Formal Objection to the Decision Regarding A “Tri-­part choice requirement for users agents.”

Dear Director, TPWG Chairs, and all participants in the TPWG,

In response to the Chairs' decision on the tri-part choice requirement 
for user agents, we are submitting the following formal objection.

Formal Objection to the TPWG Chairs’ Decision Regarding A “Tri-­part 
choice requirement for users agents.”

For the reasons set out below, we object to the Chairs' decision with 
regard to a "Tri-­-part choice requirement for user agents," (see 
http://lists.w3.org/Archives/Public/public-tracking/2012Sep/0197.html)and respectfully 
request the Director and the Chairs to reconsider and instead adopt 
Option B, the tri-­-part requirement, as the decision of the working 
group. In sum, Option B is the only option that meets the group’s aims 
and the requirements of the charter, because it goes to ensuring users 
have an adequate range of choices available to them.

The decision followed a working group poll in which two options were 
presented, and working group participants were asked to state objections 
to either of the options. Option A was "silence"; User Agent's (UAs) as 
currently defined in the draft specification would not be required in 
the specification "to provide a DNT:0 option that is equally easy to 
turn on as DNT:1." Option B, on the other hand, would create mandatory 
guidelines requiring that a "user agent MUST require equal effort to 
configure their agent to each of a minimum of three choices for a Do Not 
Track preference: 1, 0 or unset."

This is a crucial issue for the specification, as it goes to the 
fundamental problem of whether a DNT signal sent by a UA, and received 
by a server, reflects an informed, deliberate choice by the user. It has 
implications for the credibility and ultimate adoption rate of the 
specification.

The Co-­-Chairs' decision misunderstands or mischaracterizes the 
objections to Option A, and the strength thereof, and implements a 
decision that is inconsistent with the aims and charter of the TPWG, and 
that is not conducive to the success of the specification.

The TPWG’s mission includes “improv[ing] user privacy and user control 
by defining mechanisms for expressing user preferences around Web 
tracking,” and the group’s charter calls for a “simple 
machine-­-readable preference expression mechanism.” The goal of 
simplicity and the infeasibility of complex user interfaces necessitate 
a limiting of users’ choice. This is natural. However, in limiting 
choices, we limit information, and run the risk that a user’s choices 
are inadequately informed. Moreover, with the ability to limit users’ 
choices often comes centralized control that can be misused to influence 
markets. This has been the case in the past in many markets, including 
telephone service, web browsers, and automobiles. Recall Henry Ford’s 
famous quote: “Any customer can have a car painted any color that he 
wants so long as it is black.”

So, therefore, the problem becomes one of degree. What is the minimum 
availability of choices that is needed to ensure users have a real 
choice beyond just “black” or nothing? The proposed option, Option B, 
addresses this issue in a minimally prescriptive way that allows user 
agents extremely broad latitude to craft user interfaces, while meeting 
simple guidelines around the relative availability of alternative DNT 
choices. Option A leaves the issue without guidance, and entirely to the 
whims of user agent designers who, for one reason or another, may not 
care to give the user much control. Therefore, the Chairs erred in 
deciding to adopt Option A over Option B.

The Co-­-Chairs' decision was organized according to "themes" that were 
elucidated from participants' comments. Although we disagree with the 
framing of the "themes," for clarity we follow their format and address 
themes individually.

_A. Support of EU and other Compliance Regimes_
Under their response regarding this theme, the Co-­-Chairs rightly point 
out that legal compliance is not a goal of the TPWG. However, the 
Co-­-Chairs missed the point of various participants' comments regarding 
consent: that ensuring users are able to express an informed and 
meaningful choice is a legitimate and necessary goal, and not just a 
matter of legal compliance. The TPWG has already discussed and is near 
consensus on the view that DNT must reflect users' choice and that, for 
example, DNT on by default would generally not be reflective of a user's 
choice. One pending option in the TCS document provides explicit 
guidelines for achieving explicit and informed consent from a user. 
Option A does not provide UA's adequate guidance regarding what is 
required to achieve users' explicit and informed consent. Only Option B 
provides such guidance.

