AMP is not the Open Web

May 1, 2018 10:57 AM

Does Metafilter have a policy about Google AMP urls in FPPs like this one? Might we consider stripping them out? Or do the benefits (mobile friendly speed, primarily) outweigh the possible costs (Google controls more of the web)
posted by gwint to Etiquette/Policy at 10:57 AM (25 comments total)

We don't have a policy forbidding them. But we're happy to replace these as people notice them, since it doesn't seem like they're functionally that different from the regular site. We'll typically do that for mobile-specific links too if people point them out.

As for auto-stripping, probably it's a default no, since we don't do that with most things and it's not clear this case is so different to require a special solution.
posted by LobsterMitten (staff) at 11:02 AM on May 1, 2018 [3 favorites]

Oops. Feel free to remove the AMP on that one. Usually I take them out myself. URL: https://www.texasobserver.org/red-state-socialism-texas/
posted by Artw at 11:05 AM on May 1, 2018

(Note that it’s still the same site, just not an AMP/mobile ready view. )
posted by Artw at 11:06 AM on May 1, 2018

Edited that.
posted by LobsterMitten (staff) at 11:07 AM on May 1, 2018 [1 favorite]

Previous AMP discussion.
posted by Mitheral at 11:30 AM on May 1, 2018

As a mobile user, I find Amp links annoying. I dislike the ugly disappearing-reappearing banner that they put at the top of the page. I know they're supposed to improve my mobile browsing experience, but in my opinion they actually worsen it. Also, it's not clear to me what happens if Google someday ends the Amp service; do all those links then suddenly become broken? If so, that seems like reason enough to strip out the redirects.
posted by Anticipation Of A New Lover's Arrival, The at 11:33 AM on May 1, 2018 [14 favorites]

BURN THEM

I hate AMP on my phone, and I darn sure don't want them on my laptop.
posted by wenestvedt at 1:14 PM on May 1, 2018 [2 favorites]

Despite the sales pitch I find them slower in practice than the links they replace. Also they're creepy tracker bullshit. +1 burn them.
posted by enn at 2:23 PM on May 1, 2018 [5 favorites]

++ on the burnination. FLAMP the AMPs!
posted by rodlymight at 4:22 PM on May 1, 2018 [3 favorites]

What is the case for not stripping Amp redirects, other than that we normally don't do that? We do alter Amazon links to generate affiliate money, so messing with user-posted URLs isn't totally without precedent.
posted by Anticipation Of A New Lover's Arrival, The at 4:41 PM on May 1, 2018 [9 favorites]

I imagine that it's not a straight forward process to un-AMP links, unlike inserting MetaFilter affil codes into Amazon links. If I recall correctly, some sites just add AMP to parts of the URL in question, while others make shorter AMP links to content otherwise available with a longer URL.
posted by filthy light thief at 7:07 PM on May 1, 2018

Un-AMPing links isn't a plain text transformation. There's a standard way to find the non-AMP version from an AMP page, but it would take fetching and parsing the page, which is extra work that Metafilter doesn't currently do. (If you're curious: <link rel="canonical" href="url" /> in the header of an AMP page should point to the plain version.)
posted by skymt at 9:18 PM on May 1, 2018 [1 favorite]

The case against stripping AMP is increased load on the site's servers. Maybe some SEO.
posted by rhizome at 9:45 PM on May 1, 2018

What is the case for not stripping Amp redirects, other than that we normally don't do that?

Primary case for me is the work required to get that stripping working sufficiently flawlessly to not bug up anybody's post in the process of fixing it. Which is the case basically for why we're not doing it now; I don't have an ideological problem with the idea, we just have limited resources and that's not high on my list when it's something we can handle manually fairly quickly. It might be something we end up doing eventually; frimble's got other stuff on their plate right now.

odinsdream is correct that it would be a dev issue, not a server load issue. (Except for the load on other people's servers in directing folks to their core site rather than the AMP cache, but whatever else there is to say about it the intent of AMP isn't to get away from actually hosting content and MetaFilter isn't gonna annihilate a modern hosting setup even on a busy day.) But I'm also not sure if rhizome is talking about us stripping out external AMP links in posts, which I take to be the thrust of the post here, or the idea of us not serving AMP links ourselves, which is a different discussion entirely.
posted by cortex (staff) at 7:42 AM on May 2, 2018 [1 favorite]

Yeah, I imagine that unless this starts to occur with much greater frequency, and we're agreed as a community that we'd prefer to avoid AMP links, just having the understanding that "In General Please Use The Canonical URL"* is MeFi policy works for me, and there's no need for frimble to descend into the darker regions of regex.

*which I think came up back during the Great URL Shortening Services Scourge of 2016
posted by gwint at 8:12 AM on May 2, 2018 [3 favorites]

It’s uncertain to me that striping “/amp/“ from the end will always result in the non-amp/mobile site. Much as how when people used to put “m.whatever.com/whatever” as a mobile url “www.whatever.com/whatever” would mostly give you the same page, but not always.

As for the full on AMP urls that start with google.com, might be an argument for flat out booing then but it seems like they would be torn down so quickly there wouldn’t really be a point.
posted by Artw at 8:26 AM on May 2, 2018 [1 favorite]

Burninating the countryside,
Burninating the peasants,
Burninating all the peoples,
And their thatched-roof COTTAGES!
THATCHED-ROOF COTTAGES!
posted by terrapin at 8:29 AM on May 2, 2018 [3 favorites]

I hate AMP with a passion.
posted by cjorgensen at 8:58 AM on May 2, 2018

Except for the load on other people's servers in directing folks to their core site rather than the AMP cache

This is what I meant. The story's link would go directly to the site hosting it rather than Google's AMP servers.
posted by rhizome at 10:09 AM on May 2, 2018

I prefer AMP pages. I think the choice should be left to the Original Posters.
posted by save alive nothing that breatheth at 4:29 PM on May 2, 2018

AMP works if an organization opts-in. It seems that if an organization wants to use AMP -in part to reduce the load on their own servers- we should be generally fine with that choice. Let the content provider provide how it wants.
posted by Going To Maine at 11:11 PM on May 2, 2018

Let the content provider provide how it wants.

Counterpoint.

Kidding, but still. I have no opinion about AMP but sometimes leaving things in the hands of the content provider can be horrible for the end user. Hence the rise in ad blockers.
posted by zarq at 8:53 AM on May 3, 2018

AMP breaks find-in-page on Safari in iOS (on every site, not just MetaFilter), which is pretty annoying.
posted by invitapriore at 4:38 PM on May 3, 2018 [3 favorites]

I hate AMP with a passion; in my experience it's a whole lot slower, possibly because of geography. Stripping it out might not be a good idea, though, because:
(a) there's some risk of directing people to URLs that aren't accessible from a site's front page, and thereby running afoul of anti-hacking laws; and
(b) it might harm the site's revenue stream from Google if they optimise ad revenue based on the way sites are accessed;
(c) an overt acts like this are sometimes held up as an example of Fighting Back Against the Man, which invites retaliation. I'm all for defending web standards, but this doesn't seem like a good hill to die on.
posted by Joe in Australia at 11:58 PM on May 3, 2018

For Firefox users the Redirect AMP to HTML extension prevents spreading AMP links.

Extension users aren't protected from the "creepy tracker bullshit" themselves because the redirect only happens after loading the AMP page but at least they won't accidentally bookmark, share or make FPPs with AMP links.
posted by Bangaioh at 2:36 PM on May 18, 2018 [1 favorite]