AMP's innovation isn't a way to make pages fast. AMP is a way to sell other stakeholders on implementing technologies that make your website fast. All the stuff AMP does is stuff you could do yourself without the extra request to amp.js and the extra work to amp-ify your pages.

But imagine you've got an advertising department that wants three different ad networks, a couple different managers that want to see stats from a couple different analytics platforms, and and the designer wants to load a font from fontsquirrel and another one from typekit and another one from google web fonts, and as a developer who wants to keep the site fast you have to fight them every single time they want to add something else that slows your site down. Having the same fight every time, with everybody else saying "oh, it's just one request. and this one is really critical" it's hard to keep fighting that fight.

It's a lot easier to say "i can't do that, it doesn't work in AMP". If you can find a better way to convince large organizations that page load speed is a valuable metric, and more important that whatever other resource they want to load today, I'd love to hear it. But from what i've seen, AMP is the only thing that's had any success in tackling this problem.

This. AMP was a blessing for me honestly. I can now maintain a version of our new site that isn't bogged down with tracking and flavor-of-the-month JS feature garbage.

I've been fighting against adding additional tracking forever, but constantly get railroaded by marketing because "they're the ones that know how to make us profitable."

Fundamentally I hate what it means for the internet, but I finally have a little power to say "no we can't do that."

It is astonishing how hard it can be to internally sell any kind of web quality features to management in both for profit and non-profit organizations.

There is also a real herd effect. Many people will do whatever Matt Cutts tells them because they think it will be good for their SEO. Yeah right. Some of the people who are good at SEO either went to work for huge brands or quasi-competitors of Google (like about.com) that might have some ability to bring Google to anti-trust court; most of the others switched to paid advertising once they figured out that Google won't let you win at SEO.

Certainly people who write for Spamium (aka Tedium) are the ones who try herd-following methods of getting traffic and they tend to be impressed when they get 100 hits on their blog.

One technique I've seen work well for getting bosses to care is showing them how much slower their sites are than their competitors - nobody wants to be worse than their competition.

In SpeedCurve, for just around $8/mo per page/site you can set up daily synthetic checks for the business and its competitors, covering multiple profiles: mobile/3g, tablet/4g, desktop/fibre.

You can use both the metrics (Start Render, SpeedIndex, Hero Paint) [1] and the filmstrip videos to literally show them how they compare side-by-side [2] – this is super-powerful as it's so visual.

Disclaimer: I don't work for SpeedCurve, just a fan of the tool.

[1]: https://speedcurve.com/features/synthetic/#benchmark [2]: https://speedcurve.com/demo/video/?tests=180102_1T_d12fa882f...

> schwinn: were the filmstrips enough to convince them, or did your bosses want to see numbers/stats like how many were dropping off your site for each x-seconds of delay?

There are some great stats on bounce rate / abandonment on WPOStats: https://wpostats.com/

This is my favourite:

> 53% of visits to mobile sites are abandoned after 3 seconds according to research from Google's DoubleClick.

One thing to note, to those stakeholders who are aware of web traffic stats, is that if a site uses client-side analytics (e.g. Google Analytics) and it hasn't loaded the analytics script by the time the user abandons the site, they won't be tracked, so the bounces won't be affected – it'll be like they were never there.

So ultimately, bounce rates in analytics tools can typically be significantly worse than reported when web performance is poor.

> One technique I've seen work well for getting bosses to care is showing them how much slower their sites are than their competitors - nobody wants to be worse than their competition.

were the filmstrips enough to convince them, or did your bosses want to see numbers/stats like how many were dropping off your site for each x-seconds of delay?

I sometimes feel like I should sell a consultancy selling EVaaS - External Validation as a Service. :)

I've been fortunate to spend my career working at sane companies, but I've talked to so many people who were in situations where management wouldn't listen to their own staff, but then turned around and listened to (and implemented) the exact same recommendations when they were made by a consultant.

I know there are often valid reasons for this, and external validation can be important. But it sure must be frustrating as hell to see your company pay $25+k to receive a list of recommendations you already made.

Management consultant here; this is literally what I do probably 50% of the time. Our buyer has a strategy they want to implement, so they hire us to help them sell it to their boss. Usually we improve them of course; but our buyers can generally handle this stuff on their own -- their bosses don't trust them, and want someone else on the hook.

Trusting subordinates who are incentivized to fudge the numbers is very difficult. Most people won't outright lie to their boss to get ahead, but they will make a bad business proposition look good to the point they genuinely believe it themselves. Leadership is hard.

I'm sure some of it can be contributed to the management itself. If you hire and trust people, and then fail, you failed when hiring and trusting them. If you use a consultant, it's very easy to shift the blame without taking any yourself.

That is mostly a problem if the people above the management are the same. For a sane company with good leadership, failures are a learning experience that makes you better.

>It is astonishing how hard it can be to internally sell any kind of web quality features to management in both for profit and non-profit organizations.

In my experience it's next to impossible. We have a bunch of inefficient bloated legacy shit that barely works and we spend most of our time bug fixing. Yet we keep adding to it while hiring more and more people to fix the increasing amount of bugs.

Why not provide the bottom / top line impact of page optimization? Your business’s profitability is the most important thing to your bosses. They do not care about making the web good, nor should they IMHO. Your job is to show why both parties’ incentives are actually aligned, not opposed.

