I agree with your sentiment and only wish that I believed it could work that way. At risk of being flamed for this view, and with a bunch of caveats below, I think we need to come up with something that allows transforming proxies to transform sites that know nothing about them, like the many sites that haven’t been touched in years.

Caveats: I’m not sure quite what the benefit of bringing stale sites to an adoring mobile audience is, not am I sure that a transformed experience of them is worth having either. However, we need to accept that different players in the mobile value chain have different perspectives and someone needs to blink in order to make progress.

Since in general (but see http://site.mobi), as we just mentioned, you have to have a degree of tech savvy, regrettably, to build a mobile site and since these are far fewer right now, practicality says to me that the change has to be that way round. Add to that the fact that some kinds of broken markup and content actually crash the mobile devices (though that is getting better) you are really not doing the end user much of a favour by saying that everthing should be sent through untransformed, unless the site says otherwise.

To build the mobile ecosystem we need to encourage users. We can do that by on the one hand saving them from the worst effects of broken web sites, and on the other hand by building great experiences for them. If we build great experiences for them we need a way specifically to tell third parties to leave those experiences alone.

Jo

I just don’t see how that order of operations works out:

“If, in response to a request, they get a Vary: User-Agent or * they MUST re-request the URI with the User-Agent they received and remember that setting for that URI (using a mechanism similar to authentication realms in which user agents are expected to remember the URI path and re-present authentication credentials without prompting the user for URIs in that path).”

If the transcoder has requested the page without the original UA then the server will most likely return a desktop formatted page (if we’re talking about current behavior of most sites that have a desktop version and a mobile version).

Why aren’t we making existing desktop only sites opt into trascoding instead of asking the people who have mobile versions to opt out? It’s much easier in that direction, there’s only one format and if it comes back with the “Yea-Sure-Screw-With-This” header then you can do whatever you want. Instead of multiple requests back and forth to the server and modifying Vary headers and sending cache control and requesting that the original user agent be returned with the next request.

It doesn’t magically enable all the web for mobile usage, but spend some time browsing around through a transcoder and I’m pretty sure you’ll think there’s nothing magical about it. It’s better than nothing, I like the whole trascoding idea. But I think it’s overstepping bounds, especially when an entire carrier starts sending all their traffic through a transcoding system (I consider Google in a separate class here, they don’t automatically get a swipe at whatever any user puts in the location box of their mobile browser) to hide the original user agent in the outbound request.

Late to the party, I just thought I would point out that the selective quote from my post to W3C Content Transformation List doesn’t quite convey the full sense of what I was saying. So for completeness:

I realise that what I am proposing puts the onus on servers that have more than one representation to do something that they don’t today, and I realize also that I am suggesting that transcoding proxies can modify the user agent string. In light of recent intemperate discussion, on this and other lists, let me repeat that what I mean is that they can do this only in order to avoid a blocking response. If, in response to a request, they get a Vary: User-Agent or * they MUST re-request the URI with the User-Agent they received and remember that setting for that URI (using a mechanism similar to authentication realms in which user agents are expected to remember the URI path and re-present authentication credentials without prompting the user for URIs in that path).

I agree with your subsequent post and the obstacles to creating content and how difficult this is even for quite technically literate people. But that is a separate subject, in my view:

What do we want? Decent usable tools! When do we want them? Now!

But first, let’s stop content being mucked up whether it was created with decent tools or not.

Jo

And did I not just see that Novarra’s proxy will stop transcoding sites with “mobile” or “wap” and so on in the domain? I think that’s great insofar as it’s addressing the underlying issue: not transcoding sites that don’t want or need it. Again, the issue is not that people should actually demand a transcoder say it’s a phone, since it isn’t, and would lead to other bad behavior.

Hmm… who started everyone on this chant of “User-Agent! I want my User-Agent!” anyway? glad to see the right things are happening anyway.

This isn’t somehow changing the rules; proxies and these headers have been around for a while. Yeah I agree the mobile technology ecosystem is complex stuff and you can’t just throw up a page and get it right. That’s why I’m most happy to see timely, constructive discussions coming out of the W3C at the moment that offer the ‘right’ solution for transcoders and developers, a way forward. I know one transcoder that’s going to improve its behavior in this regard as a result.

Once we have agreed that:
a) there’s a number of sites providing cool mobile content that should not be touched
b) there are many sites that do not provide a decent mobile experience

How do we make the two sides happy?

With all due respect…

As mentioned here:

How Vodafone and Novarra Killed Mobile Commerce
http://www.seoprinciple.com/how-vodafone-and-novarra-killed-mobile-commerce/20/

and here:

Vodafone UK, Doing A Lovely Job Of Supporting The Mobile Web By Breaking It
http://mobhappy.com/blog1/2007/06/06/vodafone-uk-doing-a-lovely-job-of-supporting-the-mobile-web-by-breaking-it/

This issue has been affecting mobile companies since at least June. While this might not seem like a burden to people at large companies I can assure you the impact on small ones is significant.

Sometimes a bit of well deserved “Name & Shame” works wonders when big co ignores the little people. Re: the comment you have been sprinkling about “The bitch-fest I see all over the internet about VF + Novarra isn’t doing much though…”

…apparently has had EXACTLY the affect that was intended.

Vodafone paid attention and the right people are talking together and sorting it out.

Note: I really like Sean and respect the smart, good work he has contributed towards mobileOK Best Practices.

Re: “be glad to see stuff like “Cache-Control: no-transform” highlighted as a way to get a transcoder to redirect the browser directly to the target site.” and “don’t you want ways to say “get out of the way, don’t transcode”? ” and “Whether you think the Novarra proxy should be there at all is a different story — it should absolutely respect a sites wishes regarding transformation”

Yes. …and in this case or others that label a site “No Transform” the User-Agent String needs to be passed intact (which I understand is the case) so the correct content, version, templates, experience etc. can be delivered.

Cheers,
David Harper
Founder, Winksite