marketing people (in some cases for good reason) already feel locked-in by and fully dependent on Google with >50% of their traffic coming from there.
I mean, embedding data into html looks bad, but is it really worse than loading a script separately? This is basically the "bundling" on the HTML layer, i.e., what HTTP/2 should offer.
For a website like Airbnb, 1.9M for total asset size doesn't seem bad...
I got 8MB for the entire site (homepage, from empty cache). I don't think it's that bad either. The webapp does also 'feel' bloated to me, but it might not be due to total transferred bytes, but rather just the JS heavy frontend.
Not to mention their website just plainly sucks, and was clearly an afterthought. Maybe nobody there could believe people wouldn’t want to install yet another app?
Its an all around bad design. Not really a UI problem, its more of a state / workflow problem.
Its levels upon levels. You keep having to drill 2 or 3 levels down, but then lose the context of the higher levels when you do. For instance, when you finally click on a property, this now opens a new tab. But now this property tab doesn't show a map...so if you are looking at 4 properties, you now need to keep the "main map" page open so you can manually cross-reference what properties are where.
Effectively its impossible to get anything done with having 3 or 4 tabs open all dedicated to Airbnb. Love the service and the places I wind up finding, but their website really needs a re-think.
Some time ago I was surprised that https://airbnb.com was serving me a revoked certificate. Then still when I checked a few months later.
I furrowed a brow and thought to myself, "this, from a company with lots of hype and accolades". But OK, certificate revocation is not very well implemented around the industry, so maybe just an honest mistake.
There's a lot of AMP hate on hacker news and I've never quite understood why. They talk about it destroying the open web, as if the existence of Facebook, Google, Amazon didn't already make a big dent in it. Really? AMP is the cross you're going to die on? I just don't get it.
There's enough animosity out there to spread around between AMP and other centralizing forces. Calling one out doesn't implicitly give the others a pass.
Facebook (especially Instant Articles and video rehosting[0]) is awful for very similar reasons.
What's pernicious about AMP & co. is seizure of control and immense abuse potential that comes with preventing visitors from hitting content creators' domains.
One thing that always I feel on discussions about AMP is that is something that is going to happen, and is not, is happening right now, regular users are already having better and great experiences on mobile.
That is the good side that I see of the project, Google was capable of do that in about 2 years. Pushing development best practices seems that was not enough, companies stills doing awful job on the performance side of their websites.
AMP was like force companies to do other version of their websites with limitations that prevent doing stupid development and design decisions like js/css/html bloat, repeat components and more.
You like that beautiful readable and organized Medium post?, is a good experience right?, well, it happens that the editor has a lot of limitations like only let you choose 3 text hierarchies, 3 image sizes, one font family, no colors, etc. For me AMP is something like that but in the development side instead on the visual design side.
The terrifying part is that Google has created a parallel version of the web who fill their needs, a stripped down version[1] that feels like an authoritarian power blocks the freedom that the "native" web always has, also happens that this parallel version has some advantages and people are loving it.
To those who think this could be a good idea: do some research on AMP, then reevaluate your opinion.
To my mind, AMP is a content reassociation tool, not a content distribution speedup tool. It succeeds in both, was designed as the former, and marketed as the latter. If Airbnb goes through with this folly, how exactly will they retain their business?
Google will be essentially filtering every booking that goes to Airbnb, controlling both the customer and the listings. I see most comments on HN are in agreement with this view, and those that aren't don't seem to meet the quality level normal to HN. Makes you wonder who exactly made this decision...
I made my personal website AMP compatible. Took me an hour in total (including research) and in the future if I don't want the google amp overhead I can remove a single line of code and my website will run fine.
The only thinking I'm depending on Google for is some post-loading. Otherwise AMP is just optimizations, most of them are just thinks I hadn't thought of before trying to make my site AMP compatible (things like only using hardware accelerated animations and such).
Consider yourself lucky. I'm guessing it's because of location. A quick search for "Google Amp" had this result near the top (an Amp version of a tech crunch article). If you open it on a computer it just redirects to techcrunch.com, but on a phone it should stay on the google domain and have a little top bar added (unless it redirects you past due to region?).
It really, really isn't. For one, your own JavaScript isn't allowed (though they just announced that they're working on that), for two, AMP is a not a subset of HTML, as it adds tons of Custom Elements in addition to the whitelist of standard HTML tags, and lastly, they're at least out on a small limb if it goes away, as you're not allowed to self-host the AMP framework and still be valid AMP.
There are many redeeming qualities to AMP, including its significant technical merit, but the subset claim is neither accurate nor the strongest argument.
>AMP is just a subset of HTML/javascript so they're not out on a limb if AMP disappers.