Death by a thousand cuts. Everyone wants to add just one little resource, until you end up with 8MB pages with 20 seconds time to interactive. They can all make a plausible argument for how their one little resource will improve the bottom line, even considering the performance impact. If you have one person or one team responsible for performance, they're constantly fighting on all fronts just to stand still. You need a business-wide performance-oriented culture, which is very difficult to develop and maintain.

"You can add your just one little resource, but unfortunately the loading time goes over the limit - you need to identify other resource(s) that can be removed first."

Hard performance targets can work, but only if you have full buy-in from management. If you don't, someone is going to persuade a higher-up to make an exception just this one time and the whole thing falls apart.

You" need a business-wide performance-oriented culture, which is very difficult to develop and maintain."

How about customer-centric culture? One where UX is #1, and not out of sight?

perhaps I've been unfortunate but everyone I've met who markets themselves as UX are people whose job seems to be to say how will the customer understand this design, how is the customer workflow, and do not give a damn about performance at the level needed to achieve the improvements being discussed here.

I'm a dev and I understand the frustration of multiple resource load and the bandwidth burden it imposes but also as an entrepreneur and business owner, I see the need for metrics when making a sales pitch.

> Your job is to show why both parties’ incentives are actually aligned, not opposed.

This!

It's your job to do just that because of your vantage point i.e. you're able to see everything more broadly than anyone else. Understanding that all the stakeholders in your organization are doing something important that allows everyone to have a place to come to work to tomorrow is critical for you as the engineer.

> It's your job to do just that because of your vantage point i.e. you're able to see everything more broadly than anyone else.

It sounds very good on paper but it doesn't really work in practice because devs are generally not recognized as important decision makers in product design, despite said "broad vantage point".

From there, it's a rather thankless uphill battle of the "no good deed will go unpunished" variety. It's a very misaligned situation.

The need for metrics is another way of saying "I don't trust you", which may be understandable to a degree but stakeholders should also understand that not everything can be measured, and that measuring things in a manner that is both accurate and useful is very hard. I'm skeptical that metrics have more value than the word of a seasoned engineer on a project. We don't have very good results with measuring other vague things, why would software development be any different?

> I'm skeptical that metrics have more value than the word of a seasoned engineer on a project.

That's a rather odd statement to hear from a dev (or developer advocate) because I've always heard fellow engineers talk about how data is more important than opinion. Why is this an exception?

> "Your business’s profitability is the most important thing to your bosses."

I think the UX is the most important thing, without it (i.e., a good one) there is no profit.

In 2018 bosses involved in web - directly or indirectly - should be well aware of this. At this point in time the web dev's job should not be to "fix" those above them. There's plenty else to do and keep up with qithout having to add risking their career to the list.

And few if any freelancers are going to speak up. Which is all the more reason the decision makers should understand the implications of their decisions.

I agree 100%. So instead of saying “The web needs to be good! We need to spend time (money) on this!” Say “I ran an A/B test and the top of our funnel had an X% higher conversion rate when the page loaded in less than Xms. Here are my ideas for what analytics to cut to get us under that number. It’ll take some work but our funnel will improve and there’s an added benefit because all execs will view a unified report, which will ensure that all future business decisions are made with one standard set of metrics. This is win win win!”

Content sites get traffic from google / facebook / etc and then monetize it. Unless the site is so bad that people just close the tab rather than read the article, usability doesn't matter. You're only going to get one page view anyway.

Is Tedium worse than Medium?

> This. AMP was a blessing for me honestly. I can now maintain a version of our new site that isn't bogged down with tracking and flavor-of-the-month JS feature garbage.

That doesn't sound like it relieves you of maintaining a bogged down version of the site, it just means in addition to maintaining the bogged down version you have to now also maintain the amp version, and keep them consistent.

Yes, and execs will give you more budget to maintain the additional site, which they're encouraged to do due to Business Deals or whatever they read on LinkedIn.

Can you simply replace AMP with GDPR?

Nah, blame the EU for all of those pop-over ads that you see on sites like Tedium. You first saw pop-overs to harass people abut cookies as required in the EU, thus everybody learned how to make pop-overs and pop-overs were legitimized because they aren't just advertising but something perceived to be required by regulation.

> You first saw pop-overs to harass people abut cookies as required in the EU

That's a rather a-historical way to describe that. The various cookie notifications used what was already common code to implement it since most people didn't want to spend much time on what they saw as a formality.

It's more to do with declining ad revenues. GDPR has a role to play, but so do ad blockers.

You're making nothing at all from 20-30% of your visitors, because they're blocking ads. If you're actually following the GDPR, then your average CPM for EU users has tanked because you can't track them and can't serve targeted ads without an explicit opt-in.

Your average revenue per page view is declining, so what do you do? More aggressive ad units and more of them. Popovers and interstitials aren't new, they're just becoming more widespread because many publishers are starting to get desperate. Your bounce rate will increase, users will eventually learn not to click your links, you'll push even more people to use ad blocking, but none of that matters right now because you've got revenue targets to hit.

Sure, blame the EU for trying to stop companies from keeping your contacts, pictures, location history or whatever just because you signed up to write a review or comment an article.

If companies can't manage to offer their service without exploiting your data and have to blacklist Europe, that's a win to me.

But is it worth the cost of forcing all EU users into not even having the option to use said service for data?

Thankfully, I don't live in the EU because I would be super pissed about all the websites blocking me because the EU no longer wants to give me the option to trade some data for a free service and I expect it to get a LOT worse once they start actually enacting fines and every company realizes that that they have really been force opting in people to data collection so it is no longer profitable to serve the EU.

In my opinion, it would be far far better to force companies to offer a "fair price" paid service in exchange for not collecting data (I'm 100% for taking them to court if they misrepresent the value your data provides so they don't overprice it forcing people to choose the free option). That way, I can choose to keep getting my free service in exchange for data and you guys can pay to protect your privacy. Does this option sound reasonable to you?

> But is it worth the cost of forcing all EU users into not even having the option to use said service for data?

Pretty much everything you stated here is completely wrong. GDPR states that personal information can only be collected on an opt-in basis. Your entire statement therefore is completely off.

Because it has to be freely opt-in you cannot just do an opt-out.

> In my opinion, it would be far far better to force companies to offer a "fair price" paid service in exchange for not collecting data

The GDPR doesn't force to use anything for free.

> GDPR states that personal information can only be collected on an opt-in basis.

I said "use said service for data" which means all or nothing, no ability to use a service and selectively decline the exchange of data that makes the company you are exchanging with money.

> The GDPR doesn't force to use anything for free.

I didn't say anything about that. I just don't like that free online content like the la times is now blocking all of the EU because their business model is incompatible with GDPR and so I would prefer if the law still allowed EU users the choice to accept said business model, or use one that contributes revenue in proportion to the old one so their business model works while also satisfying your wants. Does that make sense/sound reasonable?

I have no idea what you're talking about.

I think you should at least know the basics of the topic you're discussing before leaving nonsensical comments.

Everything you're saying is completely wrong.

What is wrong about it?

> not even having the option to use said service for data

Companies are blocking all of the EU because of GDPR which is denying them the "option to use said service for data". Here is one list so far: https://gdprcasualties.com/

> Thankfully, I don't live in the EU because I would be super pissed about all the websites blocking me because the EU no longer wants to give me the option to trade some data for a free service and I expect it to get a LOT worse once they start actually enacting fines and every company realizes that that they have really been force opting in people to data collection so it is no longer profitable to serve the EU.

I don't know of a single company that now gives users to opt into each data collection so if they actually start throwing down huge fines (which I fully believe they will based on their history), I expect it to get a lot worse.

Are you the sort of person who likes paying protection rackets?

Assuming (correctly) that web sites are untrustworthy data collection bandits, why should they behave well only because a user proved their submission (and their gullibility) by paying them?

> protection rackets

Didn't know what that was until you said it and still not sure how it applies.

> Assuming (correctly) that web sites are untrustworthy data collection bandits, why should they behave well only because a user proved their submission (and their gullibility) by paying them?

I think you are misunderstanding me. I love everything about GDPR except 1 thing and that thing is not allowing companies to tie providing their service in exchange for data so that they can make money because it denies people the CHOICE (key word here) of using said option.

I proposed an option (just pay for the lost revenue the company no longer makes) so that everyone still gets the option to use current revenue models, pro privacy people can pay their fare share for the service, and companies don't blacklist all of the EU for what a portion of the people want. What does not sound fair about that?

It's the other way around. The EU is blacklisting data collection bandits; their "revenue models" are unacceptable and therefore forbidden by law.

Bandits implies stealing which is disingenuous as you have the option to not accept their terms of using the website.

I'm curious, how do you propose to fund the web? Should every site now have paywalls?

> trade [...] for a free service

You cannot do that not because of the EU but because of basic logic.

maybe temporarily, but as soon as all the third-party js that slows down your site start advertising themselves as GDPR-compliant, that goes away.

Seems like it is more of a problem with how the company is ran than a positive of using AMP.

Just like every other issue in every company ever.

It's a problem with how almost every company is run.

> If you can find a better way to convince large organizations that page load speed is a valuable metric, and more important that whatever other resource they want to load today, I'd love to hear it.

Yes...do all the same things but without requiring Google proxy it. There are several things orgs will do to optimize SEO, conforming to AMP rules can be another. Gaming SEO is a concern for all SEO requirements so surely that's a poor excuse for proxying.

EDIT: But it did give me an idea on how to tell websites you aren't happy with something: https://github.com/cretz/software-ideas/issues/91

Just off topic. Love the software ideas as GitHub repo idea.

Am I only one who feels this is a modern-day reincarnation of WAP? We've been here a decade ago.

https://en.wikipedia.org/wiki/Wireless_Application_Protocol

Yes, but this time you have to donate your soul to Google. Otherwise, your page will just rank lower. WAP didn't have that direct connection with Google search results.

Wow, this is one of those flips in perspectives that leaves me a bit dazed. Thanks!

Or slow sites could just die on their own? Is there really a need to try to force the issue? Can't users just get tired of waiting and start going to sites that don't make them wait?

Honest question

I hate needlessly heavy sites. I just avoid them. I'm also it a fan of AMP. I especially hate the URL is wrong and trying to share the correct URL is tedious

How many times have people tried to get sites to be lighter and failed? There is no evidence that we'd just end up with small sites naturally.

How is a user supposed to know which sites are slow in a search results lists? Mental list of bad domains to not go to?

