The End of AMP | Hacker News Reader

The End of AMP

lafoo.com

352 pointsrmason154 comments

Previous discussion (208 comments).

https://news.ycombinator.com/item?id=26560498

I am so happy I have been able to avoid AMP my entire career and this ensures that I will probably will never have to bother with it in the future. This 'web inside the web' stuff is so antithetical to what the web is supposed to be, I hope Google will just pull the plug like they do with just about everything else that isn't the search engine or G-mail.

Or youtube, although that makes a lot of money so probably not anytime soon.

Or ads, 'cause why do they have search at all if not to search you ads?

Or maps, because geo searches are a source of revenue too :)

> Or youtube, although that makes a lot of money so probably not anytime soon

Does it? Alphabet quarterly reports don't list YouTube revenues and expenses separately, and i imagine something like YouTube costs a shit load of money in storage and bandwidth.

I'm sure compute isn't a small amount either. Video encoding is not cheap.

Yeah, they went as far as designing a custom chip for encoding.

Yeah, come to think of it I guess it’s not obvious is profitable and definitely not extremely profitable.

Technically (and by accident) what I originally said is still true— it makes a lot of money (but may cost a lot of money) and so it still looks good on quarterly reports (esp. if they bury the costs) so they’ll keep it around.

Plus, I wouldn’t be surprised if my Google search results were informed and improved by their data about my watch history.

Please don’t let your excitement take your focus off this part of the story:

> The good news gets even better; non-AMP pages make considerably more revenue per pageview than AMP pages. Initially, I assumed this was due to the nature of how ads load on AMP, however, recent Antitrust lawsuits have proposed that hindering ad competition was a feature and that all non-amp ad tags, such as my company, Ezoic’s, were delayed by 1 second to make them less effective. It is also alleged that Google let their own exchange win, even when someone else bid more!

Somewhere, in a parallel universe, people are going to jail and paying substantial fines for this behavior. In this universe, however, politicians don’t seem to think there is a moral hazard in the fact that those of us who work in this industry see what happened and think, “if you get big enough, it doesn’t matter if you get caught.”

There are active lawsuits going on by something like 49 out of 50 states. That's how damning evidence like this -- as well as Facebook offering to share unencrypted WhatsApp messages with Google (through the "backup to Drive" feature) -- came out to light.

Unfortunately a lot of HN readers are dismissing this or calling it partisan grandstanding, just because it's Texas that filed it.

> Facebook offering to share unencrypted WhatsApp messages with Google

where can I read more on why they allowed this?

I haven't found any statements about it from either company, but here's a decent article: https://www.tomshardware.com/news/whatsapp-google-drive-back...

According to that article, they also snoop on the content of any files stored in Drive.

To think, just a decade ago, this article seemed paranoid, to the point where even the author seemed to think it was an oversight in the Terms of Service:

https://techland.time.com/2012/04/26/will-google-drive-snoop...

Interestingly, they give (gave?) themselves redistribution and performance rights for Drive contents at the same time. How long until they start openly using those rights, I wonder?

Where does the article say they snoop?

Drive is unencrypted because it is indexed for personal search. What is the snooping part?

> redistribution and performance rights for Drive

This is completely standard and required so that they can provide syncing and sharing.

It's written in your whatsapp backup settings.

I meant leaks or new information exposing how/why FB intentionally made backups unencrypted when storing on Google Drive.

I am aware that Whatsapp does this and have avoided Whatsapp’s Google Drive backup overall.

It is partisan because Oracle told us so. They have bragged that they ran a decade long well financed effort to make all these actions against Google happen, and what we see now is the culmination of that.

I don't doubt they "won" in the end, but I'm baffled that they brag about "competing" like that.

Here's some evidence...

Oracle for years secretly lobbied states, Justice Department to sue Google - Bloomberg (2020-12-23)

https://www.bloomberg.com/news/articles/2020-12-23/oracle-fo...

I’m as much against AMP as the next YC user, but jailing seems really excessive.

This is clearly a separate discussion, but it’s interesting to consider why we’re always so eager to jail people for street crime, but it “seems really excessive” when it’s a white-collar crime.

We empathize with "people doing their job" perhaps a little too much? But part if is must also be just how inhumane jail sentences have become. Prisons are terrible places in the US, and sentences really disproportanate. With that framing, maybe it feels harsh to send somebody to prison for years for "just playing hardball a little too much".

I don't know, but the US penal system needs a revamp anyhow.

The question is still the same, why do we feel it’s fair when blue-collar criminals are treated inhumanly, but harsh when applied to white-collar criminals?

The same reasons lie behind why we also tend to ignore many external costs.

The personal violation you feel as a victim of a mugging or a burglary is more visceral and real, whether someone steals a dollar or ten thousand dollars. The perp takes something from your psyche, and we want to extract the same.

When some character fudges a spreadsheet or sets up a viscous cycle in business, it is more abstract. We’ve all done it to some degree (received an undeserved discount, more change than owed, etc), and the black and white “wrongness” varies.

Another consideration is that the punishment for white collar crime is very severe in terms of loss of agency and autonomy. Once you’re convicted of a felony, you’re kryptonite — you’re trading your comfy middle class existence for something riskier and worse. Typical suburban punter is losing most of your friends and probably your home and spouse. The kids lose their nice school or get stuck in a grubby apartment. White collar criminals have more to lose, so the punishment standard is different.

Let me unpack this. Do you think any given person is more likely to be considered “kryptonite” because they committed insider trading than because they did, say, a smash-and-grab? If not, your last paragraph is not justifying more lenient punishments for a certain class of crime, but for a certain class of socio-economic status. To put it in other words, only poor people should go to jail.

There's a nice line by Terry Prachett in his book "Going Postal". This is near the start of the story, as the protagonist, a conman, is being confronted with the effects of his cons.

> "I Worked It Out. You Have Killed Two Point Three Three Eight People," said the golem calmly.

> "I have never laid a finger on anyone in my life, Mr Pump. I may be–– all the things you know I am, but I am not a killer! I have never so much as drawn a sword!"

> "No, You Have Not. But You Have Stolen, Embezzled, Defrauded And Swindled Without Discrimination, Mr Lipvig. You Have Ruined Businesses And Destroyed Jobs. When Banks Fail, It Is Seldom Bankers Who Starve. Your Actions Have Taken Money From Those Who Had Little Enough To Begin With. In A Myriad Small Ways You Have Hastened The Deaths Of Many. You Do Not Know Them. You Did Not See Them Bleed. But You Snatched Bread From Their Mouths And Tore Clothes From Their Backs. For Sport, Mr Lipvig. For Sport. For The Joy Of The Game.”

White-collar crime has effects. Every time people are encouraged to clock out, then clean up their station, that's 15 minutes of wage theft. Heck, sometimes this becomes so common that the criminals work to make it legal. Every time that Amazon has their mandatory daily exit screening, that's 30-60 minutes of time that should be on the clock, but is completely unpaid, legal and SCOTUS-approved, and wildly unethical.

It seems likely that "street crime" mostly refers to physical violence or the threat of it, while "white-collar crime" refers to breaking some rules. While much more real damage might be done with rule-breaking, such activity doesn't provoke visceral fear, it's harder for most people to be emotionally invested in retribution.

"Street Crime" can also consist of selling single cigarettes: https://en.wikipedia.org/wiki/Killing_of_Eric_Garner

Or of hanging an air freshener on your rearview mirror: https://edition.cnn.com/2021/04/12/us/police-shooting-air-fr...

That's because HN's audience is composed of a large majority of white collars, and from a larger perspective, 99% of politicians, judges, lawyers, journalists, columnists, editorialists and commentators are white collars. It's just empathy (in this case, misguided empathy). We can identify more easily with a white collar criminal than a blue collar criminal.

No, it’s because a violent pickpocket puts my loved ones in more immediate, direct danger than some ads loading a little slower on a website.

How about a poor guy caught with a few grams of marijuana, only looking to pleasure himself?

On the other hand, Wall Street put many people's loved ones in clear danger, leaving them in the street, without jobs and without health insurance in 2008 and none of them went to jail.

I would guess that most people on HN would favor legalizing cannabis possession.

It is difficult to weigh actions that greatly harm one person against tiny annoyances to millions, but you could argue that the tiny annoyances to millions is a larger offense than violently robbing a single person on the street.

> you could argue that the tiny annoyances to millions is a larger offense than violently robbing a single person on the street.

Yeah, no, you really couldn't. It sounds like you've never been the victim of direct, physical violence, and the psychological effects it causes.

It's an easy distinction for me. Incarceration shouldn't be used for anything but violent offenders. The torturous impacts on the inmate's mental state ring so many cruelty bells that I can't accept it as anything other than the last resort for people that have to be kept out of the public for safety reasons.

Who exactly is “we” in this discussion?

Even a few weeks would likely be at least some disincentive; and anything less than that - like fines - runs the risk of being "costs of doing business" (e.g. most of the EU's laughably low fines, and US fines are very unpredictable, though at least somewhat higher).

If it's not going to be imprisonment, then the fines need to be much quicker, and much higher; enough to threaten a corparation's viability. For some perspective the largest fine/court settlement in the past few years is VW US's, which was 37 billion, and vs. a tech firm the 2012 FTC settlement with facebook at 5 billion. EU's largest antitrust fine ever (vs. google in 2017) was just 2.4 billion. Those kind of costs are so low that they're unlikely to be preemptively discouraging, especially given how rare they are; alphabet's revenue in 2020 was 180 billion, and it's market cap is a little less than 10 times that.

Of course absurd fines that are rarely imposed are more like a reverse lottery than anything else; not exactly ideal either, really.

> e.g. most of the EU's laughably low fines

Which is why in new legislation, fines are proportional - GDPR has fines at max 4% of global yearly revenue or 100 million euros, whichever is higher.

Which is at least a factor 10 too low under the assumptions that fines of that size are the result of a lengthy legal process and thus that such fines are not levied more than once every few years. 4% is peanuts for a massive tech corps that can afford to take such massive hits because their margins are so huge, and their fixed costs quite controllable. As long as the risk cannot be existential, this becomes an accounting question, and the inevitable rarity of the fine unfortunately means the optimal strategy is likely to skirt as close as possible to illegality and pay the fine and ask forgiveness rather than being too proactive - that way they only need to bother with behavior changes that really matter. 4% is just too little for large corporations, at least.

Then of course there's the fact that the GDPR - possibly due to a bit of regulatory capture - very strongly encourages venue shopping, and with resources like that, bending a small venue (like Ireland) to your will even slightly is conceivably feasible, and valuable. Personally, I'm not too impressed by the implementation of the GDPR, even if the fact that it's finally on the agenda is a good thing. It may even exacerbate misuse of personal data, because the burdens it imposes are much harder to bear for small parties than for large ones; and that's a competitive moat that encourages centralization of private data, and that encourages shifting the balance of power toward the corporation away from the individual (who says no to google's privacy policy?), and the combination of "willing" users, a lax privacy policy, and huge centralization encourages misuse. Proportionally your local sports-club is likely more impacted than Google, which is absurd.

> GDPR has fines at max 4% of global yearly revenue

In other words, laughably low.

> or 100 million euros

Unless you're a small business, in which case they'll happily bury you.

Then apply the "death penalty" to the company instead of people. A "death penalty" would also discourage gigantic companies, in general, as one out of hand division could kill the others.

And if you claim that the workers don't deserve it, well, then exactly who DOES deserve it?

Or "jail" time for the company... Suspend them from operating or getting new clients while they are in "jail". And maybe have them forfeit all profits earned during their "jail" time.

Really the possibilities of solution to some of these issues are endless, limited only by our creativity. We're just stuck as a society with what we already have and there is huge push back against anything out of the ordinary or creative or different when it comes to government and our governance models.

I think this is society's greatest threat... We are unwilling to take risks to solve problems.

laws need to have teeth or people will violate them

I believe that there's evidence that the best legal system is one where there's a high probability of lawbreaking being punished, though not necessarily harshly. In contrast, the situation prevailing in the US, where there's a high probability of not getting punished at all, and a small probability of getting the book thrown at you, is not a great way of administering justice.

How else are we supposed to deal with repeat offenders who flout the law?

The idea that white collar crimes don't damage people's lives is easily debunked: https://www.hsph.harvard.edu/news/press-releases/economic-do...

Ads being delayed seems like a win for consumers though.

Not at all. It makes the page slower and jumpier.

> jumpier

Not on AMP pages. AMP's design prevents user-visible reflow.

(Disclosure: I work on ads at Google, speaking only for myself.)

Normally, the way that late loading adds cause reflow / layout shift / jumpiness is that you don't know what size an ad slot is going to be until after the auction. You could run an auction just for a single size, but you'd have fewer bidders and make less money, so publishers generally choose to offer a bunch of sizes. On a regular page, this looks like an ad coming in and shifting all the other content on the page.

AMP had several design goals, one of which was that it should be impossible to create pages with certain types of bad user experience, including reflow. An AMP page requires that every element on the page be sized only by information directly visible in the HTML (including inline CSS, but no inline JS) which allows the browser to lay the whole page out immediately. When an ad loads off screen there can be a request to the AMP runtime to resize the slot, but the runtime will reject the request if the resize might cause user visible effects.

More details: https://blog.amp.dev/2020/04/16/cumulative-layout-shift-in-a...

Not anymore. Many ad networks have rolled out their 'CLS' solutions for the upcoming Google Web Vitals algo change, and ads don't jump about anymore on these networks.

Example ad networks that work well with delayed ads (that don't jump) include Mediavine and AdThrive.

Just use a pihole.

It’s literally changed my internet experience in my home.

I wanted to like pihole so much and ran one in my home for a while. Unfortunately I had to move back to per-device adblockers.

I had multiple instances of sites just not working because of the pihole was blocking something. This would result in me putting it in non-block mode for a little bit, rinse and repeat. I can't have that kind of hassle when non-technical people use the internet too. At least with per-device adblockers it's a single click to disable/enable the adblocker if a site is acting funny.

Basically, it wasn't any faster than just blocking ads on the device itself & added another thing on the network for me to manage.

Yeah I agree, so what I did was (and it took a bit of time), I whitelisted all the sites I hit that with and since then its been good.

It took about 2/3 hours of work though.

Right, but what Google did was to slow down only the non-AMP ads. If the goal was to build a better user experience, they should have slowed ALL the ads.

So basically you want people to go to jail for not blasting ads into your face immediately and actually loading web content first?

