Hacker News new | past | comments | ask | show | jobs | submit login
Apple critique of Google's Topics API based on bad code (theregister.com)
19 points by beardyw 57 days ago | hide | past | favorite | 9 comments



For those, who don't know, Topics API is a browser API that allows one website to save user's interest in a browser and then another site can retrieve this data. So basically browser helps websites to collect and store information about you and your interests.

It is difficult to understand why browser should help advertisers collect data rather than protect users from them. I hope Firefox won't support it.


"Firefox added [ad tracking] and has already turned it on without asking you" https://news.ycombinator.com/item?id=40954535


If you read the information about that feature, it is about ad attribution. A website can ask your browser "Did you see these ads" and Firefox will send the result to a 3rd party. That 3rd party periodically reports a summary of ad conversion results to the ad buyer, supposedly with "noise" to provide "differential privacy".

Probably some way to to glean some info, but that is absolutely not what people are claiming it is.

Meanwhile you all still use Chrome so what's your complaint?


This is not ok. Browser should help to block ads, not to help display it more efficiently.


This feature does nothing to prevent you from blocking ads entirely, and Mozilla has not signaled that they intend to destroy the ability of extensions like ublock to run arbitrary code to decide what to block in the way that Google and the Chrome team have explicitly said they will do.

This feature does nothing more than allow advertisers a very basic form of "fifty percent of people who you showed these ads to interacted with them" in a way that doesn't require things like third party cookies. I'm not convinced that it will go anywhere, as Google and Facebook, the majority of the online advertising market, do not want to give up the powers they have currently for attribution that are also great at powerful tracking, and as you show, the public will continue to play word games about how Firefox is "destroying privacy" while continuing to mostly use AND DEVELOP FOR chrome. Or worse, a version of chrome that they pretend is somehow not entirely controlled by google.

If you want to block ads, Firefox is still the obvious browser choice over anything that branches off of Chrome. Firefox will NEVER possibly be pure enough for the people who still for whatever reason just use chrome. It's insanity.


I mostly agree with this, but I think there is also a very pragmatic angle here where we need to acknowledge:

1. Advertising is really the only successful economic model of the web at this point and has been for more or less its entire existence (at least in so much as it had economic relevance to begin with).

2. The current state of play is incredibly bad and we now live in a world where shady organizations can and do buy things like location data to track individuals movements over time. This is a capability that has historically been the purview of intelligence agencies and is now available to anyone. Here is a great example of that in action: https://www.404media.co/heritage-foundation-claims-to-use-lo...

3. The ability to strip individual identities from those datasets and make the level of targeting aligned with something very course grained like interests would be a huge win for privacy without killing the economic model of the web.

So in that sense… putting my feelings of Google aside, I can actually see how you can make an entirely good faith argument that this API is trying to align incentives rather than some sneaky trick.


> The current state of play is incredibly bad

Then browsers should provide less data for fingerprinting, not more. Every new feature that provides data for fingerprinting, should be behind a permission. Then any old features allowing fingerprinting (like reading canvas content or getting GPU name) should be gradually put behind permissions.

What Google does, is it just provides more options for fingerprinting under excuse that someday in future (maybe, just believe us) it will help to improve privacy.

> shady organizations can and do buy things like location data

Why it is not forbidden by law?

> The ability to strip individual identities from those datasets and make the level of targeting aligned with something very course grained like interests

You can show ads matching page content or user's request and it will be in most cases relevant. You don't even need cookies for this. You don't need to know someone's interests. If a person is watching a GPU review, add a link to the store selling this GPU with a discount. If a person types "buy 3090" in Google, put a link to the store in the search results.

Also reality is, in many cases advertisers want to advertise some product nobody really needs. Often ads have a click-through rates below 0.1% or even less. This means that there is no need in such ads in the first place. Instead of trying to sell it maybe you should make something that people will want to buy at a reasonable price.


> Advertising is really the only successful economic model of the web at this point and has been for more or less its entire existence (at least in so much as it had economic relevance to begin with).

This is false.

Self-publishing on the web was successful to the point of dominance for a decade, and remains successful for many. In fact, "pamphleteering" has been successful for centuries without ads:

https://en.wikipedia.org/wiki/Pamphleteer

Economic relevance of idea and knowledge sharing is unmeasured but not irrelevant.


Honestly I don't think this follows at all.

1. Advertising has been so successful arguably at the expense of users. Because it has been made easy, this has distorted the marketplace to the point where it's hard for any other model to thrive.

2. The organizations (principally google) behind this are the shady organizations who sell location data as well as using it for themselves. This is just about them entrenching their market advantage.

3. The ability to strip individual identities from those datasets and make the level of targeting aligned with something very course grained like interests leaves google as the only people who can identify the individuals in those datasets given the stuff they add to their browser. I don't see how that's net positive for users in any way.

Floc[1] was roundly rejected and this topics API is just an attempt to do exactly the same thing and pretend that it's different.

Whenever people make this argument it's worth pointing out that Ad spend in far less personally tracked contexts (eg billboards, static advertising, broadcast ads etc) is massive. Advertisers spend money even if they don't know much if anything about who's seeing their ads.

[1] https://en.wikipedia.org/wiki/Federated_Learning_of_Cohorts




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: