How the attribution cartel misleads advertisers: Q&A with Rick Bruner

The attribution cartel at W3C has lots of questionable decisions to worry about, but just for a minute let’s ignore implementation issues like centralizing on US-based platform companies or the environmental impact of all that additional processing of data for obfuscation purposes. Even if the Big Tech companies managed to go international and sustainable with this thing, they still have two big fundamental problems.

• Even if the mathematical properties of the Attribution proposal make it infeasible to reidentify users using just the attribution data, in the real world this new form of tracking would add user privacy risks—byhiding the process of “stealing attribution” and creating more incentives to collect data in riskier ways.

• The attribution cartel can hide the halo effect of advertising on trusted sites, and drive more ad budgets to search, social, and app store ads.

One of the responses I got to the second one was, well, at least the Attribution proposal is something, and something is better than nothing. I doubted that, since in general you’re better off knowing what you don’t know, so I askedRick Bruner from Central Control some questions.

Here’s what we came up with: Q&A: Marti & Bruner discuss the problematic W3C proposed measurement standard ‘Attribution Level 1’ — Central Control. And it turns out that the situtation is even worse than what I covered last month.

• Even if fraud operators are unable to exploit the complexity of the Attribution proposal to add more costs, risk and noise (which fraud hackers have been able to do every other time complexity increased)…

• Even if the Big Tech companies somehow decide not to unfairly favor their own products (for the first time ever)…

• Even if the extra processing and resource consumption required for the Attribution proposal’s mathematical goals turns out to be affordable and politically acceptable…

• And even if the design is bug-free and browsers implement it perfectly…

Then the reports that the attribution cartel can produce will still be actively misleading.

For what it’s worth, I don’t think the developers of the Attribution proposal are deliberately putting deception into the design of the system. I’m assuming that the developers are well-intentioned and that the design and code somehow achieve their goals perfectly. But it’s still important to consider the Attribution proposal as it would interact with theWeb as it is in the real world, not in isolation.

And yes, I asked if it’s possible to tell (1) a pizza ad that actually sold someone a pizza apart from (2) a pizza ad placed using speech to text, targeting people who are already talking about pizza. Rick’s answer is informative—please read the whole thing.

Update: [attribution] Methodological concerns regarding Attribution Level 1 and causal measurement by Rick Bruner, on the public-patwg mailing list at W3C. (AlsoThe Measurement Problem Inside the W3C Attribution Standard on LinkedIn)