Seems like a horribly awful world to me - would you extend this to Ad blockers as well? And when was the last time you've seen how horrible mainstream web is, with its autoplaying video popups?

AMP isn't the best solution, but what you're proposing is just horrifying. If anything, we should standardize this behaviour in browsers and make sure the ads never interfere with loaded contents. Let's not destroy the web even more so ad peddlers get more money.

I downvoted you because you are deviating the conversation. I find it hard not to attribute hidden intentions behind what you posted.

The parent comment quoted saying they were loading their ads before ads from competitor networks.

Noone said anything about when the content was loaded. Google is the largest ad peddler.

You are obviously trying to shift the conversation away from how Google controls AMP and apparently used it to gain an advantage in the ad market (basically an antitrust issue) towards a generic conversation of ads are bad, but Google is good.

The parent was complaing about corporations being large enough to ignore antitrust legislation.

> You are obviously trying to shift the conversation away from how Google controls AMP and apparently used it to gain an advantage in the ad market (basically an antitrust issue) towards a generic conversation of ads are bad, but Google is good.

That is a really horrible accusation of something I haven't done. The OP is the one proposing jail-like punishment for actions that only lower the ad revenue and actually benefit the customer. It has nothing to do with defending Google (especially since I'm also talking about AdBlockers, which could easily fall into "coporation allows adblockers to lower revenue" interpretation).

Please don't look for hidden agendas where there aren't any.

> for actions that only lower the ad revenue and actually benefit the customer.

False, it only does that for their competition.

Are you spamming this comment using multiple accounts?

Do we want people to go to jail for selling us cheap bananas, or do we want them to go to jail for illegal dumping or using slave labor to pick the bananas?

> Somewhere, in a parallel universe, people are going to jail and paying substantial fines for this behavior.

My fear is that they are not going to jail or paying fines, they are being exiled... to our universe!

> all non-amp ad tags, such as my company, Ezoic’s, were delayed by 1 second to make them less effective

This is really not true. I was able to find the previous discussion on HN where someone linked to the Github issue about this design decision: https://github.com/ampproject/amphtml/issues/3133 . Note the root goal of this was actually to get non-AMP ads to run faster. This was all fully transparent, and was not some nefarious hidden functionality that Google tacked on in the hopes of tanking other ad networks.

Look, I hate AMP as much as your average HN poster, and I don't really disagree that Google is willing to make "technological improvements" as long as they hurt the other guy, and not Google. But I think the discussion gets better, not worse, when folks are more honest about the real reasoning behind why something is implemented the way it is.

Having AMP controlled and intermediated by Google was always a bad thing. It tainted what ultimately was a good effort, which is taking the enormous scope of web technologies we have now and winnowing them back down again for optimized publishing purposes.

And this article does what is often done in the AMP discussion and (disingenuously) demonstrates that you can make a non-AMP page that's smaller/faster than an AMP page. No one ever contested that. No one ever doubted it. But AMP isn't the alternative to your minimalist page, it's the alternative to the grotesquely overloaded media site where over time every single web technology gets smashed into a monstrous, massive pig of a page. Comparing the AMP pages of a news source and what they serve in their "full" web experience and it's absolute night and day.

I do wish we had a "web lite" mode that we could turn on, if only for certain domains.

AMP was about keeping news content on the web competitive with Facebook instant articles.

When Google was forced to cut a deal with News Corp last month (the Australian debacle) that would bring News content from the open web into their News app, AMP became much less important to them anyway.

All in all, yes, another loss for the open web, but I think the picture is more complicated then the "AMP is evil" meme.

My problems with AMP are:

1) Google clearly leveraging it to take over more of the web,

2) The scroll hijacking that made it feel worse-performing (plus just otherwise unpleasant) than middle-weight websites normally do,

3) That some of the sites I most commonly accidentally end up viewing through AMP are useless in their AMP version, so it just adds a hunt-and-click for no reason (Reddit is a big offender)

Google is one problem, one is a technical implementation problem, and the other is an ecosystem problem.

> I do wish we had a "web lite" mode that we could turn on, if only for certain domains.

There are some options for activating reader view even before loading a web site[0]. I imagine it would be a pretty light lift from there, to create an extension that keeps a domain list which automatically prepends the reader view URL prefix.

The magic prefix is

    about:reader?url=

[0] https://www.howtogeek.com/268116/is-there-a-way-to-force-ena...

Pity that Google’s amp “host page” breaks reader mode on iOS Safari. It’s why I use Duck Duck Go on my phone, so I can actually read the pages undistracted. I don’t like the DDG search results as much as Google, but Google ruining my web experience drove me away.

I wonder if amp intentionally breaks reader mode, to force viewing of Google served mobile ads.

> I do wish we had a "web lite" mode that we could turn on, if only for certain domains.

You get to choose what browser and plugins you use. That goes a long way.

The thing that gets me is that AMP isn't even necessarily that fast. It appears fast because the Google search results page preloads AMP articles, which it's able to do because they are served from the same google.com domain.

Overall it just feels like hack upon hidden hack with a sliver of standardisation on top.

> It tainted what ultimately was a good effort

Really? Scroll jacking is good effort?

Re-inventing a bunch of tags, just because, is a good effort?

It had a good claimed goal: make heavy sites load faster, particularly on low bandwidth/high latency connections.

The approach taken was, like everything Google does, heavy handed, tone deaf, and instantly defended by fanboys waiting for another chance to suck on the google teat.

Sounds like just the next calculated step.

1. Introduce $badThing with claims about page load performance.

2. tell sites they need to use $badThing to get into the “carousel”

3. Scale back the requirement of #2 just soon enough that you can argue against any claims of market manipulation via your monopoly.

4. (Continue to) Profit, because 80% of the sites that implemented it aren’t going to just rip it out now that it’s there.

i disagree with 4. Sites need maintenance and evolution. i don't know a single dev that won't eagerly drop something as impacting as amp as soon as it's an option.

Not every site has an active dev, devs are hella expensive.

Vast majority of owners will have a maintenance guy to call up if something stops working to patch it, until it gets too dysfunctional and needs a re-write.

However most of the time it's not the devs making the decision on what needs to be done for a site.

(a) that assumes the dev wants to. Plenty of developers still see Google as the all-knowing all-loving mother titty that can do no wrong.

Even for those who acknowledge Google may do shitty things, but are more interested in pulling their dick or what have you, rather than doing the work to make a great $thing, AMP gives them an easy out. No interpretation required, no need to worry about timings or what a "core web vital" is. Just follow this bouncing ball. Monkey see, monkey do.

(b) that assumes the dev is allowed to. Plenty of middle and upper management will always defer to "the industry" over any amount developer insight - particularly when the pitch is essentially "lets remove this thing so we can do some more work ourselves to achieve the same result".

(c) even if the dev wants to, and management OK it - maintenance is literally that: maintaining the status quo. Evolution, particularly on the scale of large sites, is a long term thing.

Big news sites are some of the worst offenders for junking up a page. Google could have "fixed" that by using the metrics discussed in the article to determine "carousel" results, from the start. But then they'd just have made a bunch of websites load faster, and wouldn't have added yet another tentacle around the web.

> that assumes the dev wants to. Plenty of developers still see Google as the all-knowing all-loving mother titty that can do no wrong.

I think I've only met one dev in the last 8 years that was that misguided, although I am somewhat limited in sociability so maybe just didn't realize it.

I'm not saying it's all, or even a lot of developers. But they exist.

Ok, plenty made me read it as a lot.

I'd say overall there are a pretty decent number of developers/tech people (not limited to just the web) who view Google as some giant benevolent thing.

I'd hope that web developers in particular would see more of the reality of Google's attacks on the open web. But then I've also seen an alarming number of "web developers" making claims like "I just wish Firefox and Safari would adopt Blink too, then we wouldn't have cross browser issues..." without a hint of irony, nor apparently knowledge of the last 15 years of web history.

But I'm also realistic. There are developers, both web- and other, who absolutely defend AMP, and Google's approach.

> Plenty of developers still see Google as the all-knowing all-loving mother titty that can do no wrong.

Preach!

Whenver i am trying to share a news article with a friend, and Google throws up the AMP page at me, I take the time to select the original article link and share that.

Websites deserve their own traffic, not Google's.

Not having to do this is a good reason to use another search engine than Google's as your default.

Is that very different from "Whenever I find a site hosted through CloudFlare, I look up the origin server's IP address and link to that. Webservers deserve traffic, not CDNs."?

Unlike a normal CDN, AMP does not serve the same content as the origin. You need to make a special, lite edition of the original web page which does not work on Safari or Firefox, and may not necessarily make your site faster. [1]

You cannot switch AMP cache to another vendor if you want Google rank your AMP site higher. [2] You also need a speical TLS certificate with the CanSignHttpExchanges extension that can only be get from DigiCert (for a price)[3] and not from Let's Encrypt [4].

[1] https://unlikekinds.com/article/google-amp-page-speed [2] https://stackoverflow.com/a/41683427 [3] https://web.dev/how-to-set-up-signed-http-exchanges/#step-1 [4] https://community.letsencrypt.org/t/cansignhttpexchanges-ext...

> You need to make a special, lite edition of the original web page which does not work on Safari or Firefox

AMP being a subset of HTML, that does not look accurate.

It is not just about the markup. Mozilla and Apple made it very clear they won't implement this (for good): https://www.chromestatus.com/feature/5745285984681984

Yes. Google are facing many lawsuits that suggest that AMP manipulated the ads platforms, in their favour.

Using a different CDN shouldn't significantly impact a site's income, but using AMP may have.

You can't do that, assuming that it was set up properly (i.e. they either set up CloudFlare immediately without disclosing the server IPs in the first place, or blocked all non-CloudFlare traffic on the firewall).

