Google Chrome browser privacy plan investigated in UK
bbc.co.uk
bbc.co.uk
It's not the targetted advertising I object to it's the tracking.
I don't want Google to solely access to all my personal data which they then use to provide anonymised information to others.
I want to have enough control to stop Google tracking me in the first place rather than anonymising things after the fact.
Something like Brave's approach to BATs is what I'd like to see them adopt.
Google just wants a monopoly on advertising.
I'd disagree. There is a trend in HN comments to remind people that a company is a collection of people. Google is a collection of people. There are those of us who care deeply about privacy at Google.
First: it's interesting talking with googlers or reading their thoughts on how Google does ads and data collection. Many people definitely feel the same way as commenters on HN feel. Lots of people are torn on the fact that ads lets us build lots of other cool products for people. I would say this helps motivate other teams, like cloud, do you find other revenue sources for the company so you don't have to be as dated collection focused.
Second: I have seen googlers fight the privacy of users even within my division (payments). Many of us want to do right by our users, and be as privacy focused as we are capable given our constraints. We also know there is a general thought that Google is data collection focused, and one slip up will cause a big drama out on the internet and in the news (Google would likely receive more scrutiny here than some other companies). This helps remind us that we need to treat user data as best we can and minimize what we do collect.
It sounds like there is little overlap between Google's employees and the people who make and influence major decisions for Google. Wasn't that recently given as one of the major justifications for unionization?
A company isn't really that, though: a company employs a collection of people to do the work desired by its owners.
Since Google is not an employee-owned company, it's not obvious that a group of employees have effective power in this matter.
The best foundation for good outcomes is to not trust a company and to reject its control to begin with.
Googlers are earning most of their paycheck from spying on people. This is a fact. And everything about google is moving into direction of earning even more - stock owners demand it and based on infinite hunger for more revenue, this is a lost fight. Whatever you do, from within the Alphabet, it is lost game.
Now some fraction of those that are not aware of reality might be really fighting for user privacy but your fight is like biting a hand that feeds you. And this is observed as crazy at best. If you want to talk the talk and walk the walk, stop helping them and allow to have public image that they deserve instead of defending them based on some strange minority that is clueless where they work.
Or if you want it differently, it is like selling weapons to Central Africa and then saying "Weapons dont kill, people do.". Sure. True. But you are making it more efficient.
Now we will probably move to the part "someone else will do it instead". This is the part of having moral and this makes a difference between someone who has it and someone who is searching for apologies.
This a popular misconception about Google. They don't need to spy on people. You're the product, but you don't pay with data (except for basic demographics) - you pay with attention.
When push comes to shove, Google _isn't_ a collection of people - it's a legal entity with the sole purpose of turning an ever-increasing profit.
The best way to ensure privacy is to not track users across the web. Hence, if Google was interested in privacy, that's what Google would do - not add extra "anonymized" tracking features.
Google's proposals actually anonymize data before being sent to Google or any other ad network. But this only works by moving parts of the ad infrastructure into your browser. Do you really want your browser run machine learning algorithms to assign you to a cohort (see FLoC), or have ad auctions take place on your device (see TURTLE-DOV)?
and TURTLE-DOV -> https://github.com/WICG/turtledove
Thanks OP for mentioning these.
(Disclosure: I work on ads a Google, speaking only for myself)
Although I understand that the revenue and reality of marketing revenue is vast and in some ways unstoppable, this is quite a bit of complex and challenging engineering work which creates more surface area in browsers in return for privacy.
It's probably worth it on the whole, as long as it's done carefully, but for someone who cares a bit about simplicity, efficiency and being able to comprehend what's happening: will there be an opportunity just to disable targeting advertising (and thus the code paths and logic associated with it all)?
(I'll try to reason about this and work it out myself from the documentation; I do already see that there's a "Disable Ad Interest Groups", but I don't know if that's quite it, yet)
I agree with you, if I type in a search term from location X, advertisers have enough targeting to serve me a relevant ad which I'm totally fine with.
[0] https://www.gov.uk/cma-cases/online-platforms-and-digital-ad...
[1] https://www.gov.uk/government/news/cma-lifts-the-lid-on-digi...
If there's an investigation it's because governments are starting to realise they are not totally fine with it.
You may be, but it's off topic. The topic is the UK.
Abuse of market position, free markets, monopolies, oligopolies, pools, trusts. Impoverishment of one nation's economy in favor of another, tax evasion, unfair business practices.
Individuals don't care about these things, but governments do.
i am not fine with that, because that's just a step from sellers offering a personalized prices for me, trying to squeeze everything out of me.
I was meaning to differentiate between a vanilla search vs tracking data augmenting the search/ad delivery process.
Obviously the search term itself is used, and often location is required for queries such as "taxis near me". It's also essential for advertisers to know if you're in their geographical market.
That's just my opinion on what kind of data is OK to share, the point was that no further information is generally needed. The CMA reports highlight the fact that insidious tracking and the subsequent ROI tends to cement the dominant position of those who use it.
He means this on a global scale: https://www.bbc.com/news/technology-18595347
And they were (are) using only User-Agent. Now imagine that google calculates some "wealth" score back to online shops, insurance companies,...
Google's search ads work so well because Google has very good and reliable demographic data (i.e. "personalization") and the users literally them what they're looking for.
None of that requires tracking.
Related to this: Google has a lot of influence regarding web standards.
For instance, you cannot make an appointment with the California DMV without using google services. Just accessing the website will try to log you into google.
[Google] wants to replace [3rd party cookies] with new tools that give advertisers more limited, anonymised information such as how many users visited a promoted product's page after seeing a relevant ad - but not tie this information to individual users.
Here's the Chromium page about the "Privacy Sandbox":
https://www.chromium.org/Home/chromium-privacy/privacy-sandb...
Quote:
We believe ... the web’s users can access that information freely because the content creators can fund themselves through online advertising. That advertising is vastly more valuable to publishers and advertisers and more engaging and less annoying to users when it is relevant to the user.
In other words, they still want to know as much as possible about the users.
https://github.com/google/ads-privacy/tree/master/proposals/...
They prob want to replace cookies with something that gives them the same functionality but not have to deal with cookie policies in the EU.
Is not hard to imagine that they are getting rid of Cookies because they now have other ways of getting the data, and getting rid of Cookies would make things harder for competitors that don't own the full pipeline.
Even with that, I'm sure that the engineers working on Chromium/Google Chrome are being told that they are removing Cookies for the greater good and don't have insights into the longer pipeline that we're now seeing the middle off.
Most of that can be leveraged by google to replace cookies.
Even if you replace cookies with something else (localstorage or whatever), you're still on the hook for all the rules both here and in the GDPR with regards to personal information and informed consent.
Remember that the 'cookie law' says
> Exceptions to the obligationto provide information and offer the right to refuse should be limited to those situations where the technical storage or access is strictly necessary for the legitimate purpose of enabling the use of a specific service explicitly requested by the subscriber or user.
So you don't need a cookie banner for a login session cookie, or a cookie that stores preferences the user actively selected. But you _do_ need a cookie banner, and a way to opt out, for all kinds of user tracking, both first- and third-party.
Of course, IANAL. Just angry at advertisers for muddying up this issue.
You might think otherwise, but for these cases I think there should be opt-in mechanisms instead of the assumption what people want. If they are that off with their ads...
1. https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELE...
Most importantly, the ePD applies to cookies even when they are not personal data. Your post made it sound like the only concern is identifying users via cookies. That is not the case. Non-identifying cookies would not incur obligations under GDPR, but they do incur obligations under the ePD.
If you log into a Google account in Chrome, you log into Chrome itself with that Google account. Then Google tools can use your account as an identity signal. This is already available as a feature in Google Ads and Google Analytics. The name of the feature is "Google Signals" - they've posted documentation on how it works and what it does.
This identity signal works cross-domain and cross-device. Google is working to kill off other identity signals with those capabilities that would be available to competitors, such as third-party cookies. (This is also why I believe them when they say they're actively working against browser fingerprinting.)
