4 ms·

I'll take a stab. Suppose your browser told every website you visited that you had the identifier "FAANG", and your friends web browser told every website they

I'll take a stab. Suppose your browser told every website you visited that you had the identifier "FAANG", and your friends web browser told every website they visited: "FAA_G" (b/c obviously they don't have netflix). Per the linked article, the browser is filling in the blanks "___N_" as you visit each website, but it's telling each website not _what_ site you visited, but that you are in the groups of "FAA_G" vs. "FAANG". The browser manufacturer is effectively seeing every user for what they are (all the letters of the alphabet, as it were), and each site is only seeing a hash/digest/cohort/group, but those group segregations are not coming from particular "networks of websites", but instead from the browser itself. The end result is that you can't trust the browser b/c it might be (ed: it is!) spying on you. Google started removing the URL bar and replacing it with the omni-box. Instead of https://google.com/ https://google.com/ => ?q=guitar, you would simply type "guitar" into the URL bar. The lesson here is to reduce the number of intermediaries between you and your customer. Own the hardware, the screen, the operating system, the network, _everything_ between your CPU's and the user's eyeballs. (or in amazon's case, the manufacturer, the boxes, the website, the shipper, the TV, etc.). (or in netflix's case, the content, the edge-cache storage tank, the TV, the remote control, the pixels, etc.)

sumtechguy 6y ago
If I am reading this right (and you seem to have read it the same way) it ends up creating a browser fingerprint. Visit 1 FLoC and they have no real idea who you really are. But if you visit several they could easily tell. Which is similar to how they use your installed list of fonts to identify you. But more formalized?
- munk-a 6y ago
  That compromise would require knowing who someone actually is between site and site - you could collect the full list of cohorts but you'd need additional information to be able to marry all that data together - let's say something like facebook injecting JS into every page on the internet to harvest cohort information and then gleefully reselling that information.
- joshuamorton 6y ago
  No, this isn't correct. You only ever belong to a single FloC. So let's for a second assume the non-adversarial use case. The advertiser isn't using any additional tools to track you. Then the only thing the advertiser has is your FloC. This doesn't identify you individually, but it might correlate with various interests. If your FloC changes, which it will periodically, the advertiser doesn't have any history, so the correlation changes and you're given different ads. Now let's assume that they have some additional information, like an IP address. Then they can say ah you had this FloC on this date, and this different FloC later. Its not more identifying (they already have your IP, which we assume is identifying enough). The FloC might let them infer things about your interests on other sites, so they have precise information on your browsing interests on this site, but only partial information on other sites. As opposed to today, where 3rd party trackers mean they have full information on the other sites too.
  - alerighi 6y ago
    It's not the same as third party cookies! Third party cookies works only if: a) every site you visit loads some Google/Facebook/whatever JavaScript to track you. That are not all sites. b) you didn't disable these scripts with extensions like UBlock (that to me is essentials these days) or use a browser that value your privacy (like Firefox) that blocks them by default. This system not only is deeply integrated in the browser and thus impossible to block with extensions (without modifying the source code of the browser, that in case of a proprietary browser like Google Chrome you can't), but also track all your browsing history, meaning that they catch even sites that doesn't include Google trackers inside. Another bad thing about this system is that is integrated inside the browser, meaning that for a closed source browser like Chrome only Google knows how it works and what exactly it does. While classical tracker scripts that uses third party cookies are implemented in JavaScript, minified and obfuscated, but still you can in theory read the source code and understand what they do. This is far worse for privacy than how things are now!
    - joshuamorton 6y ago
      > a) every site you visit loads some Google/Facebook/whatever JavaScript to track you. That are not all sites. Right, but it is, to a first approximation, all of the sites that are going to be showing you personalized ads. > Another bad thing about this system is that is integrated inside the browser, meaning that for a closed source browser like Chrome only Google knows how it works and what exactly it does. While classical tracker scripts that uses third party cookies are implemented in JavaScript, minified and obfuscated, but still you can in theory read the source code and understand what they do. I'm not quite sure what you're getting at here. The setting and reading of a cookie by a client, yes, will be written in JS. That doesn't mean that you can know what its doing (there are likely cookies on your system that contain encrypted payloads that you can't read). The system that "actually" reads those cookies is hidden behind an API, so you don't have access to even the binary code. From that perspective, FloC is no worse, and is usually better, as the payload itself is generated and managed on your machine. Tangentially, FloC is similar to but probably more privacy preserving than the way Brave Browser does advertising today, and that was heralded by many on HN as a huge privacy improvement.
bsanr2 6y ago
Of tangential note, you may be able to better make your point by assuming that User 2 doesn't use Apple products rather than Netflix.
TeMPOraL 6y ago
> but those group segregations are not coming from particular "networks of websites", but instead from the browser itself. Wait, I thought the whole point was that third parties could define what a "cohort" is, and the browser would handle determining whether or not you belong in them?
- tmottabr 6y ago
  no, the browser that generate your cohort based on all the sites you visited on the last week.. it refresh it every week.. and it will include all sites from the last week, whatever you used adblocker to prevent tracking or if the site did not had any social media buttons that would allow then to track you there in the first place..

