Fingerprinting about gpc-spec HOT 4 CLOSED

FWIW, I'm mostly fine with @martinthomson 's suggested text, with the following change:

Change:

This represents a privacy cost that is justified by the privacy advantage provided by sending the signal.

to

Depending on the browser and implementation, the GPC signal may impose a privacy cost, though one intended to be justified by the privacy advantage provided by sending the signal.

I think this change is important bc the previous sentence recognizes that in some deployments the GPC signal doesn't provide additional fingerprinting-signal. My suggestion aims to just carry that recognition into the next sentence.

If that all sounds good / no one opposes it, i'll update the PR and merge

from gpc-spec.

@pes10k further revised the language.

from gpc-spec.

@martinthomson The changes we now have merged in were agreed upon as potentially needing additional text beyond what is there. Would you like to suggest something that would improve clarity or close this issue?

from gpc-spec.

I'm still unhappy with the "we tried not to make it worse, really" implications of that last statement.

Why not just say something like:

Exposing a user's preference (in the HTTP header field or navigator object¹) potentially divides users into two groups in a way that might increase the information available for fingerprinting. This extra information is available unless the signal perfectly correlates with other signals ². This represents a privacy cost that is justified³ by the privacy advantage provided by sending the signal.

from gpc-spec.