Did iOS 15 kill Google AMP | Hacker News Reader

Did iOS 15 kill Google AMP

lapcatsoftware.com

79 pointsddeville84 comments

>So, is it possible that Google has given up on AMP in Safari on iOS 15 because of the popularity of AMP blocking extensions?

This doesn't make any sense. If AMP is strictly beneficial for google, why would google give up because of a few people with AMP blocking extensions? Surely it's still worth going after the stragglers? You don't see them giving up on ads just because adblock is popular.

I suspect the reason is far more benign: AMP is enabled on a whitelist basis, and they haven't whitelisted ios 15 yet.

The page also says:

> I know from my own testing that Google search results still included AMP links for the first week after iOS 15 was released on September 20

I guess the revised explanation would be "random regression in ios 15 safari that affected AMP"

Why is this a regression in iOS? Changing the user-agent string is not a regression.

As in, a change/feature added to the rendering/javascript engine caused an regression.

Except AMP still loads in Safari 15 if you spoof the User Agent, per TFA.

"regression" doesn't have to mean "totally broken". maybe it's a little laggy. Maybe it breaks in certain edge cases.

I don't know that I would classify "rendering someone else's feature" a responsibility of iOS 15 or Safari.

Perhaps, although "it was just random" is arguably an even worse explanation.

Seems unlikely. How many people have installed AMP blockers specifically on ios15? How many people have installed AMP blockers on any other platform that's not being UA-sniffed?

While it's only a Bezos number, a paid AMP blocker (perhaps from OP?) was in the list of top paid apps when iOS 15 went live.

Question might be if it inspires more blocking elsewhere.

It is pretty bad news for AMP that Ampolosion exists. It would be even worse if Samsung noticed that popularity and implemented something similar in their browser, perhaps even by default.

Amplosion got a big boost by being made by the same guy who did Apollo (pretty much the de facto Reddit client on iOS).

I don't get bothered by AMP pages that much, but I paid the few euros for Amplosion just to support Christian's work and give a middle finger to Google's attempts to take over the web.

Apollo is not the defacto reddit client. Reddit's own app is.

Reddit's own app is an adware & malware that happens to display some Reddit content.

Source for the malware claim?

I believe (s)he wanted imply reddit's official app displays a lot of inline advertisements and other stuff that makes it difficult to determine what is legit subscribed content and what is sneakily inserted by reddit.

Maybe I am wrong.

From Wikipedia the Free Encyclopedia:

> Programs are also considered malware if they secretly act against the interests of the computer user.

There are two criteria that software must meet to fit this definition:

1. The software must act against the interests of the computer user.

2. This must be done "secretly".

I unfortunately do not know much about the Reddit App for iOS so I can't make a judgement as to whether it fits definition 2. I think I can say with confidence it fits definition 1.

It tracks you for advertising purposes. That's malicious in my book.

It doesn’t advertise to you in secret (how would that even work?). The ads are front and center. It’s user hostile, but not malware.

The ads are. The data being collected to target those ads is not.

I guess a non-malicious way of collecting the data for ad purposes could be to collect it locally and offer the user to review it before sending it off.

I'd be happy if someone would bake mobile teddit or other "proxies" app. I know site still works but this could be done more handy and with aiming at additional features, like pushing yt videos through invidious "pipe".

Nah. Reddit’s client is the official Reddit client, but Apollo is the de facto one for people who want to actually use Reddit.

Also Apollo now has a Safari extension which will automatically redirect Reddit links to Apollo. Previously this was a manual process after the page already loaded after telling Reddit for the umpteen-billionth time you want Safari, not the official Reddit client.

This. I think I would actually give up on reddit without apollo, the official app and website are user-hostile. I wonder if they will tighten the API up and eventually kill third-party clients.

If Apollo is the interface, what is Reddit besides a JSON document store and object storage for media? Once Apollo is big enough (network effects, critical mass, whatevs), it could spin off of Reddit as the backend (Cloudflare R2 for objects, workers and KV for comments). You can even backfill from the numerous Reddit archives (Internet Archive, Pushshift, etc) out there.