Same here. In some cases, AMP screws up the article formatting as well at least in my experience. I never share the Google/AMP link if I can help it.

You don't need to do this. Clicking the Share button will grab the canonical URL, which is the original source article.

Why does this seem to "confirm" that AMP is dead?

The article just "hopes" it will die, it doesn't point to any specific source that it will.

Quite the opposite, it says "Google will continue to support AMP" on the Google Page Experience update.

The key part of the article is this:

> I’ve had the pleasure of working with more than twenty thousand publishers in the five years since AMP’s launch, and I don’t believe I’ve ever heard a single reason that a publisher uses AMP other than to obtain this priority placement.

If there's no reason to use AMP other than to waste money, it's very likely it will be used less and less. And then it's just a matter of time before Google pulls the plug.

While that's certainly a good step on the way, I'll be holding my breath until it's official, It's the most annoyingly long-lived piece of shit I've ever had the displeasure to work with.

From Google's perspective that is an excellent reason to continue. What I don't understand is why all these parties instantly sign up to any abomination Google decides to foist on us without thinking through the consequences and realizing that they just handed Google even more control than they already had.

Because if they do no do it they fear some of their competitors will. And that is how Google makes everyone implemnt AMp despite it being bad for everyone who is not Google.

Because google already has enough control to destroy the business of these websites. Not getting into the carousel is a pretty serious dent in the business of many websites (because their competition will, and any website below the carousel is going to get a fraction of the clicks of one in it).

If google required a blood sacrifice to get on the first page of results, a huge number of websites would reply "animal or human?"

It is not the end of AMP.

In fact, I suspect it will only affect the small minority of sites that work better without AMP, the ones everyone who dislike AMP talk about. But for most mainstream news sites, I find AMP better than the typically bloated non-AMP version. And I bet Google will have the same opinion when ranking sites.

This is excellent news, but unfortunately, I don't think it will change the situation much. I wish it would, but most websites owners simply don't care about performance.

“Avoid unnecessary JavaScript, plugins and bloat, and make your site easy-to-use.”

If only web developers would follow this advice...

If there's ever free CPU cycles, you bet devs will squander it.

Exactly. If website haven’t been so bloated in the first place, no one would have believed the AMP propaganda

I've seen it happen more than once that web devs create a website that's very performant, only to be maimed and slowed down to a crawl by the marketing department's trackers and ads...

Hilarious advice considering Google's own non-adherence to it. Opening Gmail today is slower than when I was opening it on IE6 in 2004.

This is the thing that gets me.

The criticisms of AMP regarding how evil Google has been with it are all valid.

But to the average user that doesn't care (Remember, HN users are anything but average!), AMP is an absolute godsend. Normally when I view a news article on mobile, it's plagued with the scroll jumping around as things load, obnoxious controls, frequent freezing, and just overall a terrible experience.

AMP for the most part fixed all that. Pages load in only a second or two, and once they've loaded, they're smooth to scroll though and STAY smooth.

I don't do web development (Professionally anyways, I've got a couple small <3K LOC pet projects), so I don't understand why the current world is such total garbage. I just went to a few news sites, and the front pages sent me anywhere from 3.5 to 6 MB of JavaScript, and that's even with my ad blocker disabled. And then some of them started sending me even more once I started scrolling.

There's no reason for all this bloat. It does not take 3.5 MB of JavaScript to load content asynchronously and create some menus. And that's minified JavaScript.

So yeah...AMP is here to stay, and the users are grateful.

Thank god and I am honestly surprised it took this long. If it was ever purely about page speed (which we all know it wasn’t but play along) then given the choice between “analyzing a one-time metric per page” and “inventing a new parallel HTML, CSS, and JS standard and still analyzing pages for usage of this standard,” I think it seems pretty obvious the former is better in about every way. Site-specific functions like login/auth still work, there’s only one URL for a given page (not to mention the countless amp.google.com/blah links scattered about forever now), less caching required on Google’s part, and you put the power of the web to express yourself back in the hands of the sitemasters. I don’t know what possessed Google to think they knew page design better than publishers but AMP pages are seriously ugly. Good riddance!

The article title is miss leading, it’s someone arguing that it should be shelved

So long as google strongly influences page views it lives

> AMP also came with the unfortunate requirement that publishers let traffic sources such as Google cache their content and serve it from their domain, such as google.com.

At least with Chrome, one related performance advantage will remain, given AMP pages will remain preloaded from Google results pages, whereas non-AMP pages won't be.

Why is this related? On mobile, Google preloads AMP pages in the carousel that are in the Google cache, which will obviously give them a page-load time advantage. Why not do that for more sites? Avoiding third-parties (i.e., third-party v. the Google results page) from being able to detect when they appear in search results and getting data (like IP) from the user regardless of whether or not they click on the result.

Is this insurmountable? Can you outperform it? Sure, but it's inevitably going to be harder as a result.