That's not all there is to it, though. There's a whole Polymer-esque AMP
JS framework that Google is pushing people to use [0]. While it does some great things, it's still absolutely an attempt at vendor lockin.
The reason to use AMP is SEO. If you want literally any benefit other than SEO, just improve the performance on your own mobile site. Then you won't have to jump through weird hoops and come up with entirely new "magic carpet" user flows, because you'll actually control your own content on your own site.
The talk attached to this Medium piece was given at "TechSEO Boost", but the article itself dances around ever using the term. So we're left to wonder why AirBnb is very worried about being able to link users to their native app, while they apparently don't care at all about whether users are using their web site. It's because of SEO.
> the Accelerated Mobile Pages (AMP) project, which has some powerful backers including Google
> It would be fair to say that Google has been a very vocal supporter of the initiative from day one.
These are really weird ways of describing a Google-owned, Google-pushed initiative.
"We could do that, but then we'd lose the special AMP placement on Google SERPs" is a much more compelling argument, and preferable to starting a proxy war between the CMO and the CTO.
Google has done a hell of a job flexing their Search muscles to encourage the behavior they want, here. All the while somehow open-washing the project so that SEO-types make the ridiculous statements you quoted, implying that AMP isn't a Google owned-and-run project.
On an engineering level, I respect AMP. From an anti-competitive / social engineering perspective, it's absolutely terrifying.
For more details, you can read the (very readable) AMP docs: https://www.ampproject.org/docs/guides/how_cached
Here’s the section on guidelines for implementing an AMP cache: https://github.com/ampproject/amphtml/blob/master/spec/amp-c...
Google engineers seem to have a hard time differentiating between a platform and a framework. PWAs and suffer from the exact same problem. They’re great if you build a greenfield product to their exact specs, but difficult to use otherwise.
How is limiting size and server round-trips of CSS not related to the problem being solved by AMP?
I take a look at the average mobile web page and thank Google that AMP brings sone restrictions.
From a technical standpoint it wouldn't surprise me if people start proposing binary formats that beat text/plain + gzip in size and speed.
Webassembly
That's exactly what they are doing with Instant Apps, they just try different strategies to control the web more at the same time.
Airbnb really doesn’t.
There's many downsides, but, for example...Google carousel results inject JavaScript that allows the end customer to left or right swipe right to a competitor...if you choose the AMP path.
I'm not clear on why Google doesn't give control of the top banner to publishers.
It's a growth hack to get better organic search traffic, with user experience performance improvements a secondary concern.
When a big company like Airbnb jumps on a technology bandwagon people notice, and write-ups like this influence the choices other organisations make. They could at least just be upfront about the reasons why they're doing it.
The search benefit isn't even a long-term one: Either AMP will go away for some reason at some point, or, if the status quo is maintained, everyone will be forced to use AMP by Google's search monopoly, rendering it a null benefit to your rankings as everyone else transitions to it.
> it's very unlikely that Airbnb wants to admit they're probably being paid for it or getting some other form of special consideration for writing this.
It doesn't look like anyone from Airbnb wrote this. But you're right, that's a good strategy - instead of the accommodation and experiences thing, they should implement AMP and get paid by Google for it. That's how business works: https://i.imgur.com/Vd6gI7X.jpg?1h
Congratulations Brian and Gil on implementing this!
Maybe in the long term, but meanwhile they get the benefits for being early adopters. If AMP fails, they will simply roll back.
AMP is slower than hand-optimized HTML.
AMP pages are often hosted on a 3rd party domain, even sending the wrong referrers to other sites.
AMP is also extremely slow on desktop browsers that have certain ad-blockers installed. If 3rd party scripts don't load, the user gets a blank page for 8 seconds. (Try it if you don't believe me.)
* Sure, but do you really care that users are using your physical servers or just that the content is out there? It's a trade that I think a lot of businesses would be happy to make. Who wouldn't want essentially free hosting complete with ad revenue?
* Do you think ad funded sites care about people who have a worse user experience with an ad blocker.
I don't disagree that AMP is bad for the open web, but it's not bad for the people publishing with it.
I heard the same thing said about Medium and Facebook News. Whatever happened to "fool me once, shame on you; fool me twice, shame on me".
It does matter that users are on your physical servers. It affects referrers and copied/pasted links, which affect brand recognition and SEO. Using AMP also puts a giant back button on your site that takes users back to the search engine results rather than deeper into your site. Many people do not want that.
I don't think that airbnb is an ad-funded site. Even if it were, do ad-funded sites really want a 3rd party dictating how they monetize their sites? If so, I doubt that they are thinking through it all the way and considering the long-term effects.
Not someone. A corporation. Think bigger companies, not tiny one person sites. Just because it's possible doesn't mean they can win the sprint cycle.
