What does HackerNews think of turtledove?

TURTLEDOVE

Language: Bikeshed

It'll be interesting to see how this goes. Google and Mozilla+Meta each have competing standards.

https://github.com/WICG/turtledove

https://blog.mozilla.org/en/mozilla/privacy-preserving-attri...

To my knowledge, Mozilla's design is the only one where someone other than the browser collects & reports on click activity, and with a fairly trustless anonymizing double blind strategy for those intermediaries.

you can target an Economist reader a week later on a different website. If FLoC works, you can still do that.

https://github.com/WICG/floc won't really let advertisers do that, this is what https://github.com/WICG/turtledove is for

(Disclosure: I work on ads at Google, speaking only for myself)

> The issue is that you cannot have good online advertising without a little bit of the stalking.

Perhaps today, but not if we design browser APIs that allow targeting without cross-site tracking: https://github.com/WICG/turtledove https://github.com/WICG/privacy-preserving-ads/blob/main/Par...

In between that claim and the end of the article I describe how I'm working on https://github.com/WICG/turtledove etc to build well-targeted advertising without sending your browsing history to advertisers
> The ads. Aren't. The problem. The stalking is.

My primary work is on https://github.com/WICG/turtledove (discussed in the post), which allows well-targeted advertising without sending your browsing history to advertisers

> their privacy is being violated and their metadata is being sold, in order for advertisers to track them everywhere they go online

This is exactly what I'm working on changing; have a look at the second half of the post?

(Or read https://github.com/WICG/turtledove)

> What about ads, but static and not-tracking?

Coincidentally, my current project involves this Chrome proposal for supporting self-contained remarketing ads without individual tracking: https://github.com/WICG/turtledove

Is there an API that allows me to see, unpack, and set my own cohort? The FloC explainer [0] suggests the basic API just returns a stable hash [0]. I am curious about the space of cohorts. Also, whether there'd be an advantage to choosing a cohort for my task, vs just being anonymous.

EDIT: that would appear to be TURTLEDOVE [1]

[0] https://github.com/WICG/floc [1] https://github.com/WICG/turtledove

The FloC announcement is https://blog.google/products/ads-commerce/2021-01-privacy-sa...

The whitepaper explains the approach: https://github.com/google/ads-privacy/blob/master/proposals/...

1: I don't think it's very common. I have a blogger friend who used to host simple image ads, but then they migrated to substack.

2: The primary method of cheating in print was lying about your circulation numbers. Anyone could easily verify that the ad was run in the agreed on spot, but the question of how many people read your paper/magazine is harder. In the US circulation numbers typically come from the nonprofit Alliance for Audited Media (formerly the Audit Bureau of Circulation). See https://en.wikipedia.org/wiki/Newspaper_circulation

3: The world is moving in that direction, but not especially because advertisers or publishers are pushing for it. Instead it's consumer advocacy groups and browsers. My primary project at work right now is (along with a lot of other people) figuring out how we can serve personalized ads without cross-site tracking (https://github.com/WICG/turtledove).

4: An ad network is a middleman between advertisers and publishers. Directly connecting pairs of publishers (P) and advertisers (A) scales at O(PA) while publishers and advertisers both connecting to a network scales at O(P+A). AdSense would be an example: an advertiser who signs up with AdSense can advertise across all its publishers without needing to make 1:1 deals.

(Still speaking only for myself)

> if I look up the specs on a pair of Bose QC-35's, I won't get internet-stalked with ads for that specific pair of headphones for the next 3 weeks whenever I pull up a weather report

The WSJ seems to be describing https://github.com/WICG/floc, but the kind of remarketing you're describing is what https://github.com/WICG/turtledove is intended to support. Advertisers would still be able to run that kind of personalized ad, but the browser API would not allow them to learn your browsing history in the process.

(Disclosure: I work on ads and Google, speaking only for myself)

My parent claimed this was possible to do with link decoration and first party cookie matching, and I'm saying it isn't.

I do agree this is possible to do with fingerprints, though (a) all the browsers are trying to prevent fingerprinting and (b) a reputable ad company would not use fingerprints for targeting. This is my understanding of why Google is putting so much effort into https://github.com/WICG/turtledove

(Still speaking only for myself)

It sounds like you're describing https://github.com/WICG/turtledove, which Google is definitely working on.

(Disclosure: I work on ads at Google, including the seller-side of Turtledove. Speaking only for myself.)

> I like your post. It does conflate Privacy Sandbox with one of the specific proposals, however: [...] https://github.com/WICG/floc

Thanks for the clarification Jeff.

> Another component of Privacy Sandbox is Turtledove (https://github.com/WICG/turtledove), [...]. Which groups a user is in is maintained entirely by the browser, and never sent to the server

Interesting. I'll plan to read up on this tonight and amend my post from what I learn.