Search engine puts a page speed icon in the result, or uses speed in ranking. Not tagging only sites that use the engine's own code.

> AMP's innovation isn't a way to make pages fast. AMP is a way to sell other stakeholders on implementing technologies that make your website fast.

There's a certain "theory of the firm" that says corporations exist because they reduce the overhead cost of operating certain enterprises.

Things like this -- along with the also oft-seen need to hire consultants to tell some people in your company to do what other people in your company already know needs to be done -- make me wonder if the limits of that theory are well understood.

"The Nature of the Firm" by Ronald Coase. He called it transaction costs. I ran across this while trying to grok one firm I worked at that managed thru over-governance, funding models, etc to maximize internal transaction costs to the point where it was easier to just hire external expensive consultants.

But you could still do that without amp.js. Google would just have to push a validator that checks whether a page is the AMP-subset of HTML and CSS, doesn't do any sync loading, etc. etc.

And then the engineer could point at validator results.

> It's a lot easier to say "i can't do that, it doesn't work in AMP". If you can find a better way to convince large organizations that page load speed is a valuable metric, and more important that whatever other resource they want to load today, I'd love to hear it. But from what i've seen, AMP is the only thing that's had any success in tackling this problem.

Have google results punish fat pages. Then we can say "I can do that, but you have to sign this the declaration accepting responsibility for our page hits tanking".

This guy. This guy has been there.

There are other optimizations AMP implements that aren't currently possible with regular web pages, notably [their ability to be automatically preloaded by the browser without compromising the user's privacy][1].

Once [cross-origin server push][2] becomes standardized, regular webpages will be able to take advantage of similar functionality.

[1]: https://medium.com/@pbakaus/why-amp-caches-exist-cd7938da245...

[2]: https://wicg.github.io/webpackage/draft-yasskin-http-origin-...

pre-fetching links is not a new optimization technique. And it doesn't require any new technologies to implement.

It's true that this exists but there are complications – for example, rel=prefetch has no JavaScript API so you either need to put rel=prefetch for all of the top results or do something hacky to trigger it for only the link the user hovered the mouse over before clicking, etc.

There are also privacy concerns – e.g. the remote server could set a cookie in a response without you ever visiting that site but if you disable cookies by default users will have to refetch the resource anyway if there's any session state involved in page generation. You also have issues like companies wanting to distinguish between traffic from actual page views and the prefetch mechanism.

That's not to say that the answer is AMP — e.g. since it puts so much render-blocking JavaScript into the critical path prefetching is often just canceling out that cost – and it's sad that the marketing push has made the discussion of alternative performance improvements unnecessarily contentious.

"Without compromising the user's privacy" is the key detail there.

No, it's not only about size and js restrictions. Here is a summary of the strategies employed (edited from [0]).

- Execute all AMP JavaScript asynchronously

- Size all resources statically

- AMP uncouples document layout from resource layout. Only one HTTP request is needed to layout the entire doc.

- All CSS must be inline and size-bound

- Minimize style recalculations

- Only run GPU-accelerated animations

- Prioritize resource loading

- The new preconnect API is used heavily

- When AMP documents get prerendered for instant loading, only resources above the fold are actually downloaded. Resources that might use a lot of CPU (like third-party iframes) do not get downloaded.

[0]: https://www.ampproject.org/learn/about-how/

OP's point was that those are all techniques that you don't need AMP to implement. So their question is: if those are the reason for you to use AMP, why use AMP and not just implement them directly?

Why not use AMP? If you don't use the special caches, it's just another library.

I really question this "inline all CSS". Consider:

- 1K of HTML with 40K of CSS in a file with a long term cache. Clicking a different page on the site downloads another 1K of HTML.

- A 41K file with everything inlined. Clicking a link downloads another 41K.

The recommendations are to inline the important (render-blocking) CSS, so at most only the part of the CSS that applies to the 1K of HTML should be inlined.

The recommendation you describe is a violation of the AMP standard, which strictly disallows this for performance reasons.

Sorry, I was commenting generally. Don’t know about AMP. How does that make sense, though? What performance reasons could there possibly be for inlining CSS that doesn’t apply to any element in a page?

Likewise, I'm sorry, I didn't mean "you have to inline CSS that doesn't apply".

However, if you have non-render blocking CSS, or CSS that's used for below the fold or generally lower down the page content, "only render critical CSS inline" is usually coupled with "and then have the rest of your CSS in an external stylesheet". Which you are not allowed to do.

Accordingly, it's ALL inline, all the time.

When surver push of resources gets better supported by browsers, I bet AMP will be updated to change this restriction.

Ok, but that does not sound like a real world example.

Why not? The Medium link we're all looking at right now has a 43KB CSS stylesheet. It's sent with browser headers giving it a long cache.

In the AMP world, this whole stylesheet would end up inlined, and you would download it again the next time someone posted a Medium link on HN.

Google creates an environment where you "need to use AMP" because they prioritize search results for properties that have AMP varietals - placing them in the "carousel" on top of search results. You can build a very performant static site, with terrific SEO and social share meta tags, etc. - but without AMP - and still be positioned below another property who's "regular" site is slow and full of cruft but has an AMP varietal.

If this isn't a sign that Google is a monopoly, I don't know what is.

