RE: extensions in Determining User Preference

Comments below. I won't belabor this conversation as I suspect there are fundamental differences of opinion here. I would like to express my opinion on this, but some of the language ("paint that pig"?) is not conducive to reaching consensus...

-----Original Message-----
From: Roy T. Fielding [mailto:fielding@gbiv.com] 
Sent: Tuesday, April 08, 2014 2:11 PM
To: SULLIVAN, BRYAN L
Cc: 'Dobbs, Brooks'; 'Walter van Holst'; 'public-tracking@w3.org'; 'Shane Wiley'
Subject: Re: extensions in Determining User Preference

On Apr 8, 2014, at 1:48 PM, SULLIVAN, BRYAN L wrote:

> The existing text reflects that no matter how originated, a signal must represent the choice of the user: "The basic principle is that a tracking preference expression is only transmitted when it reflects a deliberate choice by the user."
> 
> What we must not do, is to limit the mechanisms of choice available to users.

Right.

> When I install a home gateway that sets a DNT header (or removes it, or whatever) according to my choice as the person responsible for setting policy in my house,

Wrong (you can do so for yourself, but not for others).  I suspect nobody
is going to care about misconfiguration of home routers, but the protocol
clearly disallows non-users from setting the policy for others.  If you
want to do that, the right place is in restricting the browser settings
on equipment that you own.

> I am exercising a right similar to that of enterprises to govern use of DNT through their proxies.

More wrong.  No such thing exists for DNT.
[Bryan] Sorry, I disagree.

> I do have an obligation to explain the terms of internet service to users in my home, as enterprises provide employees with codes of conduct and other policies that apply to terms of employment. But beyond that obligation, I must have the freedom to provide internet service as I choose, in my private network.

Completely wrong.  You can try to paint that pig in nice colors, but the
basic fact is that a network provider does not have the right to change
the content of messages to be deliberately misleading.  Drop the messages,
yes, but not change them.  It fails the protocol requirements in every respect.

[Bryan] HTTP is designed to support intermediaries of many types, and for many reasons. Home gateways are probably one of the best examples, as they can play a fundamental role in helping ensure the safety and privacy of users and devices in the private network. This is not a "network provider's" network, unless you are meaning me the owner of the home network. When my proxy or gateway for example adds a safesearch flag to a web search, I am not breaking anything - I am using a particular approach to ensuring consistency of user experience as I deem proper for services through my private home network.

> Users that choose to access internet services through it explicitly or implicitly agree to those terms. There are clearly mechanisms for explicit agreement (similar to how you accept terms of internet service at hotels before getting internet access) that can help ensure that users are aware of and consent to terms of use - and that represents their choice.

Deliberately changing the semantics of messages as they pass through a
network means that the network is not acting as a common carrier.

[Bryan] In my home I am not a common carrier. Further if the change reflects intent, it is more accurate than if it did not occur.

This, of course, has nothing to do with the discussion of plug-ins,
which are part of the user agent as defined by HTTP.

....Roy

Received on Tuesday, 8 April 2014 21:37:02 UTC