(Apollo user with an interest in a distributed web platform)

When it was just text, Reddit was cheap enough to run that it was kind of questionable why they needed so much VC funding.

> If Apollo is the interface, what is Reddit besides a JSON document store and object storage for media?

A community, or if you prefer, a community of communities. That part is essential.

Yeah, agree. I'm arguing for disintermediation of the community from the corporate entity attempting to squeeze as much value out of the community as possible (and in the process, being a hot potato from one company to the next). Communities deserve better (Stackoverflow and Wikipedia come to mind), and the technology primitives are stupid simple. Perhaps I've been reading too much Neal Stephenson and Cory Doctorow lately!

Philosophically? Sure. But with how the masses act and behave and take interest in, getting the critical mass of non-technical people you would need to make it work is a massive hurdle.

Facebook, Twitter, YouTube and Reddit will skate by with inertia for years to come no matter how suicidally they behave because they have mass appeal. Losing one audience doesn’t matter because they have millions of other audiences. You start up a social network that interests mainly technical people and the journalists that report on them, well it’s just not going to thrive because after the honeymoon period is over, the network just starts to look stale and slowly die off. E.g. Google+.

I concede these are all good points and I genuinely appreciate the discourse. Speaking only for myself, I see the challenges as:

1. Preserve the content and culture of these platforms for users and future historians (in the case of Reddit, Pushshift and the Internet Archive). UGC ("user generated content") belongs to users first and foremost.

2. Continue to find opportunities to reduce the moats of incumbents and the friction of standing up alternatives. If a site disappears, there should be little impact to anyone other than employees and shareholders.

3. Be ready to jump in and scale up when incumbents misstep (Signal and Telegram seeing huge membership jumps during Facebook's extended outage earlier this week, Digg's redesign helping Reddit's growth, etc).

That’s great but that’s an archival project, not a social network. There’s room for smaller networks to exist on the fringes but we’re unlikely to see a Reddit that looks like Reddit overtake Reddit on its own merits. The same for Facebook, or TikTok or Twitter.

I wonder if Lemmy instances are similar enough that the app could just be pointed toward them.

They should just hire the guy. I wouldn’t mind Apollo with ads from Reddit official.

Well Twitter on iOS started life as Tweetie which was an absolutely fantastic client on both iPhones and Macintoshes. I wouldn’t call it that now, even if some of the original Tweetie code is in there. Apollo with ads doesn’t seem like it is in Reddit’s business plan and after some time I think Apollo would go the way Tweetie did.

Edit: Something from the dredges of my memory just came back up and reminded me of something. The official Reddit client did start life as a 3rd party client: Alien Blue. [1]

So the real risk is Apollo becoming the next Alien Blue. Personally I would rather it stay independent.

[1] https://appleinsider.com/articles/14/10/16/reddit-buys-unoff...

The official Reddit iOS app is already the result of Reddit buying an existing, excellent third-party app (Alien Blue), which they proceeded to ruin.

I have no issue with the official Reddit app and I can use Reddit just fine.

> Apollo is not the defacto reddit client. Reddit's own app is.

Reddit's is the official client.

> Defacto: in fact, or in effect, whether by right or not

I think in this case the validity of the word "defacto" depends on your sample population. Apollo is very popular (and a fantastic reddit client, I wish there was an equivalent for HN) but I can't imagine it's marketshare is anywhere close to even 50%. Among tech enthusiasts I'll bet it ranks pretty high. Similar to an app like Overcast. If you poll HN or other tech communities I bet it's well represented but in the broader ecosystem it's a small fish.

Reddit pushes the official client hard from the mobile website. I'm sure its marketshare is huge.

100% agree. Thanks to Safari extensions (and the afore mentioned AMP redirector) you can now jump right into Apollo from a Reddit link and it's the best experience.

I hope the marketshare of the official client is at the very least high enough to prevent them from crippling APIs so third-party clients are gimped, like Twitter has done.

