Re: ACTION-390: alternative UA affordances for DNT choice

Thanks David. A few things come to mind after reading your and Ed's
examples.

The radio app running on a computer in your example can be seen as an
extension of the radio service's UA and therefore compliant so long as it
otherwise meets the proposed guidelines. If you don't feel that the proposed
language address your use case, I'd be happy to work with you on language
that addresses that concern.

That said, It seems imprudent to spend too much time constructing a standard
around a hypothetical edge case unless there's a compelling benefit and
little risk of creating a loophole.

I'm sure that regardless of how this standard is constructed, there will be
use cases that don't fit perfectly within the spec. This is undoubtedly an
area that is rife for future learnings. If having these guidelines in place
causes unnecessary heartburn for some with no demonstrable benefit for
consumers, we can adjust down the road ­ particularly after consumers obtain
more of an understanding of what DNT does. But starting the DNT experiment
with virtually no disclosure requirements on UA's seems less than ideal,
given our shared goal of ensuring informed consent.

Alan

From:  David Singer <singer@apple.com>
Date:  Thursday, April 18, 2013 11:58 PM
To:  Alan Chapell <achapell@chapellassociates.com>
Cc:  "Edward W. Felten" <felten@CS.Princeton.EDU>,
"<public-tracking@w3.org>" <public-tracking@w3.org>
Subject:  Re: ACTION-390: alternative UA affordances for DNT choice
Resent-From:  <public-tracking@w3.org>
Resent-Date:  Fri, 19 Apr 2013 03:58:54 +0000

> 
> On Apr 18, 2013, at 21:30 , Alan Chapell <achapell@chapellassociates.com>
> wrote:
>>> No one is suggesting your strawman -- that the user was not informed *at
>>> all*.  They are suggesting that the user may be informed by something other
>>> than the UA itself, and that the DNT signal still result from an informed
>>> choice.
>> 
>> Can you provide some examples of how this would work for your UA's?
> 
> I am sure that my imagination is poor compared to the products people envisage
> or invent, but I can try.
> 
> As Ed suggests, audio-only UAs are an obvious case.  Imagine an internet radio
> -- it has audio out, and a network connection.  When it is running, it is the
> UA (the user agent, that makes HTTP requests for audio segments).
> 
> Now, imagine a radio service that puts some or all of its requests through a
> re-direction service, so that they can track people and what they are
> listening to. The radio station gets some revenue in return for allowing this
> tracking.  This means that some users may be interested in a more private
> radio experience, and may desire a DNT setting.
> 
> Finally, imagine that the internet radio appliance is set up using an
> application you run on a personal computer of some sort.  There is no reason I
> can see to prohibit the device from having a privacy setting that enables DNT,
> that is configured and explained in the setup app that users run.  That setup
> app is not a user-agent.
> 
> (There are some practical issues (which are out of our scope), such as that it
> may be prudent to cause the device to do a 'test fetch' to each of its
> stations while the app is running, to see if they are willing to provide
> service with DNT turned on.)
> 
> 
> I feel sure that there are other cases, as well.
> 
> David Singer
> Multimedia and Software Standards, Apple Inc.
>