sumtechguy 6y ago

If I am reading this right (and you seem to have read it the same way) it ends up creating a browser fingerprint. Visit 1 FLoC and they have no real idea who you really are. But if you visit several they could easily tell. Which is similar to how they use your installed list of fonts to identify you. But more formalized?

munk-a 6y ago
That compromise would require knowing who someone actually is between site and site - you could collect the full list of cohorts but you'd need additional information to be able to marry all that data together - let's say something like facebook injecting JS into every page on the internet to harvest cohort information and then gleefully reselling that information.
joshuamorton 6y ago
No, this isn't correct. You only ever belong to a single FloC. So let's for a second assume the non-adversarial use case. The advertiser isn't using any additional tools to track you. Then the only thing the advertiser has is your FloC. This doesn't identify you individually, but it might correlate with various interests. If your FloC changes, which it will periodically, the advertiser doesn't have any history, so the correlation changes and you're given different ads. Now let's assume that they have some additional information, like an IP address. Then they can say ah you had this FloC on this date, and this different FloC later. Its not more identifying (they already have your IP, which we assume is identifying enough). The FloC might let them infer things about your interests on other sites, so they have precise information on your browsing interests on this site, but only partial information on other sites. As opposed to today, where 3rd party trackers mean they have full information on the other sites too.
- alerighi 6y ago
  It's not the same as third party cookies! Third party cookies works only if: a) every site you visit loads some Google/Facebook/whatever JavaScript to track you. That are not all sites. b) you didn't disable these scripts with extensions like UBlock (that to me is essentials these days) or use a browser that value your privacy (like Firefox) that blocks them by default. This system not only is deeply integrated in the browser and thus impossible to block with extensions (without modifying the source code of the browser, that in case of a proprietary browser like Google Chrome you can't), but also track all your browsing history, meaning that they catch even sites that doesn't include Google trackers inside. Another bad thing about this system is that is integrated inside the browser, meaning that for a closed source browser like Chrome only Google knows how it works and what exactly it does. While classical tracker scripts that uses third party cookies are implemented in JavaScript, minified and obfuscated, but still you can in theory read the source code and understand what they do. This is far worse for privacy than how things are now!
  - joshuamorton 6y ago
    > a) every site you visit loads some Google/Facebook/whatever JavaScript to track you. That are not all sites. Right, but it is, to a first approximation, all of the sites that are going to be showing you personalized ads. > Another bad thing about this system is that is integrated inside the browser, meaning that for a closed source browser like Chrome only Google knows how it works and what exactly it does. While classical tracker scripts that uses third party cookies are implemented in JavaScript, minified and obfuscated, but still you can in theory read the source code and understand what they do. I'm not quite sure what you're getting at here. The setting and reading of a cookie by a client, yes, will be written in JS. That doesn't mean that you can know what its doing (there are likely cookies on your system that contain encrypted payloads that you can't read). The system that "actually" reads those cookies is hidden behind an API, so you don't have access to even the binary code. From that perspective, FloC is no worse, and is usually better, as the payload itself is generated and managed on your machine. Tangentially, FloC is similar to but probably more privacy preserving than the way Brave Browser does advertising today, and that was heralded by many on HN as a huge privacy improvement.

bsanr2 6y ago

Of tangential note, you may be able to better make your point by assuming that User 2 doesn't use Apple products rather than Netflix.

TeMPOraL 6y ago

> but those group segregations are not coming from particular "networks of websites", but instead from the browser itself. Wait, I thought the whole point was that third parties could define what a "cohort" is, and the browser would handle determining whether or not you belong in them?

tmottabr 6y ago
no, the browser that generate your cohort based on all the sites you visited on the last week.. it refresh it every week.. and it will include all sites from the last week, whatever you used adblocker to prevent tracking or if the site did not had any social media buttons that would allow then to track you there in the first place..