God forbid they add a token limit

I don't care too much about AMP links but I bought Amplosion just to show my support for Christian, the author of Apollo. That app alone makes Reddit bearable. The official Reddit app is terrible and actively harms the browsing experience and makes it harder to find actual useful content on Reddit as opposed to all the pushed spam content.

For anyone on android, "Reddit is fun" is hands down the best I've used.

I seriously don't see how people put up with the main app + non old.reddit.com site. It's just miserable and ad filled.

Soon they won't have a choice as old.reddit.com is already set for deprecation. The whole site is an advertising platform now.

I actually generally like AMP, but I paid for the extension because AMP breaks native app URL listeners, so from an AMP page it's impossible to open the page in the native iOS app e.g. YouTube or Reddit. (FYI, free AMP blocker extensions quickly emerged after that paid one went viral.)

People who

- use iOS 15 (including betas)

- hate AMP

- still use Google for search

- went to the trouble of finding and installing such an extension

is probably a number that rounds to 0.0% of Google searches.

Not meant as a critique of the extension writer, just agree that this explanation seems highly implausible to me.

I agree – the number is probably insignificant, but I don't think you have to restrict your estimate to people who hate AMP. Basically an average iOS user (likely using Google as it's still the default last I checked) who in the past installed an extension to block unwanted content (ads/malware/etc) may be blocking AMP links without even knowing what they are.

How many normal users have even heard of AMP? Even among those technical enough to know or care about AMP, how many have installed and use these plugins? I haven’t.

I don’t see how blockers installed by a small percentage of techies would have much impact.

Google launched "Core Web Vitals" a while back as their "open" alternative to the requirement to provide AMP versions of pages. It's possible they're running some kind of A/B test with AMP vs CWV calculation and filtering based on iOS version.

Maybe an odd thing to do but IMO no less odd than giving up on AMP in iOS 15 because of browser extensions, I can't believe the number of people using AMP-blocking extensions is that high yet.

>I can't believe the number of people using AMP-blocking extensions is that high yet.

by what metric? I think it being high on the app charts is more indicative of that category being deserted and therefore easy to be on top, than everybody and their mom downloading it. A few days ago I saw a post mentioning that Amplosion was #1 on the "utilities" chart. Now it's #27, and the top apps are 1) a proxy manager 2) calculator 3) dark mode extension for safari.

Right: I mean in absolute numbers of Safari users. Even something at the #1 spot in Utilities for a few days is (in my assumption) still barely a dent.

My suspicion, having worked at google for a while, is that this is a parser / version eligibility bug. Someone probably put the list of major versions where this is supported, and 15 was not there :)

Things like this happened every new ios release. Going from 9->10 was a problem for example because regex used was limiting major version to single digit etc.

Also AMP it's disabled for Firefox on Mobile (Android)

There could also be a newly discovered bug or change in Safari 15 that affects AMP pages.

Except AMP links were showing up in iOS 15 in the beta and for the first few days after official launch.

I think that Apple Relay is a more probable cause

Which betrays AMP’s purpose as tracking.

How does apple relay affect AMP?

Apple uses their own IP address when accessing sites to protect privacy: https://support.apple.com/en-us/HT212614

Maybe the majority of iOS 15 traffic coming from Apple servers instead of individuals is causing trouble with Google's current AMP implementation somehow, so they check the iOS 15 user agent and disable it?

IMO more likely Google just has a user agent whitelist or regex that misses iOS 15 and AMP was just showing up initially while Apple was sending an old user agent during the iOS 15 betas or something.

Majority of Private Relay traffic comes from Cloudflare and Fastly. That's one of the relay's privacy features: the server don't know where the client comes from.

I agree that this is probably more likely.

I do not think so. I have tried to switch off relay and still it does not show AMP on Safari browser for me.

It's a communication setup, which does not affect browser agent string, which in turn can be manipulated to allow access to AMP over the same connection.

If the author's theory is correct, I guess Google's worry is that if they annoy users with AMP they might come looking for extensions and install not just AMP blockers but also ad/annoyance blockers.

