- From: Mike O'Neill <michael.oneill@baycloud.com>
- Date: Tue, 23 Jul 2013 14:25:54 +0100
- To: "'Rigo Wenning'" <rigo@w3.org>, <public-tracking@w3.org>
- Cc: <rob@blaeu.com>
Rigo, You do not need to single out (use a UID) to count objects or detect unique visitors. These things can be done in other ways as I have shown. UIDs do not help with the recognition of bots either. There may be other things that need UIDs but we have not been told what they are. We should have the reason for them, and therefore why there needs to be a permitted use, clearly explained. Mike -----Original Message----- From: Rigo Wenning [mailto:rigo@w3.org] Sent: 23 July 2013 13:47 To: public-tracking@w3.org Cc: Mike O'Neill; rob@blaeu.com Subject: Re: Change proposal: new general principle for permitted uses Mike, On Tuesday 23 July 2013 08:10:46 Mike O'Neill wrote: > If user profiles are not used or built then why the necessity for > singling-out? Why have we not been given a definitive reason for > collecting/using UIDs? As far as I have understood, one needs to know the single user/device to be able to count on objects. If one doesn't know whether it is the same user or device, counting doesn't really work as it may e.g. be a bot. The second reason is that the audience measurers could make all of this up and just assert the result. So instead of having a burdensome inquiry with measures and beacons and things, one would just make things up. If the client has suspicions, they would say: No, we deleted all for privacy. Given that outreach determines ad-pricing, there needs to be a safeguard against manipulation. This safeguard is done with reproducible data sets so that one can prove the correctness of the results to the client. I think, on the long run, we can do better than we do now and without UIDs, but I personally believe this would overstretch the industry if written positivistically into a specification. The bar for implementing the TCS would become too high. But this is surely something we could imagine for an incremental improvement (version 1.1) for the Specification. --Rigo
Received on Tuesday, 23 July 2013 13:27:02 UTC