You can get damned close--and even exceed it if you're a minimalist--by throwing every bleeding-edge performance optimization under the sun at it, but nothing is going to stop marketers throwing Adobe's render-blocking VisitorAPI in the <head> to offset many of those optimizations, either. (Seriously, and you can't async it, either. Marketers get angry and Adobe doesn't support it.)
The advantage of AMP dissipates rapidly as one tests in 4G and low-end cable speeds, however. At that point, hitting a 1s-2s SpeedIndex isn't terribly difficult if you're using front-end performance best practices, even with a framework bootstrapping itself to the DOM.
Yep, because as that demographic becomes the majority they will then be forced to move from an ad funded model solely or perish.
AMP stands for Accelerated Mobile Pages. Being extremely slow on desktop browsers with certain ad-blockers seems like an edge case of the least importance.
Loading 3rd party JavaScript without restriction is extremely unsafe and not an edge case. It's basic Web-browsing safety.
But the focus of the project is on the mobile experience. Complaining about the lack of optimization on the desktop experience, with an ad-blocker to boot, for a mobile specific project is an edge case, in my opinion.
The more people treat it with contempt, the higher its value becomes for the others due to its only advantage, getting a SEO boost from Google.
Also, marketing people (in some cases for good reason) already feel locked-in by and fully dependent on Google with >50% of their traffic coming from there.
They shouldn't put AMP at the core of their strategy. It's enough to just make proper pages (the page referenced in the link is now much slimmer)
For a website like Airbnb, 1.9M for total asset size doesn't seem bad...
Seriously, use it on a connection with packet loss and it will corrupt your application state so hard you will have to switch browsers to resolve it.
Its levels upon levels. You keep having to drill 2 or 3 levels down, but then lose the context of the higher levels when you do. For instance, when you finally click on a property, this now opens a new tab. But now this property tab doesn't show a map...so if you are looking at 4 properties, you now need to keep the "main map" page open so you can manually cross-reference what properties are where.
Effectively its impossible to get anything done with having 3 or 4 tabs open all dedicated to Airbnb. Love the service and the places I wind up finding, but their website really needs a re-think.
I furrowed a brow and thought to myself, "this, from a company with lots of hype and accolades". But OK, certificate revocation is not very well implemented around the industry, so maybe just an honest mistake.
Facebook (especially Instant Articles and video rehosting[0]) is awful for very similar reasons.
What's pernicious about AMP & co. is seizure of control and immense abuse potential that comes with preventing visitors from hitting content creators' domains.
---
[0] http://splitsider.com/2018/02/how-facebook-is-killing-comedy...
One thing that always I feel on discussions about AMP is that is something that is going to happen, and is not, is happening right now, regular users are already having better and great experiences on mobile.
That is the good side that I see of the project, Google was capable of do that in about 2 years. Pushing development best practices seems that was not enough, companies stills doing awful job on the performance side of their websites.
AMP was like force companies to do other version of their websites with limitations that prevent doing stupid development and design decisions like js/css/html bloat, repeat components and more.
You like that beautiful readable and organized Medium post?, is a good experience right?, well, it happens that the editor has a lot of limitations like only let you choose 3 text hierarchies, 3 image sizes, one font family, no colors, etc. For me AMP is something like that but in the development side instead on the visual design side.
The terrifying part is that Google has created a parallel version of the web who fill their needs, a stripped down version[1] that feels like an authoritarian power blocks the freedom that the "native" web always has, also happens that this parallel version has some advantages and people are loving it.
I see AMP as a temporary patch of the web.
[1] https://ampbyexample.com
To my mind, AMP is a content reassociation tool, not a content distribution speedup tool. It succeeds in both, was designed as the former, and marketed as the latter. If Airbnb goes through with this folly, how exactly will they retain their business?
Google will be essentially filtering every booking that goes to Airbnb, controlling both the customer and the listings. I see most comments on HN are in agreement with this view, and those that aren't don't seem to meet the quality level normal to HN. Makes you wonder who exactly made this decision...
FUD.
Is it something I have to turn on or is it restricted to certain countries?
I browse from Norway.
https://www.google.com/amp/s/techcrunch.com/2018/02/13/amp-f...
Maybe they are afraid that EU will smack them down for anti-competitive behaviour.
The link posted above will show a prominent "techcrunch.com" at the top - but the URL is still Google. I am not redirected to their own site.
There are many redeeming qualities to AMP, including its significant technical merit, but the subset claim is neither accurate nor the strongest argument.
That's not all there is to it, though. There's a whole Polymer-esque AMP JS framework that Google is pushing people to use [0]. While it does some great things, it's still absolutely an attempt at vendor lockin.
[0] https://www.ampproject.org/docs/reference/components