Hacker News Reader

Comment by akras14 | Hacker News Reader

>However, there has been some misunderstanding about how AMP works. One widely circulated blog post written back in October claimed Google was stealing traffic from publishers via its AMP pages.

I am really happy to see this change, as the author of said blog post :)

> But that wasn’t true. Google does display the AMP URL in the search results, which serves up the page content from Google’s cache, but the traffic remains the publisher’s, and the content is served from the publisher’s site.

So which one is it? Does it server content from Google cache or from publisher's site ;)

Link to the original blog post: https://www.alexkras.com/google-may-be-stealing-your-mobile-...

View on HN

I received the same sort of self contradicting response from Cloudflare. Their AMP bot mirrors and rehosts any pages one of their subscribers with AMP enabled links to. And their FAQ is so badly written it seems like english may have been their second language: https://support.cloudflare.com/hc/en-us/articles/11500063530...

>Will Accelerated Mobile Links if the AMP links are pointing to content publishers who are not part of Cloudflare customers?

>Yes. Only the discovery site (the website that has the links to AMP content) needs to be a Cloudflare customer.

When I saw their AMP-bot in my server logs I emailed them about this. 2 weeks later I finally managed to talk to a human. That was about 4 days ago and they still haven't responded. If you're not a Cloudflare customer they don't care that they're re-hosting and serving your content.

robots.txt and dmca seems appropriate

Doesn't DMCA have specific exemptions around caching.

my understanding is possibly, but not in violation of robots.txt

Yeah. After I saw the bot I added it to my robots.txt and I blocked all of Cloudflare's IP ranges to the server. But the bot had already come and gone. And the email I'm waiting back on from Cloudflare is about what to put in the robots.txt since it's not on the FAQ page. As of right now I'm trying,

User-agent: Cloudflare-AMP

Disallow: /

They are confusing two different meanings of traffic: - Advertising: where is the user coming from - Networking: where is the data being served from

So my guess is that people were worried that Google would steal the advertising traffic and, therefore, revenue.

That's definitely how Google is acting.

Most of the negative feedback I've heard is because it makes it very difficult to bypass amp on mobile (impossible?) and, yes, copy the URL of a google result without linking to google.

So if I have AMP on my site and a user clicks an AMP link, does my web server record a hit? That's what I care about, since I use traffic stats to negotiate deals with sponsors. I don't run third-party (even Google) ads on my site, it's all native "sponsored content", so making sure traffic is recording correctly is pretty important to me.

The first question I hear when I start a conversation about a sponsorship is "how many hits did you have last month" and the second question is usually "how many hits do your sponsored posts usually get?". I need to be able to answer those questions, and prove it too.

Server-side logs are extremely uncool; you need to use JS to record analytics these days (even if the JS just does a POST to your own server).

> So if I have AMP on my site and a user clicks an AMP link, does my web server record a hit?

That can be very easily answered with "no". This is intentional in this design, for faster loads.

Defective by design sure rings a bell....

The server itself doesn't (you don't get extra load), but analytics is forwarded to you. So you can still get the numbers you need to convince the higher ups, just not directly from your server logs.

At first my distaste for AMP has been as a user: broken inertial scrolling, wasted space, and broken links.

But as I learn about it AMP becomes even more distasteful. Waiting for an intermediary to forward analytics to you is horrific.

What's going on at Google these days? Everything about AMP sounds like a bald-faced attempt to destroy the web. At every point there are horribly unjustified architectural decisions that hurt the web but also help Google.

The decisions are justified (although years late) but you either didn't read them or don't agree with them.

which discounts people who use various content blockers specifically to stop intrusive analytics. these seem to be becoming more popular, so it's only going to be more of an issue to measure traffic

Why woul;d sites want to count those users anyway?

> I don't run third-party (even Google) ads on my site, it's all native "sponsored content"

This is a case where adblocking doesn't really happen. So those users are welcome to come along, and view the content.

Does it only forward analytics using Google Analytics? If I use other forms of analytics, anything from Jetpack to AWStats, it sounds like I'd be missing out on recording traffic.

it supports a bunch of analytics, but from my (tiny amount of) research it looks like you'll have to write your own config for those two:

https://www.ampproject.org/docs/reference/components/amp-ana...

> Does it server content from Google cache or from publisher's site ;)

Or does it serve part from Google's cache and part from the publisher's site?

For example, it might serve page text and images from Google's cache but serve ads from the publisher's ad network.

Since the #1 cause of latency I experience is waiting for ads to be fetched from a publisher's ad network, I'm pretty sure AMP would not work at all if it worked this way.

AMP only starts loading ads after the content loaded anyway.

Ads can be served from the publisher's server, but this is rarely done.

> but the traffic remains the publisher’s

I bet "traffic" here means clicks counted by AdSense, so you still have to pay for a click even if the user never really visited your website ;)

In any case, google is just trying to copy Facebook Instant Articles here. They want people to stay within the walls, because they realized they make more money that way.

>so you still have to pay for a click even if the user never really visited your website

Don't you mean so you get paid for a click even if the user never visited your website?

I meant AdWords. But I guess both.

That makes me even more curious:

what is the market share of adwords things (my understanding this is mostly going to be people offering services) that is impacted by AMP? An ecommerce site won't support AMP so won't be affected, and are news sites and blogs going to be using adwords to advertise their news and blog offerings that much?

E: (and then does it even matter if you're ads are still shown on your AMPed site so you get paid per hit?)

Yes. My understanding from conversations with those in the know, and our limited testing at AMP launch, is that you treat the CDN as a proxy. All the downsides of losing visibility to your domain and none of the upside from offloading traffic.

That being said, we did notice some single digit percentage traffic bumps when we rolled out AMP, and it felt really good telling the hordes of product/business people that I literally couldn't add the dozens of analytics scripts they wanted.

it's all so oddly and guardedly worded it looks like it was drafted by lawyers or a north korean tourist guide

Reply on news.ycombinator.com