Basically, the only tracking solution that becomes viable is their own.
No user-agent strings is interesting to me, to me they seem like a minor concern privacy-wise, and doesn't a large portion of the web use them to maintain compatibility between browsers, detect your OS for downloads, and etc?
To detect the OS for download, either JS will have to be used, or the new granular Client Hints[0], specifically User-Agent Client Hints[1]. You can use Sec-CH-UA-Platform and Sec-CH-UA-Arch to figure out the OS and CPU Architecture of the client. However, browsers may refuse to honor this, depending on the privacy budget.
Seeing User-Agent go away will be a net positive for web compatibility. That it also improves privacy is just a nice-to-have.
[0]: https://developer.mozilla.org/en-US/docs/Glossary/Client_hin...
For me personally I use different browsers for different things, and if I don't want to be tracked and have browsing artifacts like cookies correlating data together and tracking me, I just go incognito and call it a day. Google's attempt to re-design how browsers work at this fundamental level is welcomed, but that means other browsers have to do the same, which I don't see happening. Firefox rarely copies Chrome features (or Chrome's anti-features).
I've got a PhD and I find the lengths required to have some modicum of privacy on the internet truly insane, and at times, a little technologically annoying (especially when you have to debug which random script broke a particular page). The other trouble is of course apps: android is just a _dumpster fire_ and every MS product causes an awful lot of blocklist entries to mobile.pipe.aria.microsoft.com.
_Something_ new would be nice, but I fundamentally think that Google is the most conflicted company possible to deliver it.
> Firefox rarely copies Chrome’s features
I hope this is sarcasm. Sadly, FF is effectively pushed by market forces to adopt most Chrome features, down to the extension mechanics. This has always been true for every player, to be fair, it just so happens that Chrome is the current reference.
Yes you may not leave traces or history on the machine, but cookies can be correlated to other activity. For example, when logged into a Google account, if the site uses Google Analytics and you are logged into Google, then Google can build a profile of you and target ads at you based on your activity.
Also: Browsing artifacts may be left on a machine anyway due to swap (on Linux) or Windows' memory paging file. This is why I advocate for using something like TailsOS[0] if you don't want to leave a trace and be as anonymous as possible.
Yes, TailsOS is a lot of overhead for most people, but worth it if you want your privacy real bad.
Cause it sure looks like so. If I remember correctly, other browsers have also removed support for 3rd party cookies too
"protecting" would be better
Google identifies users using Chrome via other means than cookies (Google Signal etc) in a way that no other competitor can or is able to access, and therefore is now pushing to remove the ability for anyone else to do it who doesn't control the browser.
Guarantee you if Google was forced to stop sharing data from its browser business to its ad business, it would not be pushing to remove cookies.
Edit: One final note is that it's a shame because there really is an opportunity to create a new alternative to cookies, better suited to the risks posed by modern tracking methods. But Google leading the charge to dictate how you can be tracked should be a huge red flag for you.
Anyways, google has been writing a spec for “first party sets” to help replace the use of a 3rd party cookie to connect domains you own. https://github.com/privacycg/first-party-sets
I believe this will need to be implemented before Chrome moves aggressively against 3rd party cookies.
"Google will effectively control how websites can monetise and operate their business,"
"This means that any business that buys or sells advertising will be reliant on Google for a part of the process, whether they like it or not."
Guess what the state is, now, 1 year later ... the declarativeWebRequest API is still on hold; and it's not supported outside of Beta Channel, and there are no plans to move it to stable. [2] Its documentation still states the same as it did half a year ago:
"Note: this API is currently on hold, without concrete plans to move to stable. Use the chrome.declarativeWebRequest API to intercept, block, or modify requests in-flight."
... which effectively means that there's no way to block or modify request/response headers in Manifest V3, which is essential for Adblockers because they tend to override the Content-Security-Policy and remove headers like "Cookie" or "Set-Cookie" etc.
And now we have the Chrome Web Store moving ahead with the Manifest V3 rollout. [3]
[1] https://developer.chrome.com/docs/extensions/reference/webRe...
