back to article We give up, Progressive Web Apps can track you, says W3C: After 5 years, it decides privacy is too much bother

In 2015, as part of a privacy review conducted under the auspices of the World Wide Web Consortium (W3C), Nick Doty flagged a potential problem with web applications. This week, after five years of debate about whether or how to mitigate this privacy concern, the technical types discussing the matter have simply given up and …

  1. IGotOut Silver badge
    Thumb Down

    Looks likes...

    Googles bullying / bribery won yet again.

    1. Anonymous Coward
      Anonymous Coward

      Re: Looks likes...

      Is ‘Google’ the universal answer for the lazy of mind?

      The doc/spec for the web app manifest was started by "Marcos Caceres, Mozilla Corporation" in 2013. Worked on in parallel since then with "Mounir Lamouri, Google Inc." and plenty of others. This isn't the first time a technology was made useful without considering security. Note the several commentatoreros saying they've turned off JS in browser.

      Words for cheap like 'radical', 'leftist', 'extremist', "Google is tha eevihl!" are not the height of insight, nor informative.

      1. Dan 55 Silver badge
        Stop

        Re: Looks likes...

        Well, it's true.

        Google takes sole stand on privacy, rejects new rules for fear of 'authoritarian' review

        Google's vote nixed the W3C privacy group after members rightly stated that Google's so-called "privacy sandbox" was a load of old nonsense. In Google's objection they gaslighted all a bit suggesting that with no "authoritarian" review body 1000 flowers can bloom, but they know Audrey II (Chrome) can come along and kill them all off.

    2. Warm Braw

      Re: Looks likes...

      Actually, it looks like the way to stop people being tracked is to have laws that effectively protect privacy.

      In any communication, you need to identify the end points and it would be an extraordinary technical feat to prevent that necessary identification being exploited in any circumstance.

      If you're concerned about evil, money-grubbing corporations, you can't really expect commerce to be the source of a technical solution.

      1. Charles 9

        Re: Looks likes...

        "If you're concerned about evil, money-grubbing corporations, you can't really expect commerce to be the source of a technical solution."

        But what if that's your only option, given commerce has enough clout to sway the government?

  2. Anonymous Coward
    Anonymous Coward

    Probably not insoluable, just not trying very hard.

    The flaw in there assertion is, I think, that the SITE should be able to specify the Start URL. Allowing(which is to say forcing) the user to edit this is not a good solution, but they were on the right track with the OS/Browser mitigations.

    Off the cuff let the SITE request the os or browser build the URL from a limited(and not unique) set of options, and a (non-unique) base domain url. Of course if you leave it to the browser then Chrome will probably allow at least themselves to track you, and Microsoft will re-badge it and point it back at their own tracking server.

  3. luminous

    Why do you need to allow question marks in the start URL? Maybe a lot of the top sites use them, but I don't bookmark URLs with identifiers in them. I just bookmark the main URL and then any redirection or identifiers are added after I visit domain.com

    1. Pier Reviewer

      Banning query parameters (the stuff after the “?”) doesn’t fix the problem sadly. A site operator could provide a start_url value like https://pwa.acme.com/start/123456 where 123456 is a unique number for each user. Voila, you can be uniquely identified.

      Whilst the article appears to bemoan the fact that the issue is being left to browser makers to fix, I don’t see a reasonable and effective alternative. The solution needs to be where it’s most effective. Letting site operators like Facebook et al decide how to prevent tracking is laughable. The problem of course is that Google intentionally broke the Chinese wall by releasing a browser. The only practical solution to avoiding tracking in this instance is to have the browser makers fix the problem and keep away from Chrome.

      1. Sampler

        Then take the spirit of the recommendation and not the letter, why allow anything after the tld?

        Now, yes, a site could spin up unique subdomains, but, that's going to be harder work and would probably put most off.

        Could not even allow sub-domains, you want a PWA, it'll have to run from the main domain.

        That said, sounds a lot like a non-issue, if you like a site enough to use a PWA and add a shortcut to your homescreen you're probably going to want to be logged in to it anyway (like a weather app so you have your location stored, or twitter so you're logged in etc..).

  4. Anonymous Coward
    Boffin

    Optional

    Only let PWAs start from the base domain URI.

    Allowed: pwa.thing.com

    Disallowed: pwa.thing.com/something

    Disallowed: pwa.thing.com/?something=something

    There. Solved.

    1. Anonymous Coward
      Anonymous Coward

      Re: Optional

      Then what happens when (not if) the PII is in the PWA?

    2. aelking

      Re: Optional

      Welcome to

      Http://123456.pwa.thing.com

      1. Dan 55 Silver badge

        Re: Optional

        Browsers already know that abc.com is a domain under the .com TLD and abc.co.uk is a domain under the .co.uk TLD. If you don't believe me check the address bar and see how the something in something.abc.com and something.abc.co.uk are either in gray or hidden.

        Therefore it could enforce this so that something.abc.com and something.abc.co.uk aren't allowed, as well as banning parameter passing using ?.

        There, not too difficult. Perhaps the issue should be reopened on github again.

        1. Anonymous Coward
          Anonymous Coward

          Re: Optional

          I can't tell whether you're being sarcastic or not. All you've achieved is to (yet again) push the tracking to a different part of the URL - presenting: tracker-12345.com

          Yeah, it'll cost me $5/year or whatever. Assuming I can't get free domains*. So all I need to do is add $20 to the price when I'm selling all your data and then cancel the domain after 3 years of inactivity.

          Also blocking subdomains would cause a bunch of headaches. Having to register a new domain for my-app.com rather than using myapp.mycompany.com shouldn't be necessary and would be a pain for all the little guys out there.