No, Google is the dominate search engine which is part of an oligopoly which while still not as economically efficient as a perfectly competitive market, they have nowhere near the power of a monopoly.

If Google replaced all of their search results with ads, people would easily switch to bing as bing isn't that much worse.

Facebook on the other hand definitely is "pretty much" a monopoly (but not completely) for certain subsets of the social media market due to network effects which is why they shovel ads down your throat on their platform and they are making tons of money.

Google has an unseen network effect...

Their search results are good because they do machine learning on data from all their other users.

A lesser-used search engine has less data, so even if they have smarter people and a better algorithm, the search results will probably be inferior.

That is fair. However, I still argue that it is not nearly as strong as it is easy for me to switch from google search to bing as most of my search results won't really change as they are basic searches but it is incredibly painful to switch social media networks as you have to rebuild your individual content and network connections for the new sight.

I actually have set Bing as my default search on my phone

... and let me tell you Bing is not nearly as good as Google when it comes to bringing me the right search results first.

But I got tired of Google asking me to “prove that I’m not a robot” by tapping on roads and street signs with every new search. I use incognito mode, and since they can’t track me, they either are punishing me or just automatically assume I’m not human.

Google replaced all there search results with ads a looong time ago. I blogged the first time I saw Google results had all ads on the visible part of the results page. Now often its all ads for the whole page. And you will not find any page of results that is not defined by Google's definition of what "organic" search should be. E.g amp, https, having an adwords account, hosted by them, tuned your page to their adwords requirements. etc etc.

Can you take a screenshot because I have never seen a page of all ads? I just did a few random searches for typical highly advertisable things and I got 1-4 ads but you could still see organic search results. Granted the "asbestos lawyer" search did have half of the original screen area full of ads but scrolling down showed most of the actual page with organic search results.

In a way, Google IS the internet. AMP prioritized searches moves them to even more centralized internet where they control everything, host everything and can dump your site at their whims.

Right. Absolutely. People should just follow some of these restrictions to get speed without AMP. I went down this AMP path though because of yes the Icon. Google was testing the Icon in Paid Ads I had read, and wanted to see if that could also improve our Click Through Rates. I also wanted to see if officially being AMP could improve things like our Quality Score in our ads, which would reduce our click cost, etc.

So yeah, just me trying to get blessed by G so everything else improves from G for us :)

> People should just follow some of these restrictions to get speed without AMP

It's easier for a dev at a newspaper to say "sorry marketing, Google AMP doesn't allow your javascript" than to actually argue with marketing.

Unless the dev has the authority to make the decision, what he should really do is explain the tradeoff: “Hey marketing, I know you want this feature, but it’s gonna cause us to drop down in Google results. Is it still worth it?”

Yeah, good luck using reason with these people. They'll just stare at you and say something along the lines of "But it's Google and they say so."

Tell them that it won't support all their various analytics scripts, tracking pixels, and A/B testing scripts that fill up their charts with vanity metrics they use to impress stakeholders.

Never underestimate the willingness of salespeople to trade potential revenue booked tomorrow for real revenue they can book today.

> Never underestimate the willingness of salespeople to trade potential revenue booked tomorrow for real revenue they can book today

Counterfactual: lots of terrible business decisions are made because someone gets on an ideological bent and runs with it. Revenue today is worth more than revenue tomorrow. Balancing growth opportunities against investment risks is the core of commercial decision-making.

Oh, I don't disagree! But the key word in your response is "balancing." That implies there is some neutral party weighing the pros and cons of each new script, whereas in reality, what usually happens is you have the marketers on one side saying "all scripts are great" and the techies on the other side saying "all scripts are terrible," and the side that wins comes down more to which has more political power in the organization than to the actual business merits of the individual script. So lots of sites end up absurdly over-monetized and others absurdly under-monetized, with very few finding a sweet spot in the middle.

What does "under-monetized" mean here? How many scripts does it take to serve an ad effectively?

It's just another case of people hiding behind a computer; For some weird reason people trust things when "computers say them" much more than when their coworkers do.

I had a thought the other day:

Is there a mobile browser that tracks bandwidth used, and can tell me how much money I'm spending on bandwidth per domain?

I'd like to know which sites are parsimonious about page size. Then I'll limit my mobile browsing to those and ditch the ones that don't care.

You can get that information from, e.g., a Squid proxy (the daily/weekly/monthly reports). Even for non-cached content.

Brave shows how many ads and trackers it has blocked, as well as a number that's labeled Time Saved. How that is calculated I haven't the foggiest, but it must be related to download size.

As far as I'm aware, Brave is open source and this appears to be the time saved calculation on Android. https://github.com/brave/browser-android-tabs/blob/b4d565f07...

Interesting. I (like others here, apparently) assumed that the time saved number was based on various tracking scripts that weren't downloaded, but if this is the relevant code, looks like it's just tracking parallel requests.

You know, I think you're right. The formula from the PR that implemented "Est. Time Saved" appears to be (trackers + ads blocked) * 50ms

https://github.com/brave/browser-android-tabs/pull/331/files...

Probably `Content-Length` of requests they didn't serve divided by your average download speed?

Brave shows this on the "new tab" page. I was surprised with how much these numbers added up over time, though I do suppose they are reasonable estimations.

> If what makes AMP fast is the restrictions on size, JS, and CSS, and you know this and want to conform to this, why do you need to use AMP? Why not just develop your site like this anyways?