[2] https://developer.chrome.com/docs/extensions/reference/decla...
[3] https://blog.chromium.org/2020/12/manifest-v3-now-available-...
The whole point of declarativeNetRequest is to make it safer and faster to use adblockers. The tradeoff is fewer rules and less expressivity. As someone who couldn't live without an adblocker, I appreciate it and look forward to it because it removes a massive security risk (image the carnage if one of the major adblocker extensions get compromised).
Your citations do not support the claim that "It's only a matter of months before Adblockers won't work anymore.".
There's plenty of reasons to not like Chrome, but this is not a justification for spreading FUD.
In fact, the very blog post you cite states that "There is not an exact date for removing support for Manifest V2 extensions" and has a quote from the Adblock Plus team, praising the collaboration with Chromium.
Yes, I agree. I removed that part of my statement.
Still, the declarativeWebRequest API does not allow to filter out tracking-related headers from incoming responses or sent requests.
I mean, you cannot declare the rules via the RequestMatcher and know what headers are going to be sent or received in advance, as the API expects full declaration of all protocol schemes and host suffixes; which is very bad if a website can pretty much do whatever it wants when it executes js code.
> In fact, the very blog post you cite states that "There is not an exact date for removing support for Manifest V2 extensions" and has a quote from the Adblock Plus team, praising the collaboration with Chromium.
You know that eyeo GmbH were the ones with the "Acceptable Ads" initiative that are literally forcing websites to pay them money so that their ads continue to work, right? Personally, I would take their comment with a grain of salt.
For their whitelisting-ads use case the API works; for the use case of uBlock Origin et al - it doesn't.
Yes, but this is a good thing. This kind of intrusive logic with full access to view and modify request data does not belong in a Chrome extension that can be updated by a single account at a moment's notice through the store, with zero peer review (unlike changes to Chrome itself).
This is much safer to implement in the browser core. If the declarativeNetRequest API is insufficiently expressive, then it needs to be improved to handle those use cases instead of sticking with the old way.
You're welcome to prove me wrong by implementing an Ad-Blocker that can override Content-Security-Policy by default; and incrementally allow to execute things.
As Google continues the Manifest V3 rollout, their priorities seem to be not on that specific degradation of this featureset; and that was all I'm saying.
I've been working on my own Extension for the last week [1], and I had to switch back to Manifest V2 because there was no way to create an Adblocker that's based on a concept that allows the user to select what should be executed; as the Content-Security-Policy header couldn't be set and malicious HTTP headers couldn't be filtered for all domains by default.
You claim that the new API is a full replacement of the featureset, so rather than saying that I'd welcome a hint to some examples or other evidence I'm probably missing here.
Chrome is a shitty browser. Google is a shitty company. Stop defending shit.
> You are the one spreading FUD here by claiming that the new API is so much better than the old one.
This kind of language is not welcome here. Please have a look at the guidelines: https://news.ycombinator.com/newsguidelines.html
I read the design documents and I'm convinced that the new API is a big improvement, and that the design decision was made in good faith and for technical, not political reasons.
The uBlock Origin developer is very trustworthy, but there's so many things that can go wrong - like a workstation, repository, signing key or account compromise.
It's a massive single point of failure. Anyone who can sneak malicious code into the extension has instant and unlimited access to millions of browsers and sensitive personal and company data.
A project like Chrome has extensive processes and safeguards to prevent this kind of compromise, including strict code review. As far as I can tell, uBlock has no code review process[1]. We need to move away from such points of failure no matter how well-intentioned they are.
[1]: https://github.com/gorhill/uBlock/commits/master
The webRequest API also forces a lot of IPC overhead and serialization in the runtime, no matter how fast the extension itself is. uBlock is very, very fast - but it's still a lot slower than native code in the core.
I'm not against adblocking or in favor of stripping away adblocker features, but why not implement it right in the core of the browser, where it belongs? Chrome is open source, if Google doesn't want to do it, another vendor or open source project most certainly could.
Find me just one person who is legimitately interested in these "legitimate interest" tracking things.
Super annoying!