Ad tech firms test ways to connect Google’s FLoC to other data • Hacker News Board

Re: Ad tech firms test ways to connect Google’s FLoC to other data

#91

Post by pdimitar » Thu, Jun 10, 2021, 1:18 PM UTC

gorhill wrote: post #76

pdimitar wrote: post #61

Earlier quoted context omitted.

Yes. Often times pages load very slowly, we're talking 5-8 seconds. I am on a gigabit connection (yes, that doesn't guarantee that my ISP has fast access to that particular page, I know) and Chrome is always loading those pages at least twice as fast. Can't describe it perfectly. The UI is responsive but page loading is just severely slowed down -- not always but often. I have a bunch of privacy extensions but again,…

> Often times pages load very slowly, we're talking 5-8 seconds [...] I have a bunch of privacy extensions If you are using uBlock Origin, you may want to see if un-checking "Uncloak canonical names" option in the "Settings" pane in the dashboard makes a difference.[1] There have been reports of slow page load with some network configurations, and this has been linked to DNS lookup in uBO.[2] Chromium-based browsers…

Super interesting, thank you! I immediately disabled it and will monitor if that improves my page loading speeds.

EDIT: pay no attention to the text below, I have misread the linked documentation. uBO isn't using external proxy for any network requests.

Not sure how much -- or at all -- you're involved with uBO. Your name does ring a bell though so I'd like to remark to you that making the users' browser use proxy is a step too far. It shouldn't automatically be enabled.

A privacy extension should do everything it could locally and stop there. If I one day figure it's not enough then I'll set a privacy VPN (or use an existing one).

I don't want that decision made for me on my own machine without my consent. :(

And apologies if my comment is misguided -- I only skimmed the linked page and I might have misunderstood.

Re: Ad tech firms test ways to connect Google’s FLoC to other data

#92

Post by ahmedalsudani » Thu, Jun 10, 2021, 1:23 PM UTC

alias_neo wrote: post #45

ahmedalsudani wrote: post #40

Earlier quoted context omitted.

Correction: they never injected ads, but they do remove ads and replace them with "attention tokens", which is an opt-out system for publishers. See also https://tech.hindustantimes.com/tech/news/this-popular-brows... . In short, please look into the questionable aspects of Brave. Brave is perhaps the most ethically challenged browser out there. Hopefully they have stopped doing this, but they were injecting their ow…

Is it Brave that uses BAT? You're being downvoted but weren't they collecting BAT for sites whose owners had never signed up, and presumably keeping it if the owner didn't claim it?

Right, that's what I was thinking of. They created an opt-out system for publishers.

The reader can guess how Brave expects to make money with a free browser that is handing out BAT :-)

Re: Ad tech firms test ways to connect Google’s FLoC to other data

#93

Post by SXX » Thu, Jun 10, 2021, 1:23 PM UTC

cheschire wrote: post #84

Dah00n wrote: post #81

Earlier quoted context omitted.

Because as it is now Chrome (read: Google) can dictate how the web works and they do. If you use a Chromium based browser everything just works and happy people doesnt often complain or see the problems. So using something Chromium based makes you part of the problem. Apple isn't much better though so there aren't many options left...

What prevents forks of chromium from sharing amongst each other? I don't think google can gatekeep to the extent you seem to be implying. I'm pretty sure Brave and Opera and MS Edge can share changes they've made amongst each other without going through the original repository first.

You underestimate how much of effort and money it takes to maintain fork of a browser. Chromium is project as big as Linux kernel (if not larger), but it's primary developed by Google emplyees so they have huge influence on direction of the project.

Re: Ad tech firms test ways to connect Google’s FLoC to other data

#94

Post by danShumway » Thu, Jun 10, 2021, 1:27 PM UTC

mavhc wrote: post #36

ece wrote: post #20

Honestly, who asked for FLoC and the MS alternative (edit: it's PARAKEET)? Why can't ad servers just follow my Do Not Track preferences? Serve me a random ad, or a context based ad if I have Do Not Track set... seems a lot simpler no? Those using ad-blockers are going to use them anyway, but some assurance that I won't be tracked would go a long way towards me turning off ad-blockers on the sites I don't frequent, an…

Do Not Track failed because a) the worst sites would just ignore it and b) Microsoft tried to use it as a weapon against Google by setting it as default

Obligatory "Microsoft Did Nothing Wrong" post:

Too busy to do a long post in detail, but short version is that advertiser's acceptance of DNT was entirely dependent on people not using it. If Microsoft had left it as opt-in but the majority of consumers had turned it on, we would have seen the same result.

You can see the same principle to explain the response to iOS's privacy changes, which are not opt-in or opt-out; they force the user to make a choice. The ad industry is still furious about this.

DNT could have only worked if no one used it, and that is not a privacy outcome that is worth pursuing or advocating for. It's not Microsoft's fault that DNT went away, Microsoft was just the excuse the advertising industry needed to avoid it. I don't think that DNT was ever anything other than an excuse the industry could use to keep tracking people and avoid legislation without changing anything. Microsoft didn't take it away from you, they just pulled back the curtain and showed you the truth about it.

