Am I FLoCed? | Hacker News Reader

Am I FLoCed?

amifloced.org

154 pointsfilleokus33 comments

I haven't consciously chosen to use Chrome in so long, I can't actually remember the last time I did.

Firefox for the win, every time!

Me too, I just never switched when everyone else did. It's always just worked for me.

I should never have switched. I spent time jumping between every browser I saw, learning their workflows and quirks, but I've returned to Firefox. The one lesson I've settled on for myself is: if I can't uninstall a browser, I won't trust it.

I had to use it because Teams ist actively ignoring years of best practice for web development and just decided to boycott any noon IE/Chrome/Chromium browser instead of doing feature detection and shims.

Most ironic screen sharing does work best in Firefox under wayland as chromium based browser still don't have proper pipewire support (it's experimental and uses an outdated pipe write api, you can adapt it and it works fine but seriously given covid-19 Google should have put some resources into).

Works until it doesnt. Last uodate on ubuntu 18 completely destroyed my firefox and i had ro switch to chromium

@Downvoters why downvote the simple truth ? Firefox is not perfect in terms of usability anyway.

"For example, a site can opt out of all FLoC cohort calculation by sending the HTTP response header:

Permissions-Policy: interest-cohort=()"

https://github.com/WICG/floc

Added to proxy config.

http-response add-header "Permissions-Policy: interest-cohort=()"

Client side there is also the DNT header which everyone, including Google is happily ignoring. So I'd imagine things will get ignored similarly here.

You are suggesting that Chrome will ignore this header sent from a site. You are suggesting that sites will not be able to opt-out.

Seems like the easierst way to opt-out of FLoC is to use a browser that does not support it. I already do that. It does not even support CSS or Javascript. I love it.

With client-side header "DNT", sites can ignore it. With server-side header "Permissions-Policy", browsers or other programs that make HTTP requests can ignore it.

Wrong syntax.

http-response add-header Permissions-Policy "interest-cohort=()"

Cloudflare and Fastly should step up and add this header for all domains by default. That would probably be enough to kill this idiotic mal-feature.

Clever name, I suppose, since the implication is that we're sheep.

Flock means "group"

...and sheep in a group are called flocks.

How does FLoC compare to prior forms of tracking? I thought the point of federated learning was to improve privacy by obscuring folks' data amongst a cohort of others, like Tor (or firing squads...)

EFF is really deciding to die on a hill on a feature where all the tracking is moved to your owned device instead of being held on some foreign servers? Despite that giving your software full control what it collects and reports to the servers?

Even Apple uses this kind of federated learning approach for their data collection needs.

Is EFF fighting to improve privacy or are they only interested in attacking certain corporations?

I could very well argue that ad sellers are trying to die on the hill where they try to track their users for increasingly better ad-targeting to the point where some instances of that targeting becomes downright scary and/or illegal.

According to the articles I've read, this will actually decrease the quality of targeting in exchange for keeping the data locally. Is that wrong?

You're assuming this is also responsible for the decrease in targeting quality. Apple and Mozilla have implemented similar anti-tracking changes in ITP and ETP without a similar compensatory system like FLoC and GDPR enforcement is slowly moving into gear hence Google's redesigned consent dialog after their latest fine which puts making an ad tracking decision on the front page.

What advertisers have is going away, whether FLoC gets implemented or not, FLoC is Google throwing the advertising industry a bone so they don't get accused of anti-competitive behaviour by implementing rules in their OS and browser businesses that will affect them less due to all their first party users when customer and regulatory pressure forces them into following Apple

How about not tracking users period? Billboard ads and street ads are still things advertisers buy even though they cannot track individuals.

Federated learning on itself is not the problem.

But just because it's federated doesn't mean it's privacy friendly or creates a massive subtle vendor locking situation.

Genuinely - what's stopping Firefox from implementing this and sending a set of data that's either chosen by you in UI or setting it to empty?

As a replacement for server-side tracking? Which part of this is not a net win for privacy?

> Which part of this is not a net win for privacy?

The way google implements it all of it!

Google will assign you a Id, sure this Id is shared with others but on a It/Statistic scale it's not that many others. Furthermore every website you and every add on it gets the Id.

If you combine that id with other identifying information you can easily extract from a browser without needing any permissions you get a pretty close to or maybe even fully unique id!

In the end while it seems to be an improvement on the first look it actually makes deeply privacy invasive tracking easier, not harder.

EDIT: At the same time it allows google to pretend to be good in front of law makers. And potentially hide behind things like "our AGBs require that you don't abuse the Id" (which given that it's send to anyone independent of weather or not they agreed to the AGBs is stupid, and also even if every one did agree to the AGBs common abuse of the android advertiser Id without consequences shows how well that works.)

A proper federated ad system doesn't provide a way to unique identify you to advertisers. It lets advertisers provide a selection of labeled add slots and the browser then pics one of them without any information besides the choice ever going to the advertiser. This (oversimplified presented approach) still can slightly be abused. But is one of the few ways you can have personalized ads without a absurd invasion of privacy.

> Genuinely - what's stopping Firefox from implementing this and sending a set of data that's either chosen by you in UI or setting it to empty?

I haven't looked at the tech but most likely it might involve certificates, drm, and legal agreements which would make it impossible for a cooperation to do so.

> data collection needs.

There are no "data collection needs" there are "data collection wants". If they asked, people would say no. To pour so much effort into manipulating people into letting themselves be monitoring and just tracking people without their consent is straight up abusive.

They asked "Hey can I have this data about you?"

We said "We´d actually rather you didn´t."

End of discussion. This is not complicated concept. Informed consent is not a complex theory, it´s possible for a class of 2nd graders to grasp it.

I don´t need new and improved ways for this dystopian garbage to be more convenient to me.

I need it so stop.

This is exactly the point I've been wondering about. What strikes me about this FLoC stuff is that we're removing a technology which was genuinely built for the user (cookies) to a technology whose SOLE purpose is tracking users.

For some reason we're not properly questioning the assumption that this is remotely reasonable: why should we have tracking at all? It's as though we replaced facial recognition cameras piggy-backing on CCTV security systems with chipping every citizen and then saying "but it's just a better tracking system" rather than "is this even reasonable?"

This all stems from people having been trained to expect everything online to be "free".

I disagree because everything on antenna, cable, satellite television or radio is similarly “free” unless it is locked somehow (e.g. cable or satellite programming packages, pay-per-view television, different radio technology like SiriusXM, etc.).

I hold the view that, on the contrary, this all mostly stems from people having been trained to expect everything they want to do online can be turned into a "business" by itself.

Yeah I don't understand their stance. If this becomes popular and everyone uses it then it will become trivial to avoid being tracked. Is that not something they want?

Could firefox (or chrome) not implement this and let the user decide what category of ads they'd like to see? Or choose to have it randomly selected per website? Or respond with some value like "random ad please".

Why would Firefox implement this? No one but Google has anything to gain from this.

Except for that Mozilla receives funding through a Google search deal, but I hope they manage to remain independent and keep tracking out of the browser.

I can see this being a plus if it's randomised. Every request posts random garbage, similar to resist fingerprinting. From my naive understanding, it sounds like it would be easier to avoid tracking compared to all the cookie trickery that Firefox does.

My thought was that it may help further an argument that the web user themself can choose and declare what type of ads they want to see if people insist on showing "relevant" ads. But perhaps have fewer than 33000 categories to chose from.

Reply on news.ycombinator.com