You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
With FledgeEnforceKAnonymity gradually rolling out to Chrome stable per docs, I've already observed several instances of confusion and lost productivity when debugging unexpectedly null runAdAuction() results for tiny, test-only buyers due to k-anon filtering.
While the solution so far has been to just disable the feature on the command line while testing, developers should receive a debug signal from the browser perhaps in the DevTools Interest Groups panel or in the JS console when this filtering occurs, especially if this filtering removed all otherwise accepted ad candidates. If we expect this to happen frequently, to avoid console clutter the browser might log each filtered candidate in the IG panel and only once in the console iff there was no winner and k-anon filtered at least one candidate.
Adjacently, I had to dig deeply to find #867 (comment) indicating that production observability of k-anon may be done through private aggregation reporting. It would be helpful to include this detail in the k-anon developer guide (although as a nit I believe it is the browser, not the seller, that rejects the bid in this case).
The text was updated successfully, but these errors were encountered:
I'm doing some testing in Canary. When starting Chrome sans cmdline flags or with --enable-features=FledgeEnforceKAnonymity, I see browserSignals.kAnonStatus="notCalculated" in reportWin. I'm expecting to see kAnon enforced. Am I missing a magic incantation?
Hey David, I haven't been able to recreate the magical incantation of flags that Chrome used to start enforcing k-anon on my test browser profile back in June. One suggestion might be to include FledgeConsiderKAnonymity alongside FledgeEnforceKAnonymity though.
With
FledgeEnforceKAnonymity
gradually rolling out to Chrome stable per docs, I've already observed several instances of confusion and lost productivity when debugging unexpectedly nullrunAdAuction()
results for tiny, test-only buyers due to k-anon filtering.While the solution so far has been to just disable the feature on the command line while testing, developers should receive a debug signal from the browser perhaps in the DevTools Interest Groups panel or in the JS console when this filtering occurs, especially if this filtering removed all otherwise accepted ad candidates. If we expect this to happen frequently, to avoid console clutter the browser might log each filtered candidate in the IG panel and only once in the console iff there was no winner and k-anon filtered at least one candidate.
Adjacently, I had to dig deeply to find #867 (comment) indicating that production observability of k-anon may be done through private aggregation reporting. It would be helpful to include this detail in the k-anon developer guide (although as a nit I believe it is the browser, not the seller, that rejects the bid in this case).
The text was updated successfully, but these errors were encountered: