Re: CT problem statement

Jo,

I certainly do not wish to trivialize your work and the work of the Group. It is because I respect W3C and this group, that I am surprised the work here was cited by Vodafone to divert attention from their User-Agent header. It was a misunderstanding on my part. I want to stress that I do realize the importance of the work that you are doing, and I thank you for doing so. I sincerely apologize for anything I said that seemed offensive or seemed to trivialize your work. I do not mean to be offensive. I speak for myself and do not represent anyone.

All I wanted to address is the controversy that is going on with Vodafone's inteference of the HTTP User-Agent header. It is pretty clear where the consensus so far is leaning. Jo, even you yourself questioned the practice in the Vodafone Forum. The Problem Statement even describes this issue. I want to know if the organization has the ability to influence Vodafone to change their practice, or, perhaps more realistically, how much clout does each of the individuals here have in getting Vodafone to change their practice?

While trying to get Vodafone to change their User-Agent via other means, I do agree that standards work such as this here is important. If this is what is needed to get Vodafone to change their practice, and to prevent other operators doing similar things, perhaps I should bring the fight here. I am new to the process of W3C and do not know the roles of a Task Force. Pardon my ignorance and I appreciate pointers to how the process works. As someone working in this space, I do follow the work in Mobile Web Initiative and in dot.mobi . It's the first time that I engage in such activity. Pardon me for storming in and acting outside of established procedures and processes. In light of the current controversy, I do have some comments about the current drafts:

On the draft Guideline itself: perhaps a Guidance for Network Operators is needed in Section 2? Or perhaps a completely new section about the operations of such Server is needed. Network Operators operates the Content Transformation Server, who does not necessarily develope the technology, but does have the ability to configure and control its use. I think the points should be about operations, not what the technology can or should do. For example, perhaps it should be mentioned that Content Transformation Server Operators should remain as transparent as possible, if it is inserted without the knowledge or consent of the User and/or the Content Author. My view is that the Operator should

1) make the HTTP request as transparent as possible, leaving intact the User-Agent and other headers that identify the Browser, and modifying headers such as the "Via" header and "X-Forwarded-For" headers if they conform to current standards or prevailing practices. It may add headers to aid Origin Servers in identifying such Transformation Server exists.
2) in HTTP response,  it should transcode the content only
  i) if the user allows it and/or
  ii) if it is certain that the returned content is aimed for desktop use, through the detection of the Content-Type header, the DTD/DOCTYPE, additional headers such as Cache-Control: no-transform or heuristics that looks at the content length, the content itself etc.
3) In general, not tempering with existing technologies that Servers use to identify the Browser and/or device. It should never transcode by default. It should transcode if the user allows it, and/or if it is certain the content returned is for desktop use.

Which leads to another question: It seems to me the Guideline written so far is more about the undelying technology and mechanism, which is great in unifying the various techniques that exists in practice today. However, does it also make recommendations that addresses the non-technical issues as stated in the Problem Statement, such as Legal, Moral, Commercial and Security issues? Or is the Guideline not the right document for that?

How about addressing specific use cases? Is that too specific for a standards document? A user downloading Opera Mini is expecting it to transcode/transform the content somewhat, so there is an implicit user agreement. Ditto for users of web transcoders like the Google Web Transcoder. What the Opera proxy can and should do is probably different from a Content Transformation Server inserted by an Operator into their network.

Also, perhaps it should be recommended that the transcoded content pass
the mobileOK Test, or conform to the Mobile Web Best Practices.



Regards,
Nigel.