The <Geolocation> HTML Element | Hacker News Reader

The <Geolocation> HTML Element

developer.chrome.com

125 pointsenz62 comments

This might be easier than refusing permission every time - it sounds like I can just not click it. I really dislike location permission things. I don't know what location will be shared, I don't use anything that needs a precise location, and I don't ever want to share my actual location. If location permission things showed me a map with where they think I am and let me click a (vague) location to share, I might use them, but currently to find nearest stores or whatever I just type in a postcode or use their map.

Edit: this has prompted me to go find a way to turn off location permission requests in the browser settings. It turns out you can do it under Privacy and Security > Site Settings in Firefox and Chrome.

> easier than refusing permission every time

Most browsers allow setting default permissions for all sites at once.

Not default permissions, it turns out, but you have a global choice between letting sites ask for permission or blocking the requests entirely.

Safari lets you differentiate. I can set all sites to deny by default, configure Home Depot to always ask for permission (resets every 24 hours), and configure still others to always allow.

This element has an autolocate attribute that will request permission automatically, plus it doesn't supersede the JS api, it simply provides a declarative alternative to it, so sites that follow this negative pattern will keep doing so.

At the same time, there is no reason to not implement this pattern today and require user intent prior to requesting the permission

According to the post, autolocate only does something after a user initiated permission has been granted.

So on the first vist you still need to click the button. On the second visit the callback will be triggered directly.

But, well, nothing prevents a big fat html modal on the page pointing to the button, now does it? If you want to annoy your product^H^H^H^H^H^H^Husers then you can always find ways to do so.

Me too. I don't share precise location but will happily click on a map if a website gives me the option. For things that i really need, i just use dev tools to manually set the location :)

This seems pretty sketchy, and I don't really understand what prevents a website from clickjacking.

The original flow is awkward, but also renders the permission element in a location that can't be clickjacked, thus offering some protection from geolocation.

It still pops up a permission confirmation dialog

It explicitly pops up the confirmation dialog even though user has previously DENIED that geolocation permission dialog. The whole thing is only created because they see "high denial rates" and want to create a way to permanently undo these denials.

The spec doesn't mandate such behaviour, that's up to the user agent to implement.

I've had to deal with plenty of people who couldn't do things like use Jitsi or other web apps because they missed or denied the permission prompt before reading them. The tiny icons in the address bar are barely recognised as clickable items by most users, which is a good thing for toning down annoyances but an awful inconvenience when trying to help people.

In a few cases, the solution to "accidentally dismissed permission popup" was "make everyone else download an app full of trackers".

Jitsi is about audio / video where this would make more sense, not about geolocation.

Geolocation based on IP address is always done in the background, so they already know what city you are from. But Google wants to have the nice high-precision location from our GPS chips so they can permanently associate the IP address, and available WIFIs/Bluetooth/network devices and all related MAC addresses to a specific building.

And they want to have this specific functionality so they can organically trick non-power-users who got accustomed to the permission popup dialogs into re-sharing their location.

The audio/video version is in the works [1]. I imagine geolocation just got to this stage first because it's far simpler.

And IP-based geolocation is really unreliable beyond the country level. It depends on ISPs using a different pool of IP addresses for each city. That seems to be the norm in the US, but is not how every ISP runs their operation

1: https://developer.chrome.com/origintrials/#/view_trial/37362...

Oh come on please stop this insincere false narrative. IP-to-geolocation works extremely well even with semi-public databases, and Google has spent billions on collecting a very powerful database of it.

Google is internally fusing data from all devices that were ever in your LAN or near your WIFI access point. It merges them based on:

  - outside IP address given by ISP and mobile phone carriers
  - list of network devices incl. IP address & MACs (do you have a 3d printer or not?)
  - any info they can extract via their Smartphone Apps both on Android and IOS
  - Android devices share geolocation data with google anyways
  - user agents & browsing history on google properties and 3rd party websites
  - every time your IP addresses hit anything in google cloud or google CDNs (e.g jquery from googlestatic.com or google fonts)
  - all data you have ever provided to them (payment data, gcloud, gmail)
  - all the tracking that is already built into chrome and other google software

Plus they procure data from third-party providers. Only a single app on your smartphone needs to have geolocation privilege and then the data (location, ip address and user data) is available for google to digest via their scripts/SDKs which are packaged into basically every smartphone app.

The large tech companies have significant incentive to mutually share data with each other, that's why you often see Javascript from one tech company included in the website or product of another one. It's enough to touch a google dc with a single packet included in the facebook app for them to associate your session with the new IP address and vice versa.

Google has years of data on how often user agents and devices behind a certain IP address change. They can very confidently say if your ISP-provided IP address has been rotated or not, and where it was rotated to. They most definitely have enough GPS positions from smartphones that they can predict where you sleep, so if your smartphone shows up under another IP address but the network devices around it stay the same they can easily deduce that this is your new dial-up IP address.