_B. Incomplete specification of user agents_
On this point – that without specification UA's "may implement DNT in 
undesirable ways" – the Co-­-Chairs' rejected objections to Option A 
(and support of Option B) under the rationale that the TPWG charter does 
not allow specification of UI. Under Option B, however, does not specify 
UI, but would provide guidelines that define user experience, as allowed 
under the charter. The Option says only that the three states must 
require equal effort; it does not specify how this is to be achieved, or 
otherwise in what form the choices must be presented. It does not even 
require that they be presented together. Rather, it requires that each 
be accessible with equal effort. This guideline is intended to ensure 
the user's choice rises to the level of informed consent required by the 
standard. This intent was raised in the objections to Option A, but 
disregarded, or misinterpreted by the Co-­-Chairs in making their decision.

_D. Ability to validate against the spec_
Here the Co-­-Chairs side with comments suggesting Option B would add 
untestable elements to the specification. If testability is a 
requirement for elements of the specification, then much of the 
specification fails. Not once yet has testability been raised as a 
requirement for any element of the TCS, and no such requirement exists

in the charter. Although testability is an obvious need for true 
technical specifications, as the W3C ordinarily produces, testability 
under many elements of the TCS document is virtually hopeless. For 
example, elements of the specification rely on such vague and untestable 
standards as "infer with a high probability," "confidence," "reasonably 
necessary," "reasonable efforts," and so on. To apply a testability 
standard to only UI guidance, but not other crucial elements of the 
specification would be an untenable double standard.

_E. Ability to express actual user preference_
Here, in rejecting the argument that the three state requirement is 
essential to ensuring fully informed consent, the Co-­-Chairs reframe 
the issue, stating that "the input received does not argue why a 
3-­-state UI is the only way to ensure that users can express 
preferences." The Co-­-Chairs go on to state that they "do not conclude 
that all user agents must offer all preferences for all users as the 
only way" for users to find ways to express their preferences. The 
decision misses the point. First, Option B does not require a 3-­-state 
UI, but only that UA's "require equal effort to configure their agent to 
each of a minimum of three choices." But more importantly, Option A, 
silence, does not ensure that users have the opportunity to fully 
understand their choices and make an informed choice. It is not a matter 
of users finding ways to express their preferences. It is a matter of 
ensuring that users have the opportunity to express adequately informed 
choices.

_F. Simplicity, Innovation and Differentiation for User Agents_
In agreeing with objections to Option B, the chairs state that 
"specifying a user interface is out of scope by our charter." This 
mischaracterizes Option B. The charter states that "while guidelines 
that define the user experience or user interface may be useful (and 
within scope), the Working Group will not specify the exact presentation 
to the user." Option B does not specify the exact presentation, but 
rather states guidelines that must be met to ensure users have adequate 
access to the full range of choices available under the standard. This 
is far from specifying UI, and leaves considerable latitude to UA makes 
to determine how the guidelines will be met.

_G. Unbiased Representation_
The Chairs dismiss concerns about bias and fairness on the part of UA's, 
saying they "have not seen evidence as to why requiring all three 
options with equal effort satisfies the goal of fairness." It is 
self-­-evident that in many cases companies with competitive interests 
at stake, and that stand to benefit from widespread DNT usage, also 
control users' experience of the Internet. Requiring that UA's satisfy 
guidelines with regard to equitable availability of all DNT options is a 
minimally prescriptive means to try to limit the extent to which 
competitive interests are not at play in UA’s influencing users' DNT 
choices, and that therefore a DNT signal reflects a truly informed choice.

_Conclusion_
For the reasons above, and in the interest of producing a specification 
that is credible, fair, and that will garner widespread adoption, we 
respectfully request the Director and the TPWG Chairs to reconsider this 
decision with regard to a tri-­part choice requirement for user agents.

We thank you for your consideration.

Respectfully,

Jason Bier, ValueClick
Alan Chapell, Chapell & Associates
Brooks Dobbs, KBMG
Marc Groman, NAI
Dan Jaffe, ANA
Peter Kosmala, 4A’s
Lou Mastria, DAA
Dick O’Brien, 4A’s
Clark Rector, AAF
Brendan Riordan-­-Butterworth, IAB
Rachel Nyswander Thomas, DMA
David Wainberg, NAI
Shane Wiley, Yahoo!
Jeff Wilson, AOL
Kimon Zorbas, IAB Europe

tri-part_objection_final