A privacy standard that fails as soon as it's turned on by default is almost worthless and it really shouldn't be advocated for in the first place.

- https://news.ycombinator.com/item?id=26821972

- https://news.ycombinator.com/item?id=24294280

- https://news.ycombinator.com/item?id=24289186

- https://news.ycombinator.com/item?id=19483149

Re: Ad tech firms test ways to connect Google’s FLoC to other data

#95

Post by robertlagrant » Thu, Jun 10, 2021, 1:31 PM UTC

ohgodplsno wrote: post #80

pmurt7 wrote: post #72

Earlier quoted context omitted.

Chromium/Blink is an open source project (forked from Webkit, forked from KHTML).

Chromium/Blink is pretty much only maintained by Google. Google decides what goes in and what doesn't. It's a pretense of open source, except when it serves them.

That's often how open source works. It may not be what you want, but it's not just a pretense.

Re: Ad tech firms test ways to connect Google’s FLoC to other data

#96

Post by magwa101 » Thu, Jun 10, 2021, 1:38 PM UTC

"Only we have valuable specific timely data to accurately link it to the users general interests via the FLOC ID which is a gradual, general indication of users interests. This is how you beat your competitors."

"Better than retargeting?"

"For sure, because general interests combined with accuracy will bring the best performance."

"Reup me please."

Re: Ad tech firms test ways to connect Google’s FLoC to other data

#97

Post by danShumway » Thu, Jun 10, 2021, 1:45 PM UTC

But this can't be, people assured me that FLoC was an improvement to privacy, and that it would provide an alternative to persistent identifiers and profiles, and that it would help end the arms race on fingerprinting.

You mean to tell me that FLoC will be used for fingerprinting anyway, and it changes nothing about advertiser's strategies and tracking techniques, and they won't self-regulate, and that it doesn't work to throw them bones of extra data and hope that they'll willingly stop their abusive behavior if we meet them halfway?

This is a shocking development.

The only consolation is that Google's next privacy compromise with the ad industry definitely won't suffer from exactly the same problems. The best thing for us to do now is to assume that this is a completely random, one-time fluke that doesn't reflect anything on the industry's character. No need to change the way we engage with the advertising industry on privacy issues because of it. We should keep offering them compromises that make it easier for them to track users, and keep assuming that they'll in good faith regulate themselves.

Re: Ad tech firms test ways to connect Google’s FLoC to other data

#98

Post by hluska » Thu, Jun 10, 2021, 1:47 PM UTC

gerbler wrote: post #18

> However, it is likely that the same FLoC IDs or range of IDs will be associated with someone. This seems all quite speculative. In the first paragraph they describe FLoc IDs as changing constantly - why would they assume new IDs are not being generated, and groups are not constantly being mixed?

It’s extremely speculative. While I’m allergic to ad tech, I’m even more allergic to poverty, so I have occasion to work with/around that side of the industry.

The best way to explain it is that a lot of companies have been making significantly more money than their technology is worth. A number of initiatives have attacked the data-in side of the equation so the underlying tech is showing how questionable it really is.

This type of research should be filed under “could be big” but at this point it’s closer to public relations for ad tech firms than “the sky is falling, become Amish.”

Re: Ad tech firms test ways to connect Google’s FLoC to other data

#99

Post by OhNoMyqueen » Thu, Jun 10, 2021, 1:50 PM UTC

This is exactly how FLoC is supposed to work, and this is exactly what privacy watchdogs have criticized about it.

The proper headline should be "Ad tech films test FLoC".

Re: Ad tech firms test ways to connect Google’s FLoC to other data

#100

Post by gorhill » Thu, Jun 10, 2021, 2:03 PM UTC

pdimitar wrote: post #91

gorhill wrote: post #76

Earlier quoted context omitted.

> Often times pages load very slowly, we're talking 5-8 seconds [...] I have a bunch of privacy extensions If you are using uBlock Origin, you may want to see if un-checking "Uncloak canonical names" option in the "Settings" pane in the dashboard makes a difference.[1] There have been reports of slow page load with some network configurations, and this has been linked to DNS lookup in uBO.[2] Chromium-based browsers…

Super interesting, thank you! I immediately disabled it and will monitor if that improves my page loading speeds. EDIT: pay no attention to the text below, I have misread the linked documentation. uBO isn't using external proxy for any network requests. Not sure how much -- or at all -- you're involved with uBO. Your name does ring a bell though so I'd like to remark to you that making the users' browser use proxy is…

> making the users' browser use proxy is a step too far

uBO does not do this, and nowhere is there any suggestion that uBO does this.

Users configure their own network settings, and it was found that in some cases when the browser is configured to go through a proxy (through either OS or browser settings), uBO's CNAME-uncloaking feature, which requires a call to the `dns.resolve()` API[1], would cause undue delay to page load. The root cause is outside uBO and outside the browser, it lies in the proxy.

* * *

[1] https://developer.mozilla.org/en-US/docs/Mozilla/Add-ons/Web...