The Technical Architecture Group (TAG) at the W3C has concluded that Google’s Topics API “appears to maintain the status quo of inappropriate surveillance on the web” and stated that it does “not want to see it proceed further”.[1]
Some of the objections made by the W3C register as fairly reasonable. For instance, those based on the lack of user controls associated with opting into Google’s business-facing solution
We likewise agree with the TAG when it sates: “[c]entralisation is a big concern here. This proposal makes it the responsibility of browser vendors (a small group) to determine what categories of user are of interest to advertisers for targeting. This may make it difficult for smaller organisations to compete or innovate in this space. What mitigations can we expect to see for this?” [2]
However, we would argue, contrary to the W3C, Topics is not a mechanism for “unwanted tracking and profiling.”
Google’s Topics API functions by assigning interest-based attributes based on each browser’s activity during the prior week, which Google then shares with third-party ‘callers’.
To describe the generation of highly generalised interest groups as “profiling” specific individuals represents quite the stretch.
The W3C’s concern that Sensitive Information, such as video files, location of financial data (see link for W3C’s comprehensive list of examples) could be compromised by revealing that a deidentified user’s top five topics included “Acting & Theatre” and “Family & Relationships”, is rather implausible.
MOW would agree that Topics API is unfit for purpose and should not be taken any further, but for the simple fact that it does not work.
Testing suggests that Google’s classifier is highly inaccurate. CafeMedia found, for instance, that 17% of “Automotive and vehicles” pages were classified by Chrome as “News”, 12% of “technology and computing pages were classified, again, as “News”, and 10% of “Health and Fitness” pages were classified as “Reference”. [3]
Topics is, thus, a bad idea, that has been poorly executed. Few of its shortfalls have much to do with privacy.
[1] Early design review for the Topics API · Issue #726 · w3ctag/design-reviews (github.com)
[2] Early design review for the Topics API · Issue #726 · w3ctag/design-reviews · GitHub
[3] Topics API data: what a million impressions tell us – Web Wide Open : Web Wide Open