- From: David Singer <singer@mac.com>
- Date: Mon, 23 Oct 2017 14:48:11 +0800
- To: "Aleecia M. McDonald" <aleecia@aleecia.com>
- Cc: "public-tracking@w3.org (public-tracking@w3.org) (public-tracking@w3.org)" <public-tracking@w3.org>
> On Oct 23, 2017, at 10:53 , Aleecia M. McDonald <aleecia@aleecia.com> wrote: > > >> On Oct 22, 2017, at 5:45 PM, David Singer <singer@apple.com> wrote: > > […snip…] > >>> To get round the transparency issue of parties having different codings for purposes, we could insist there is a mandatory field in the TSR (which is designed to be dynamic i.e. it can depend on the whole DNT header) which contains a URI to a page that explains the specific purposes the user has agreed to in human readable terms. >> >> yes, I agree that the TSR should somehow explain the DNT extensions, if we go this route > > […snip…] > > +1 on having a URI point to a human-readable decoder ring. The text on the page MUST match the text as presented to the user when requesting consent, and MAY contain additional information. I had not got as far as a URI. I am having trouble parsing what you wrote. > It occurs to me that URI then *becomes* the unique prefix that I’d been looking for. What URI? I had floated that during registering the exception, you could supply the extension-string that will be sent with the DNT:0, so for example registerSiteSpecExcept( forDomains: “ads.com;R”, “audits.com;Xs” ) causes DNT:0;R and DNT:0;Xs to be sent. Mike is saying that ‘somehow’ the TSR of ads.com explains what an X extension-string allows, so in this case, the TSR on ads.com explains “X” and on audits.com explains “Xs”. Not sure how. > I hear Shane’s scoping argument, but I think it will save great angst to avoid accidental collisions. If we know “a:1 b:0 c:1” goes with www.iab.com/dnt.html, vs. www.eff.org/dnt.html, there’s less chance someone will get confused while writing / maintaining code. Given large-scale privacy invasions and lawsuits are possible outcomes of getting it wrong, I’d like to support defensive programming. This is another place you lose me. > > Bonus points if the URI *itself* contains version information, since these things may change over time, and orgs will certainly want to manage what it is the “a” in a:1 meant at the time consent was given/revoked. There are many ways to do so, hence only “bonus points." > > Translated, I’m suggesting a MUST for a URI that accompanies the extension, plus an example that shows www.example.com/dnt1017.html with a date or version number embedded and discussion of that as a best practice, rather than even a SHOULD. > > Aleecia > David Singer singer@mac.com
Received on Monday, 23 October 2017 06:48:48 UTC