Maciej Cegłowski (aka idlewords) recommends exactly this is in his (2015!) talk "The Website Obesity Crisis": http://idlewords.com/talks/website_obesity.htm

His specific comments about AMP: "AMP is a special subset of HTML designed to be fast on mobile devices. Why not just serve regular HTML without stuffing it full of useless crap? The question is left unanswered."

AMP is just a fork of HTML, creating more pressure and work for publishers to implement.

The reason why they do it is because of better search page placement in Google. It's unfair and wasteful, since site speed can easily be measured by the most advanced search engine that already has tools and reports doing exactly that.

AMP is not a search ranking signal. Non-AMP pages will end up in the carousel in 2019... if sites can easily be sped up, why do 80% of websites still suck so much and are slow? AMP won't be for specialized dev teams, but it helps plenty of average websites.

> Question: If what makes AMP fast is the restrictions on size, JS, and CSS, and you know this and want to conform to this, why do you need to use AMP? Why not just develop your site like this anyways?

I think aside from the icon and the special treatment mentioned above, don't AMP pages get served directly from google and get preloaded on the search results page? So that's the other benefit, I guess. Whether a few less milliseconds from preloading on an already fast page really buys anything is another question.

> Whether a few less milliseconds from preloading on an already fast page really buys anything is another question.

On the technical side, it hardly matters. If the sales and marketing teams have bitten, they sell to management, and tech is informed they will be implementing it.

That's often how these things go, anyway.

AMP can be served by anybody and browsers will preload likely destinations regardless of AMP or not.

Actually the main performance gains is by google search preloading everything (basically putting it in an iFrame before you click on it). Which practically means that pages load instantly.

https://timkadlec.com/remembers/2018-03-19-how-fast-is-amp-r...

Google could easily adjust their ranking algorithm to better reward non-AMP pages (which they've obviously already crawled) which they know are 1) below say 75 KB and 2) contain a limited number of resources to fetch. But instead of doing that, they choose to attempt to strongarm everyone into using AMP.

In the article they say they're planning on doing exactly that.

Hopefully they actually follow through on that.

It looks like the only reason to use AMP is to bypass ad blockers and sell more advertising, and this is also the only reason why Google developed it.

Personally, I think it's stupid from content creators, because if I visit the website I might read more articles, share, or sign up to the newsletter, and that doesn't happen with all AMP implementations I've seen.

One lesser mentioned feature of AMP is when they are found through Google search, they are prefetched. So in addition to being size restricted, they are probably already partially if not completely downloaded in the background before the link is clicked. Giving them an extra advantage over other search results.

You can implement this feature yourself too: https://developer.mozilla.org/en-US/docs/Web/HTTP/Link_prefe...

They have a little carousel at the top of the search results containing only AMP pages, right?

Ah, yeah, that's true too. Wasn't critical for us on this landing page stuff. But yes, that's another reason blogs and news sites are AMPing their stuff.

Sorry if this distracts from the topic, but one of the biggest reasons I, and a few people I know, don't like AMP is that - this will sound trivial - but they're a real annoyance when you want to cut and paste a link into an email or a text message when you're on a phone. I'm not going to screw around with finding a share button or what have you on a mobile site, nobody knows what JS is behind those.

So, lack of UX with the links.

It drives me bonkers and makes me irrationally angry.

I don't avoid them out of principle, but since JavaScript is turned off by default in my browser none of them work for me to begin with. I can't whitelist AMP pages like I can non-amp pages I browse. AMP pages over-all are a terrible experience.

Sadly ironic that the format doesn’t allow JavaScript to ensure pages are lightweight and fast-loading, and yet disabling JS stops them from displaying.

> ... why do you need to use AMP?

Because you need to convince business folks about stripping down bloat from your webpages. And doing that by citing sticks and carrots from Big-G is much easier than doing it on your own. So we can attribute this to: laziness of developers (who cannot make arguments to reduce bloat), boneheadedness of business folks (who cannot quickly understand why bloat is bad), tragedy of commons (my competitors are gaining valuable traffic by having their articles in special Google carousel at the top of search results, I am missing out so I must do AMP) etc.

There's no laziness or boneheadedness involved. Without the AMP carousel, making a fast website is a bad business decision. With the carousel, it's a good one.

The usability of your site just doesn't really matter if your business is to get traffic from Google and monetize it, which is what we are talking about here.

I always avoid AMP pages but we are a tiny minority. Most people I know like AMP pages because they are fast.

I mostly click on AMP links if I'm on mobile. I know those load fast, so if I see one then that's what I hit.

Regarding your recommend point, I think Occam's razor applies. Seems unlikely there is an unconscious collective purposely avoiding AMP links.

It's much more likely that the changes done to a website to satisfy AMP affected the conversion rate.

The author was wrong. AMP is a way to prerender web pages in a way that plays nice with analytics. The rest makes the preload fast, but the user doesn't experience it because when they navigate to the page, it is already rendered.

TBH until yesterday when I saw the article I did not even notice the lightning bolt, my brain was automatically filtering it, and now that I know it similarly to you I probably will start skipping those links (similarly what I do with responses tagged as "Ad"). The reason for it is the same why I use uMatrix, I don't want to provide Google with more information about me.

So unless Google actually adjusts the ranking it is useless to get AMP. I wish though that many web sites would chill down with including so much of JS on their pages.

If it was up to developers, it would be. It is not up to devs.

You're not the only one. I will load AMP pages and then tap through to the source page just to make sure I'm not letting Google control the experience.

The only thing that made people use AMP is that there were rumours that Google would rank AMP sites heigher.

I think that's the only reason.

Rumors? Just about every Google search I make brings back a carousel with exclusively AMP near or at the top. It's absolutely prioritized.

Is it possible for Google to factor in page load speeds into their search algorithm? Hopefully in a way that actually encourages publishers to optimise for performance and not focus on fooling googles webcrawlers

Because nothing will be faster than the Google-cached copy of your page, and that’s only served up if it meets the AMP spec.

Sometimes its not always economical, especially if you need your engineers to focus elsewhere (like the actual product as opposed to the marketing site), or if you have a large organization and it is difficult to lock down website discipline. Should sites be engineered to be efficient? yes. Is everything that should happen actually realistic to achieve? Not exactly.

I never click on AMP links. Google ≠ the internet.

HN is utterly incapable of discussing AMP honestly or coherently, so you're hardly going to get many reality-based replies.

AMP is offered as a promise to the user, not the publisher. When I see AMP it guarantees that the site has minimally invasive JavaScript (and the scourge of nonsense that comes with many sites now), will load quickly, etc. The site cannot suddenly bait and switch and start mining bitcoin or spamming pop-overs because it has been technically restricted from doing so.

So yes, sites can be every bit as fast as AMP. And industrial manufacturers can be every bit as clean as required by regulators if they had no regulation. But they aren't, and they don't.

We need an HTML Lite as a mode in the browser that winnows down the enormous featureset of HTML. Not for all content, but for that content where text is king.

Pardon my naivete, but Google validates whether a site is AMP-eligible by validating it conforms to its requirements, right? Can the Google bot that crawls the site not perform this same validation? I mean they can bait and switch the title of the page, the content, etc too but we don't require a Google proxy to make sure they don't. Can anyone else build a proxy that gives the same AMP guarantees and gets the same ranking benefits? More than a promise to the user, AMP is a promise to everyone where the bits are coming from.

Google is going to get in trouble stewarding their gateway into the internet in this fashion with these rules. Rules about slow page load times can be seen as universal. Rules about using a Google proxy however are just begging for political intervention. As someone with a small business, I'm upset that these companies, under the guise of helping users, are giving reasons for these liberal governments to act. It negatively affects me when these inevitable regulations come down because companies like Google can't remain provider-neutral. Arg!

> We need an HTML Lite as a mode in the browser that winnows down the enormous featureset of HTML. Not for all content, but for that content where text is king.

I concur here and have been thinking about this recently especially in the context of web browsing in the terminal. But it has to be driven by user adoption, not rules written on the walls of a few companies' gardens.

In the meantime, I would ask that the search engine not give preference to their proxy over anyone else's but instead define the guidelines that we can all reasonably meet.

> Can the Google bot that crawls the site not perform this same validation?

From what I am aware, in practice there are a large number of sites which provide different content or different behavior in response to the GoogleBot user-agent.

And yet we are still OK with its results everywhere else right?

The benefits are nebulous in other instances. Here they are much more concrete, and thus will undoubtedly (in my mind) be abused to a much higher degree.

It is a web platform security restriction that prevents Google Search from coordinating with non-Google caches for prerendering.

Pinterest runs its own AMP cache for prerendering pages linked from Twitter. Bing runs its own AMP cache to prerender pages linked from Bing. Yandex runs its own AMP cache to prerender pages linked from Yandex. They can't use each others' AMP caches and get the same benefits.

I like AMPages when they are served by the content creator, but I hate AMPages when they are served on Google's CDN.

Honest question: would the author of the linked article have experienced the same issues if the users had actually visited his own site's AMPages?

Author here. No, not at all. Well not the first two problems. The extra chrome is just when google is hosting it, and the blank pages only occurred in testing from the CDN.

We need an HTML lite but AMP doesn't get us anywhere close to it because it's a custom dialect of HTML driven by a bunch of proprietary Javascript. The google search spider enforcing a 'lite' subset would have gotten us closer, if not there entirely. We know they can do it because the spider runs JavaScript, they could examine loaded content and measure time to first content. They don't because it's way more effective to get content providers into a walled garden where they can be monetized.

Re: "HTML Lite" - I think there's already a pretty solid solution. Hand-coded HTML 3 without any CSS, ECMAScript, or other dynamic elements. That always worked pretty well for me, and pages that could load acceptably on 28.8 would be absolutely mind-blowingly fast now. Of course, that would put the power of web design back in the hands of the average user rather than exclusively designers and devs. Because of that, I don't think we'll ever see that becoming an option again.

Or, based on the browser I use when I care about speed, can we design pages around the constraints of Lynx/Links/eLinks? The CSS layouts work when called for, menus work in text-mode, but without images, *.js, and all the other Web annoyances, everything loads quick even on the worst connections.

The concept of HTML Lite is about restrictions that the source is forced to live within. As with AMP, it would not be an honor system. It would not be something that maybe that random link would conform to, but way more likely it wouldn't.