It is a bug, and Google is unfortunately working on a fix

https://twitter.com/dannysullivan/status/1445799446139191298...

Side note:

Does anyone have a decent link that explains exactly what AMP is to someone non technical? (Bonus if it explains why its bad, but I think I was able to at least get that part across)

I have been trying to explain multiple times but I am continuing to realize that there is certain terms that I just "know" and don't really know how to translate.

Techcrunch usually has this type of articles: https://techcrunch.com/2018/02/13/amp-for-email-is-a-terribl...

Since this is site is usually read by C-levels and some VC's, the language is less technical but somewhat editorialized.

Anyway AMP is bad because it proposes an Internet without the freedom to push dynamic content (JavaScript, webassembly, etc...), cached at "google servers" for fast delivery but with features specifically built so Google Analytics and others can keep working, it doesn't solve anything that a good css style wouldn't solve plus content creators lose control of their data.

TBH this is how I summarized AMP being bad, but some are curious more exactly what it is:

"Google tried to push something that gives Google more control, Google used the control Google already had to strongarm websites to start using it or their search results would go farther down. "

Thank you for the link, will check it out and forward it.

This is not a good explanation. AMP4Email is different from AMP and serves a different purpose. https://medium.com/@pbakaus/why-amp-caches-exist-cd7938da245... is an explanation of what AMP does.

That is exactly the type of example/article/blog OP didn't wanted, too technical, TC article is about AMP for email but first its describes what AMP is and does, why is bad and why google is pushing AMP for literally everything it can. It's less technical more down to in tone for less technical savvy readers.

You can't understand any of those things unless you understand what AMP does, and (despite what you claim) the TC article doesn't explain what AMP does. Not understanding what AMP does leads you to make nonsensical statements like that AMP can be replaced with a CSS stylesheet, so users continue to prefer AMP because there remains no better alternative for doing what it does.

You are looking at the problem from a developer perspective, not from a user/reader or content creator, your article just gives tops 1 paragraph to the issue at start talking about syntactic validation and micro-optimizations that the user/reader doesn't care. The point is we don't need AMP, we can deal with caches and the bullshit micro-optimizations ourself, ppl don't care if the page loads instantly or in 4s-6s, despite any claim Google does.

On the contrary, the user cares that the article loads instantly, and my article explains how AMP achieves that. Since you do not understand the problem that AMP solves, you think you can solve the same problem with micro-optimizations, but micro-optimizations can't achieve instant loading.

It is not just Google (and Microsoft and the other companies that implement AMP caches) that believes users want instant loading. Apple and Facebook have products competing with AMP that give users instant loading but that are far more restrictive on the publisher and on the reader. The only reason that publishers and readers put up with that is instant loading.

Update in the post says a Google PR person said it's a bug

I'm an AMP noob but why would someone block AMP pages?

Because they want to visit the actual website and not Google's scraped version of the website delivered on Google's infrastructure.

My experience on iOS: scrolling feels wrong because it is non-native (different inertia and velocity), reader mode stops working, and if for whatever reason I want to share original link that won't be easy.

Because the AMP pages are often functionally crippled compared to the real site and require clicks to get off and on to the real web page (see Reddit, The Guardian etc), as well as the negatives of giving Google that level of control.

By this logic safari content blockers enable ad blocking, so why hasn't google exited the advertising business yet?

TIL there are AMP blockers on Safari...brb

I moved to duckduckgo because of how AMP breaks moat sites on iOS .. I too would like one of those AMP blockers

I don't use google search but every once in a while I get sent or stumble upon an amp link - so even for those limited times a $4 extension is a no-brainer.

I did the same thing, that is the quickest few bucks I have spent on the App Store ever (and then shared to all of my friends).

Did not even need to think about it, I have wanted this since APM first started being rolled out.

yeah, I didn’t even think twice about dropping $4 on it

Just because you drop the "?" from the end of the headline doesn't mean it's not still subject to Betteridge's Law.

Reply on news.ycombinator.com