All of this not even discusses unethical or clearly illegal ways these companies have acquired data by abusing a lack of security measures on the smartphone operating systems. An example is Facebook uploading entire smartphone contact books to their servers to fuel their "organic" growth - Google most definitely has done exactly the same.

The "careless people" book highlights that Facebook deployed spyware with their smartphone apps which monitored what other apps the users were using - this is how they figured out that WhatsApp was going viral and based on this data they did the surprise acquisition of WhatsApp. I'm confident that google abuses the same security vulnerabilities in order to further collect data.

Maybe I’m missing something, but Google does not offer an IP Geolocation service or publish their database.

You are incorrect. Google offers IP geolocation service at https://developers.google.com/maps/documentation/geolocation...

Of course they don't share the final database, because it is their core asset. And if the common public catches a glimpse of the data that google has saved they would be really upset.

That is not IP Geolocation.

One of the first times I used a Google product to show me my location (over a decade ago), it showed me a zoomed-in satellite picture of the parking lot right next to my apartment.

(It didn't take me long to realize the implications.)

> Jitsi is about audio / video where this would make more sense, not about geolocation.

It's still permission-based. And the article mentions that the same ux is being done for media objects.

> Geolocation based on IP address is always done in the background, so they already know what city you are from

They don't. Maybe the trackers do, but most websites place me 100km to the west. When I want to look up opening times for supermarkets near me, I like the "show nearby" button.

Even Google gets confused because I VPN back to my home network. Every time I spend a few days somewhere else, Google's IP-based geolocation is broken again.

Even though the other browser vendors have positive reaction, note how this follows the same pattern Chrome has followed for over a decade:

- scribble on a napkin (explainer)

- ask others for their position

- ship regardless of position or any outstanding issues

- claim it's a new standard

I'm not familiar with how this process went in this case, the blog post seems to suggest that feedback from other vendors was incorporated. Can you elaborate on the outstanding issues?

If you follow the links to browser positions, there are still discussions about various concerns that don't seem to have been addressed.

Exactly, as sites start using this new tag users will see this breaking in the other browsers who will then be "pushed" into the Google way.

The line about being open to feedback while they're encouraging sites to start building using this really feels disingenuous. At least they could own their choice that they want to go it alone on this.

we're planning to push ahead with our implementation in Chromium, continuing to iterate with your feedback in mind.

https://github.com/WebKit/standards-positions/issues/545#iss...

I'm a bit confused about how it actually works, and somehow they decided to not include a demonstration video.

If clicking on it does trigger a location permission prompt: what's the point? The "issues" with prompts getting denied can already be solved by web developers doing this themselves, rather than just blindly firing off a request on page load.

If clicking on it does not trigger a location permission prompt: have we forgotten about the Line Of Death [0]? Clicking random website-styled elements should never result in dangerous actions being taken - and leaking the user's physical location is definitely dangerous. Sure, they are trying to restrict the styling, but that's a fools' errant: somebody will just make a browser game where the button looks to refer to something ingame, but actually leak your real-world location.

Besides, who's actually asking for this? Location is perhaps useful for Google Maps-like websites to save you a few seconds of scrolling, but in practice it has mostly been spammy websites trying to get me to "subscribe to local news". Making geolocation easier is the last thing I want in my browser!

[0]: https://textslashplain.com/2017/01/14/the-line-of-death/

> The "issues" with prompts getting denied can already be solved by web developers doing this themselves

Does that mean identifying the browser and trying to tell the user how to go into the browser settings and un-block permission prompts?

No, I mean adding a "use your location" button yourself which the user has to click before it uses the geolocation API, rather than just blindly requesting it on page load.

The only reason people block it in settings is because they get sick of nagging prompts they never asked for.

Ah, gotcha. So this change is giving developers a more standardised way to follow that "add a button, pop up permission dialog" pattern that will hopefully drive more of them away from the bad pattern?

You get prompted to supply your location just like normal.

Using geolocation on the web is not something I do daily, but I do use it every now and then. The "locate stores near me" button for looking up store closing times is a lot easier than manually panning across a map.

I find Chrome's current implementation (on Android) to be acceptable as long as measures are taken to prevent clickjacking and such to automate repeating prompts after denying permissions. I expect other browsers like Firefox to be more conservative in showing popups like that.

Presumably at some point they'll turn off script-triggered location access or make it less undesirable for sites in some way.

The main intent behind this seems to be constantly reminding users that enabling geolocation is an option, and that it is an ordinary thing to do. Dear user, why don't you still use it, everyone does it. Just click the button!

Yes, because users don't like when things don't work, and they don't think about that time they blindly dismissed a permission notification and don't understand how to go undo that easily.

So we make it easier for the user to actually do the thing they intend to do. Seems good to me.

Contextual permissions are a big improvement over early and uncertain prompts. I will never agree to grant my permission when first loading a page, however, I may do so if intentionally activating a map widget. At least then I understand the context by which it's being asked, and can make a more informed decision.

I'm curious to why the polyfill example uses unpkg.com. It is quite unreliable and has broken sites many times.

jsdelivr.com is much more reliable (Multi-CDN, Multi-DNS). Comparison: https://www.jsdelivr.com/unpkg

I am not affiliated in anyway to jsdeliver or unpkg. I simply used to be a user on unpkg.

Maybe for tracking purposes, because most google-affiliated CDNs are widely blocked and unpkg might be small enough to not immediately raise eyebrows with devs. Unpkg removed their SPONSORS.md file recently, and their README claims to be hosted at fly.io which seems to be behind Cloudflare.

I just don't get it. Why is this needed?

But I have no doubt there is a play happening here.

Probably it will change over time to make gathering data easier?

Or something else that makes Google money.

google.com/maps to capture your location across the whole domain easier than before

Google maps on mobile is directing the user to the native app more and more. So I doubt this. Geolocation on desktop is mostly useless.

The demo crashes Chrome on Android for me.

https://permission.site/geolocation_element.html

Works on my phone (after enabling the experiment in chrome-flags). Even includes localised permission prompts.

Also crashed for me on Chrome on Mac

This mostly changes how location is requested, not what you can do with it. Instead of imperative JS calls, location access becomes declarative in HTML, which gives browsers more context for permission UX and auditing. Your app logic, data flow, and fallbacks don’t change, and you’ll still need JS to actually use the location. Think of it as a cleaner permission and intent layer, not a new geolocation capability.

Main purpose of this seems to offer a way for undoing "previously blocked location access" by the user.

> If a prompt appears unexpectedly, users may block it reflexively or accidentally, unaware that this decision creates a permanent block that is difficult to reverse. This context gap—rather than the feature itself—is a primary driver of high denial rates.

> If a user previously blocked location access when browsing a site (perhaps by accident or lack of context), clicking the element triggers a specialized recovery flow. This helps them re-enable location at the moment when they actually want to use location, without the friction of navigating deep into the browser's site settings.

Google sees "high denial rates" when they try ask users for their geolocation. This is a problem for Google's customers, the advertisers. So they introduce this <geolocation> HTML tag so that dark patterns can be employed to trick users into permanently sharing location even though they have blocked location sharing before.

If the Google engineers who are working on this feature would actually give a damn about users who decided to block geolocation access, this feature would be designed as a "temporary access to geolocation for duration of browser session".

So basically it is all about more tracking and less data privacy.

It's overdue that skilled engineers provide better solutions than this crap, but of course it's much easier to be apolitical and become a millionaire working for a bunch of tech bros who visited Epstein's island.

Yeah they cited Zoom as a successful case study, but why does Zoom need access to my location in the first place? Asking me when I click the <geolocation> button will not change my decision to block.

Also I’m not sure about the argument of context disconnect. Properly designed websites will only ask for (and prompt the location permission modal) when it really needs it.

The Zoom mention was about camera and microphone permission with <permission>, not location:

> Zoom reported a 46.9% decrease in camera or microphone capture errors

This seems completely unnecessary. If they want to solve the issue of permission prompts on website load without user intent then disallow permission requests that are done outside a click event handler.

I assume an adblocker can remove those.

That would be a net benefit because pages requesting location for no reason end up in the block list and don't annoy me anymore.

This is pretty cool. But what gets me really excited is the new generic <Permission>[0] element. I had to implement a webcam element one time for some CV pet project and I had a lot of trouble getting the basic api to just work (Highly likely a skill issue). So seeing that this will also expand to webcam and other IO seems like a really good UX improvement.

- [0] https://github.com/WICG/PEPC/blob/main/explainer.md

Why not just expose a GeoLocationEvent and call it a day? why add another element people will wrap anyway?...

Adding a new HTML element probably leads to a better bonus at Google.

There probably is a bonus associated with each web proposal you create and get accepted. It's most frustrating when they take an existing proposal by an outsider chop it up into several separate proposals, and somehow still manage to miss the point of the original while simultaneously making it all more complex.

Yay! Next up, <usermedia>, which is even more vital a permission to be able to flip on and off!! https://github.com/WICG/PEPC/blob/main/usermedia_element.md https://groups.google.com/a/chromium.org/g/blink-dev/c/jQgYo...

As noted in the intent to ship for both, these are a very specific narrow cases chipped off a bunch broader attempt to offer declarative ways to handle permissions in general, a <permission> element.

Intent-to-ship for geolocation: https://groups.google.com/a/chromium.org/g/blink-dev/c/GL0Bk...

Earlier Page-Embedded Permissions Control (PEPC) proposal: https://github.com/WICG/proposals/issues/113 https://github.com/andypaicu/PEPC/blob/main/explainer.md

The root problem is that permissions right now are super hard to adjust for users (and the way they are exposed to the page is not very good at dealing with users turning permissions on and off either). It's imo very good that we are finally leaving this awful tarpit of design, & moving towards permissions being more fluid. I get that other browsers wanted to be conservative & not do a generic <permission> element, but given how big an improvement this feels like, I sort of wish it'd gotten the pass.

I wonder if this new element is used by MapLibre.

Reply on news.ycombinator.com