This has changed too:

Signed exchanges (SXG) allow Google Search to prefetch your content while preserving the user’s privacy. In practice, this means that both AMP and non-AMP results shown on Google Search may prefetch a few key resources (such as HTML, JavaScript, CSS, images, or fonts) in a privacy-preserving manner, if the associated website supports SXG.

https://developers.google.com/search/docs/advanced/experienc...

(Disclosure: I work for Google, speaking only for myself)

Another entry for the Google Graveyard.

Kind of depressing when you think of all the money and engineering hours that went into trying to make websites work with AMP that could have been spent on just making their existing code less shitty...

Then again, probably most of them wouldn't have bothered anyway given what the rest of the mobile web landscape looks like.

Is there a process for off boarding AMP? We serve them now but want to stop. Can we just shut it off or will Google punish us?

"Who is gonna go first" seems to be an issue. I'm unable to produce a query that sticks any non-AMP result into the front of the carousel. So, everyone seems to be waiting for some notable publication to go first.

It would be such a blessing. Not even going into the obvious ethical and political implications of an AMP-powered web, it simply has broken UX on mobile. On iPhone it is possible to horizontally scroll the pages and then they are stuck in an unreadable state until reload. Seeing only "google.com" in the header is just the icing on the cake. Instead of one click to go to the source page you need to open a menu to go there. It makes me furious every time. IMHO Google really jumped the shark with AMP. Claiming it is for the users good and speed of the web is just hypocrisy. We would not need "accelerated" mobile pages if it was not for the garbage 10mb of trackers that are on every page because it helps them monetize with Google.

I'm flagging this article because of a very misleading title. This is definitely not the end of AMP.

The title is somewhat misleading, but if Google do not come up with a compelling reason to use the widely hated AMP, its future sure doesn't look rosy.

There hasn't been a compelling reason since day 1, that did not make a difference so far and since this is a way for google to play outsiders against each other it will take more than just a few stragglers to turn the tide. Only a mass coordinated boycott of AMP will do away with it.

The most frustrating issue with web development is fads. Every month there's a new must have technology that spreads with the most powerful marketing technique: FUD - fear, uncertainty, doubt.

"Did you hear about AMP? Yeah we switched to it last week. I heard Google would start weighting pages that didn't use it down. Wait, you haven't made the switch yet? Well... I don't know, we'll see what happens..."

If I smell FUD as a marketing technique, I'm out. The odds that you're trying to sell me something I don't need are astronomical -- and in four years, management will be driving me to the next big fad under FOMO.

Make no mistake: Google could have page metrics used in search ranking years ago. The reasons they introduce these now isn't because of the kindness of hearts or because they changed their minds.

Most likely (one, or some, or any of these reasons):

- there's a potential court battle regarding AMP, and Google wants to get ahead

- publishers are leaving the platform

- there are new advertising opportunities that bring as much money to Google, or more, than AMP (or are better than potentially losing money on courts and to publishers leaving)

Google has done that in the past, I worked through the whole PageSpeed introduction, measurements, dumb decisions made to products just for SEO sake, etc.

They had this before, they could have just expanded it, they preferred to introduce AMP instead. Worse, introduced AMP with the whole corporate backing of it, they knew what they were doing.

Dupe: https://news.ycombinator.com/item?id=26560498 (from 40 days ago)

The article says "The largest and most talked-about item in the update is Google announcing that sites with passing core web vitals will receive a ranking boost on mobile."

Can anyone please explain what "passing core web vitals" means? It links to very lengthy articles full of SEO jargon that I don't understand.

Use the Google Search Console to check for Core Web Vitals problems. Look for the "Experience" section on the left menu.

Unfortunately, that requires a Google account, which requires a valid phone number.

Is this different from PageSpeed Insights? If not, why doesn't PageSpeed Insights suffice? All I'm doing is creating websites that are as accessible and easy to use as possible. I shouldn't need to sign up and give my data to Google to do that.

HTML5 (with CSS and JS) and HTTP is all I should have to deal with, registering for search engines seems backwards to me.

A phyrric victory for a Web that is basically ChromeOS.

Ever heard of Web Bundles?

https://web.dev/web-bundles/

https://github.com/WICG/webpackage

How is Web Bundles different from .mht, which I "think", if I remember correctly is the same as MHTML. And Safari somehow support something else.....

This is a genuine question. It is such a simple feature, zipping an HTML file with all of its resources. Why do we need that many standards for it. Is it NIH syndrome or are there technical reasons in doing so?

Zipping an HTML file and the resources is not equivalent, since the browser behavior very much depends on HTTP headers etc, which MHTML doesn't capture. From the existing formats, that leaves pretty much just WARC, but that's also already kind of niche, has a a bunch of interoperability issues in practice, would probably need extending/reinterpreting to fit, and just like MHTML uses base64 for binary files, which is needlessly inefficient for the purpose here.

The more important things to specify are the browser behaviors when working with such files and the signature mechanisms, and those specifications you'd also need if you reused a format.