If I were looking for song lyrics, or a recipe, or schedule information, etc, I would be browsing in HTML Lite mode. I don't want popovers, subscribe now boxes, animated bullshit, etc.

My root post is sitting in the negatives right now, and I own that and embrace it because it is how discussions about AMP always go on HN. A bunch of web developers herd in to tell us how terrible AMP is while the web gets more and more bloated, more and more destructive and tragedy of the commons, and we all layer on various shoehorned, half-assed solutions to fight back (e.g. Ad blockers, nuisance blockers, tracking blockers, etc).

Well, I'll say this much about AMP and how it is: I actively avoid it; but then again, I don't use GMail, I search with DDG or Exalead most of the time, etc. I like the idea, I hate the implementation. I don't like a single corporation determining the course of the Web. I like the idea of a constrained standard. Then again, I also like the unrealistic idea of a single monolithic standard. When I got into making webpages, you had HTML and that was it. Then they said, "well, if you want to do some cool stuff, use CSS in addition to HTML". And now, they're trying to make it so that creating a website without using CSS or JS is completely impossible.

Really, what you're saying you want just links back to what I suggested - disable remote CSS (preferably also inline but that's a pipe-dream), disable JavaScript, disable images, and turn your content blocker on full blast. Either that or just use Links. It works for me and blocks all the crud so I can get the content I actually want. The only thing I'd like to do with Links is to change the terminal colors it uses (which I don't think I can do in Windows - if you know how please do tell me). HTML was a fine language for what it does, and it still serves as a very solid core for a Web made of pure content. Only a Web made up of utter corporate nonsense needs scripting.

Well, speaking as a web user, I hate AMP, and actively avoid websites where there's no other option. The reasons? Most of all, I don't like how they break the Back button in strange ways (I still haven't figured out how exactly it's supposed to operate, but it's certainly not working as it usually does in a web browser). And I don't like that all URLs are showing up as Google.

In abstract, I also don't like the notion of giving Google that much power. Even if what you say makes sense, and regulation is necessary, a private company doing it is the worst of both worlds - they are not accountable to the rest of us, and they have their own goals that contradict the public good.

> HTML Lite

Couldn't agree more with this. People approximate it with e.g. noscript, it would be awesome to have a standard which describes a minimal HTML Lite spec though.

Edit: It would be awesome as another option in the link context menu: open in new tab, open in incognito tab, open in lite tab

Having this enforced by the browser as a mode would be brilliant. The HTML stack has grown enormous, with it the potential for a abuse and misuse. If you are just looking for a digital version of paper, so to speak, 95% of that is simply completely unnecessary.

Right now the AMP CDN enforces the restrictions of AMP (you can't simply pretend it conforms to AMP to Google and then feed the client something else) which is something that is fundamentally misunderstood when this is discussed on HN. AMP is a protection and guarantee of behavior for the user, it is not for the publisher, it is not for the developer. Google has a profound vested interest in trying to maintain people's interest in the web, especially in the face of various alternative walled gardens like Facebook feeds, Apple News, etc.

> Google has a profound vested interest in trying to maintain people's interest in the web, especially in the face of various alternative walled gardens like Facebook feeds, Apple News, etc.

To me, AMP is just Google trying to turn the open web in its own walled garden.

there's NOTHING technically requiring google to serve AMP from their own domain. None of the benefits you cited answered that. AMP could just be another standard where the basic lib is preloaded.

I feel like this conversation, like all AMP conversations, is going in circles, spittle-filled comments denoted by unnecessary angry screeds of all caps.

Yes, right now there is a technical need for an intermediary because it guarantees AMP is actually AMP. It is trivial for a page to say it's AMP, load the AMP standard library, and then do everything disallowed by AMP. There is absolutely nothing preventing that but that intermediary rejecting non-compliant content.

Now of course we've talked about an HTML Light and that would be the browser enforcing that limited sandbox. It could send a relevant cookie and then reject content that steps outside the bounds. But we don't have that right now.

> Yes, right now there is a technical need for an intermediary because it guarantees AMP is actually AMP. It is trivial for a page to say it's AMP, load the AMP standard library, and then do everything disallowed by AMP. There is absolutely nothing preventing that but that intermediary rejecting non-compliant content.

That's because the standard is not well defined. If you had a doctype just for AMP at the top, that would work. I don't see any technical reason why google's servers are necessary. I see a business reason for Google of course but there's nothing technically that makes sense.

> Now of course we've talked about an HTML Light and that would be the browser enforcing that limited sandbox. It could send a relevant cookie and then reject content that steps outside the bounds. But we don't have that right now.

Well, AMP could be just that if it was standardized.

Good idea. As a starting point, let's define "HTML Lite" as "HTML without Javascript".

Then "HTML Fat" and "HTML Light" could even be served as one document. If you execute the Javascript, you get the fat version. Else, you get the light version.

If you use Javascript not just to improve the behaviour but also to display content at all, you can provide fallback content that is only displayed in light mode. The fallback content is marked by a special tag. Let's call it `noscript`.

My point is: Making it easier to serve sane and lightweight HTML will not help fix the problem. Publishers don't want to do that, they like all that crap. If we want them to make sane web pages, we must force them.

Reader Mode on Safari maybe?

Maybe roll back to an old browser?

Along with all the security risks? Or do I get to choose just the good parts? :)

AMP promises me that the site won't load.