mirror of
https://git.sr.ht/~seirdy/seirdy.one
synced 2024-12-25 02:02:11 +00:00
Add gemtext version of FLoC article
This commit is contained in:
parent
1e7f1260d7
commit
7429b31fec
2 changed files with 76 additions and 0 deletions
75
content/posts/permissions-policy-floc-misinfo.gmi
Normal file
75
content/posts/permissions-policy-floc-misinfo.gmi
Normal file
|
@ -0,0 +1,75 @@
|
|||
This post was written in a hurry in response to some misinformation about Google's newest Web antifeature, Federated Learning of Cohorts (FLoC). Google's FLoC is an attempt to track users even when their browsers (rightly) block third-party cookies.
|
||||
|
||||
A certain blog post ¹ has been making rounds:
|
||||
|
||||
=> https://paramdeo.com/blog/opting-your-website-out-of-googles-floc-network Opting your website out of Google's FLoC network
|
||||
|
||||
It instructs webmasters everywhere to add the following HTTP response headers to all their pages:
|
||||
|
||||
```
|
||||
Permissions-Policy: interest-cohort=()
|
||||
```
|
||||
|
||||
Strictly speaking, this advice is correct; including this header will prevent your page form using Google's FLoC. What the advice misses is that the header isn't always necessary and isn't even the ideal way to prevent your page from doing nefarious things.
|
||||
|
||||
That advice comes from a post by Plausible Analytics:
|
||||
|
||||
=> https://plausible.io/blog/google-floc#how-to-opt-out-of-floc-as-a-web-developer-set-a-permissions-policy How to fight back against Google FLoC
|
||||
|
||||
The author of that post concluded that FLoC is an "opt-out" antifeature by reading the "Opting Out of Computation" section of the WICG's FLoC README:
|
||||
|
||||
=> https://github.com/WICG/floc#opting-out-of-computation WICG/floc on GitHub
|
||||
|
||||
If you check that README yourself, you'll notice the following statement one (1) paragraph above the mention of the HTTP permission policy (emphasis mine):
|
||||
|
||||
> A site should be able to declare that it does not want to be included in the user's list of sites for cohort calculation. This can be accomplished via a new `interest-cohort` permissions policy. This policy will be default allow. Any frame that is not allowed `interest-cohort` permission will have a default value returned **when they call `document.interestCohort()`.** If the main frame does not have `interest-cohort` permission then the page visit will not be included in interest cohort calculation.
|
||||
|
||||
If your website does not include JS that calls `document.interestCohort()`, it will not leverage Google's FLoC. There will be no need to opt-out.
|
||||
|
||||
As per a post on Google's web development blog, web.dev, FLoC also will be enabled during an origin trial if the page "loads ads or ads-related resource"; i.e., anything that Chromium's ad tagger classifies as an ad. If your website loads third-party ads, it might end up getting opted-in even if those ads don't call `document.interestCohort()`.
|
||||
|
||||
|
||||
=> https://web.dev/floc/ What is FLoC? (web.dev)
|
||||
=> https://chromium.googlesource.com/chromium/src/+/master/docs/ad_tagging.md Ad Tagging in Chromium
|
||||
|
||||
## How Permissions Policy works
|
||||
|
||||
I don't think that every webmaster should have to read every single W3C spec. I do, however, think that people who offer prescriptive advice and interpretations of a spec should be expected to read the relevant spec first.
|
||||
|
||||
Here's a copy of the Permissions Policy spec:
|
||||
|
||||
=> https://www.w3.org/TR/permissions-policy-1/ Permissions Policy (W3C Working Draft, 16 July 2020)
|
||||
|
||||
Essentially, the `Permissions-Policy` header allows a webmaster to whitelist which parties (if any) are allowed to leverage certain APIs. If I make a website that doesn't perform any geolocation directly but I load a third-party widget that does, I can forbid all parties from using Geolocation APIs by setting the following Permissions-Policy:
|
||||
|
||||
```
|
||||
Permissions-Policy: geolocation=()
|
||||
```
|
||||
|
||||
In other words, it provides a global override for any page contents requesting too many permissions. It's ideal for situations in which authors aren't in control of what content is being loaded.
|
||||
|
||||
## Better advice: how not to opt-in
|
||||
|
||||
Here's how not to opt-in to Google's FLoC:
|
||||
|
||||
* Don't load untrusted third-party content that might get classified as an ad (only applies during the origin trial)
|
||||
* Don't call `document.interestCohort()`, and don't load third-party scripts that might call it either.
|
||||
|
||||
If you have to resort to adding a Permissions-Policy header to opt your site out of FLoC, it means that you've accidentally opted yourself in by including a tracking script (malware) on your page. I'd wager that opting into FLoC wasn't the only nefarious thing those scripts do; chances are they bundle a host of other fingerprinting measures.
|
||||
|
||||
To be extra safe, you can whitelist exactly what scripts can run with a Content Security Policy (CSP); seirdy.one, for instance, has a CSP that blocks all scripts, forms, frames, and third-party resources from loading.
|
||||
|
||||
|
||||
=> https://developer.mozilla.org/en-US/docs/Web/HTTP/CSP Content Security Policy (MDN)
|
||||
|
||||
If you're concerned about Google breaking the spec and opting you in even after you've not done so yourself, what reason do you have to believe that they'll stop there? There's nothing preventing Google from ignoring your `Permissions-Policy` header, either.
|
||||
|
||||
Don't tack a new header onto every request because someone read a post by someone else who read part of a GitHub README.
|
||||
|
||||
## Take a breath
|
||||
|
||||
Please, don't spam maintainers of web server/backend software to tell them to include this header by default. Don't tell webmasters that they have a *moral obligation* to add a Permissions Policy header either.² You don't need to add this permission policy to every request, just as you don't need to wear a helmet for every form of physical activity.
|
||||
|
||||
¹ This isn't the only post making rounds, but it did hit the front page of a certain orange-colored website. I'm not blaming the author; if I hadn't encountered the Permissions Policy spec earlier, I probably would have also taken the advice the author read at face-value.
|
||||
|
||||
² I've noticed both of these behaviors on several threads online. I've decided against linking to them because I think the discourse there has heated past the point of reason.
|
|
@ -5,6 +5,7 @@ description: Recently, people have been telling webmasters to add a Permissions-
|
|||
how the Permissions-Policy header works.
|
||||
outputs:
|
||||
- html
|
||||
- gemtext
|
||||
footnote_heading: Notes
|
||||
title: "Misinformation about Permissions Policy and FLoC"
|
||||
---
|
||||
|
|
Loading…
Reference in a new issue