AFAIK bundles are designed to enable Google's CDN to serve up signed bundles from their servers and masquerade the bundles as coming from your site's domain. That's not something MHT files were designed for. (This actual feature is separate.)

It would be nice, though, if a site owner could use an offline private key to sign a web bundle, and then instruct browsers to only load their web app if the signature matches its public key.

The key would probably have to be delivered in-band, requiring a TOFU security policy, but that's not so bad.

Ideally the browser would also keep track of which version of the app it loaded last time, and give the user the option of fetching a newer version if one is available, to prevent malicious updates.

I’m rebuilding a leaderboard for news articles, switching from WebPageTest to Lighthouse.

The next step is to highlight when your need page hits the right benchmarks to qualify for “Top Stories”

My worry is that most sites won’t qualify and AMP continues

https://webperf.xyz

So I know nothing about AMP, but I've skimmed through the Google's documentation of signed exchanges, and it seemed like a good thing. I assume that it is the foundation of AMP. Isn't the point of signed exchanges that you, the publisher, don't have to invest in beefy servers and CDNs, and instead rely on third parties (for fee or free) to store the (signed) content? Of course AMP being a Google service the monopoly thing is hard to ignore.

> AMP created all kinds of problems, from analytics to ad serving to logins.

Oh no, it hinders competition with Google for surveillance and ad spam!

It's a pity you're being downvoted (perhaps because of the sarcastic tone of your final sentence), as I think you are right about Signed Exchanges. The technology could have some useful non-evil applications.

I wrote another comment in this discussion about this:

https://news.ycombinator.com/item?id=27024415

I was hoping I'd start seeing notable amounts of non-AMP stories in the carousel. Just tested it with some news type query strings, and it's not the case yet.

> This effectively means that publishers will no longer be forced to use AMP and can instead provide fast, rich experiences on their own domains.

Yeah I won't hold my breath.

Better late than never.

I have always detested AMP. But there was no other way around - if you wanted to appear in google results on top you had to use it.

What about the toxicity that’s AMP for email?

Google doesn't seem to have the same kind of leverage to push that idea around. Or at least, hasn't chosen to use it.

2 years ago, I openly wondered why Google gives AMP preferential treatment when they already have the ability to rank pages by speed.

AMP is about control, not speed: https://unlikekinds.com/article/google-amp-page-speed

As an aside, blocking AMP with uBlock origin and using this userscript removes it entirely, including the annoying delay:

https://gist.github.com/ilikenwf/0c732fc3e657f29a0b6f2bad4c4...

We should all throw a massive party to celebrate the death of AMP

It's not dead yet.

My favorite way to avoid AMP is to use DuckDuckGo. :)

If that were the solution, amp might not be so bad. But people reuse amp URLs - it's not just a search engine thing.

My least favorite way to avoid AMP is to use DuckDuckGo.

It seems like a pretty good search engine for desktop (research, tech, etc.), but for anything I want while on the road (community stuff, culture, restaurants) it's not good at all. But I suck it up and use DDG on my iPhone anyway, better that than risk losing my temper and throwing my phone into traffic.

Every year when WWDC rolls around, I hope that Apple is finally ready to announce WebExtensions for iOS. I'm pretty sure that building an AMP buster would be child's play. Keeping my fingers crossed… again… X,,

I'm here for this! You have my axe

Ironically, Google News was totally ruined by AMP.

what about AMP stories

I detest AMP. Every time I fall into it’s clutches I can’t help but feel like I’ve wandered into a trap.

> AMP also came with the unfortunate requirement that publishers let traffic sources such as Google cache their content and serve it from their domain, such as google.com.

I think this is backwards. Google offers to host your content on their domains, and if you want that it requires you to use certain markup, and only certain ads.

If your job is monetizing content, or if you want to raise some abstract "time spend on domain" metric, then I can understand how this can be unpopular. But if your goal is to have your content read by many people, have very fast load times, and spend a bit less on traffic, then it is beneficial.

> But if your goal is to have your content read by many people, have very fast load times, and spend a bit less on traffic, then it is beneficial.

But that was not the problem. There's nothing intrinsically wrong with Google offering AMP as a content hosting service subject to certain syntax and content limitations.

The problem was that, however fast your load times were in practice, all else being equal Google's search engine would penalise your ranking compared to using Google's AMP—all under the pretence of favouring "very fast load times".

I created a - very - opinionated Wordpress Theme that out of the box - after following the install instructions - fulfils the Web Core Vitals. this was the failed campaign for it https://www.kickstarter.com/projects/franzseo/fastest-possib...

we created it anyways. now i am looking for alpha testers. my email is in my profile.

oh yeah, it is AMP free as AMP is .... a horrible developer hostile, brittle and free internet hostile technology, not even mentioning the QA overhead it burdens on every project.

For those interested, my own website is based on it https://developers.google.com/speed/pagespeed/insights/?url=...

Reply on news.ycombinator.com