          The fact that there's money involved now means that this scheme would disproportionately affect individuals / small / open source projects while leaving the huge corporates (who are the ones doing the most tracking and slurping the most data) almost unaffected.

          *if I'm large enough, I'll just buy a TLD (or use the one i already have, *cough* .google *cough* .amazon) and not pay that $5/year/user in the first place.

        2. Charles 9

          Re: Optional

          Have you looked at the current URL? Sure, the "forums" part is grayed out, but it's still significant, given putting that in puts you in a different part of El Reg versus the front page. Meaning you can't ban the "forums" part, for example, because there's too much risk for collateral damage, to the point you could end up blocking or banning the app itself.

          1. brotherelf

            Re: Optional

            As if "there'll be collateral damage" ever stopped a browser vendor. Basic auth in iframes, not showing www/m prefix even if what's shown has no A/AAAA record, the default changing from rel="opener" to rel="noopener", abandoning the CA/B forum, the list goes on and on. We are rapidly heading back towards "best viewed in X/Y/Z because Y/Z/X is broken".

            1. Charles 9

              Re: Optional

              "We are rapidly heading back towards "best viewed in X/Y/Z because Y/Z/X is broken"

              Such as it is. Are you gonna just throw up your hands and say, "Stop the Internet! I wanna get off!" Because it's not like we have any real say in the matter anymore.

    3. Anonymous Coward
      Anonymous Coward

      Re: Optional

      You can also encode identifiers in the hostname, e.g., something.thing.com, or 12345.thing.com. So that doesn’t help much either.

    4. Blane Bramble

      Re: Optional

      Which breaks things such as:

      www.mysite.com/my-pwa-app1

      www.mysite.com/my-pwa-app2

      etc.

    5. Anonymous Coward
      Anonymous Coward

      Re: Optional

      Also allowed: 354c9a65-1139-4f87-adac-5d4fcfd8783c.myapp.example.com

    6. Brewster's Angle Grinder Silver badge
      Facepalm

      Re: Optional

      Allowed: user1234.thing.com

  5. This post has been deleted by its author

  6. phillipao

    Isn't this something that native apps can do trivially, by generating a uid and storing it locally? Or is the issue that a server generates the url, and it has access to other info? I'm curious about that isolation mechanism you mentioned in webkit.

    1. Anonymous Coward
      Anonymous Coward

      When they get to the point where they claim bookmarks are tracking you, you have to step back and wonder what the beef is with Apple WebKit developers trying to throw shade on webapps.

      Why don't they just say: "we don't support that because we want to force you to use the AppStore"? It's not like the AppStore isn't tracking your AppleID too...

  7. Will Godfrey Silver badge
    Unhappy

    Poison the well

    It's the only thing you can do, and it would have to be done by piggybacking the mouse and keyboard, so the browser has no way of knowing you didn't enter the requests manually.

  8. cd

    Could a browser extension salt/scramble the id on the way out?

  9. Paper

    Just remove the manifest too?

    When the user goes to "Clear browsing data", just delete the manifest file, or start_url at the least, too... Next time they visit redownload. Problem solved. User may need to have an internet connection for the initial revisit.

  10. Anonymous Coward
    Anonymous Coward

    WarW3C, what is it good for?

    1. Charles 9
      FAIL

      Absolutely NOTHING!

      And they know it, too, because they now possess too much power for any one authority to control. This is the "Sprawl" solution to government oversight: become too big for governments to oversee. Now they can just dodge jurisdictions and so on to avoid regulation. Anyone gets too uppity, use your leverage to take them over.

      If sites want to track you, they have the ability to do so completely server-side, meaning you're in a Hobson's Choice, which may not be a choice at all if your job (the only one at hand in this mess) requires your use of this privacy leech.

  11. Anonymous Coward
    Anonymous Coward

    I wonder at what point I became a Luddite?

    Can't think how I got along all these years without PWA's. Or mobile phone apps, come to think of it. Or anti-social media sites </sarcasm> Yeah, I know, just because I've never felt any inclination to use such doesn't mean that others don't find 'em useful, and good luck to them! But I feel sad that my one-time wonder at the fact that markup-languages made things like web pages possible has changed over the years to a combination of frustration and anger at the multiple invasive assaults on privacy that have resulted since. The future just isn't what it used to be! Sigh. OK, here's me shuffling back to my flat in the local home for mildly bewildered old biddies (Flat 1980, Vulture Eyrie Tower, Sodoff Street, Birminster, Barchestershire, UK)

    1. Will Godfrey Silver badge

      Re: I wonder at what point I became a Luddite?

      Got a spare room by any chance?

  12. Ozzard
    Big Brother

    Fixable with browser extension

    "Trivial"* fix with a browser extension** from someone like EFF, where the extension sends the start_url (with no other user information) to a central server. If the server sees a high proportion of unique start_urls for a particular second-level or top-level domain, it responds that there may be fingerprinting going on, and the extension can warn the user.

    * Relies on user installing extension, but the same is true of pretty much anything of this kind, including Privacy Badger.

    ** Assumes a browser extension can examine (but not manipulate) such requests. If not, perhaps the browser manufacturers should consider that extension point.

    1. Charles 9

      Re: Fixable with browser extension

      The problem is if the start_url itself contains the user information, such as it was before stateful web features like cookies appeared. Plus this does nothing for when the fingerprinting takes place IN the PWA.

  13. Danny Boyd

    Strangely, anything called "Progressive"...

    ... means " For the greater good of mankind, and fuck all those individuals that get caught in the crossfire."

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like