W3C home > Mailing lists > Public > public-tracking@w3.org > March 2012

Re: Redirect chains and DNT:0 / Exception:* (ACTION-146 re ISSUE-111)

From: John Simpson <john@consumerwatchdog.org>
Date: Wed, 21 Mar 2012 06:41:39 -0700
Message-Id: <43A00B45-7AEB-4D18-90AC-78F45FEE234B@consumerwatchdog.org>
Cc: "public-tracking@w3.org Group WG" <public-tracking@w3.org>
To: ifette@google.com
I'm still contemplating the full implications of this chain, but initially I think Ian's analysis makes sense.

On Mar 21, 2012, at 5:51 AM, Ian Fette (イアンフェッティ) wrote:

> I will also say I think this same argument holds for ACTION-147. I think at the core of my concern is that I don't think it's actually useful in practice to grant exceptions to some enumerated list. I think it should be a binary (third parties on this site get an exception en-masse [ "*" ] or they don't).
> To be useful to the website, you want to know whether your site is going to work. (And please don't get into the rathole argument of "but I use lynx" or "I have JavaScript disabled" -- we're talking the common use cases here.) If you are getting DNT:1 and you want to request an exception for third parties, presumably you will do something different if you don't get the exception. (Show different ads, request payment, block access, who knows.) 
> It's not clear to me what we are all expecting "request an exception to do". Some people seem to be interpreting it as "check a user's pre-defined list for exceptions", others interpret it as "ask the user with some prompt." I think this needs to be understood if a site is to figure out how they are to request exceptions. The notion that you might try to enumerate some subsets is perhaps plausible (if still unlikely) in the case that no UI is generated, but if UI is generated, you've probably only got one shot. 
> As an example, let's say you want to request an exception for Facebook and Google+ because you use their widgets, as well as your ad network. The user objects to one of these. What are you supposed to do? Especially if you're requesting an exception for N third parties, you've now got 2^n enumerations to try? Ouch. Unless you know which the user objected to, you're SOL (and knowing which the user objected to also has implications on the UI).
> The other option would be to request serially, but this would also be a giant PITA.
> In short, I don't think requesting exceptions for an enumerated list makes sense. I think it ought to be "I request exceptions for third parties on my page" Y/N.
> -Ian
> 2012/3/21 Ian Fette (イアンフェッティ) <ifette@google.com>
> Upon reflection, this is probably just further discussion for ISSUE-111. I also can't seem to find the canonical text that ISSUE-111 is proposing. That said, my understanding of the proposal is essentially allowing for negotiation of (on this site, X can track me) where X is a single third party, list of third parties, or all third parties.
> My main concern is that, as a website author, you may include ads from a given ad network (be that doubleclick, yieldmanager, adecn, or whatever) but have no idea what other third parties those ad networks syndicate to. You want higher quality ads on your site (which presumably translates to more revenue for the site), so you request an exception for the third party ad network you use directly. But, you have no idea, in the presence of syndication, what the final ad provider will be, so you have no way of requesting an exception.
> It seems like the only meaningful thing is to request *, at which point I wonder why we're making this so complicated, rather than just two options -- "I request an exception for myself" vs "I request an exception for myself and third parties on my page."
> -Ian

John M. Simpson
Consumer Advocate
Consumer Watchdog
1750 Ocean Park Blvd. ,Suite 200
Santa Monica, CA,90405
Tel: 310-392-7041
Cell: 310-292-1902
Received on Wednesday, 21 March 2012 13:41:57 UTC

This archive was generated by hypermail 2.3.1 : Friday, 3 November 2017 21:44:46 UTC