Re: ACTION-128: Draft @summary voting text in conjunction with PF

-www-archive
+public-html

Ian Hickson wrote:
> On Mon, 6 Jul 2009, Shelley Powers wrote:
>> On Mon, Jul 6, 2009 at 6:42 PM, Ian Hickson<ian@hixie.ch> wrote:
>>> On Mon, 6 Jul 2009, Shelley Powers wrote:
>>>>> http://lists.w3.org/Archives/Public/www-archive/2009Jul/0001.html
>>>> Now, if you want to create a new email with the same set of questions,
>>>> and post it in the proper group, public-html, I will attempt to answer
>>>> each and every question, as thoroughly as I can, carefully, quietly, and
>>>> civilly.
>>> Please consider the above e-mail to be that e-mail.
>> Then consider this your answer.
> 
> So you refuse to answer the questions I asked in that e-mail? Really?
> 
> Why?

Shelley explicitly asked that the questions be asked on public-html. 
Your response that she should consider your prior post on wwww-archive 
to be on public-html was clearly non-responsive.

I won't volunteer to attempt to answer each and every question, but 
here's my take on the first...

> On Mon, 29 Jun 2009, Shelley Powers wrote:
>> >
>> > What studies do _you_ plan to take to demonstrate that summary="" is 
>> > the superior solution?
>> 
>> Actually, I don't believe we need to, not unless there's a plan to do 
>> the same with every last bit of the HTML 5 spec.
> 
> Ok. So if we shouldn't base decisions on research and data, what _should_ 
> we base our decisions on?

Shelley did not say that we shouldn't base decisions on research and 
data.  Instead she noted that there are features in the HTML 5 spec that 
do not have studies that demonstrate that such features are the superior 
solution, and that there isn't a demand that such studies be provided.

To directly answer your question: I believe that the W3C HTML Working 
Group should base decisions on consensus.

In the specific case of the summary attribute, I see it highly likely 
that there will consensus on the following:

   * HTML 4 did not provide the optimal guidance for the usage of
     this attribute, and that there is widespread misuse.
   * Highlighting effective alternatives, through the spec and
     through tools such as conformance checkers is a good thing.

What I do not believe that there is consensus on is the position that 
each and every usage of the summary attribute has to be treated as 
non-conforming.  Yet this is the position of the current spec, and that 
position is taken without the benefit of studies that demonstrate that 
marking this attribute as non-conforming is the superior solution.

I believe that Rob and Maciej and others have tried to make this case. 
No, not with formal studies, but with intelligent and reasoned debate. 
Joshue and Shelley and others seem willing to participate in that debate.

Even though I am not aware of any formal studies on the matter, I happen 
to believe that there is ample evidence that people will ignore the 
advice of conformance checkers when such proposes theoretical 
superiority over demonstrably interoperable behavior.  As an example, I 
will now belatedly respond to the following:

   http://lists.w3.org/Archives/Public/public-html/2009Jun/0471.html

I previously heard another representative of your company, namely 
Marissa Mayer, give a quite different answer to a similar question when 
it was asked by the audience following a presentation that she gave. 
Her response (and I'm admittedly paraphrasing from memory so I may have 
some details wrong) was that conformance was just one factor, and that 
many other factors are considered when producing pages in general, and 
that one particular page in specific.  One such factor that Google 
places a high premium on in general, and again on that specific page in 
particular, is bandwidth.

Consider the following:

   http://html5.validator.nu/?doc=http%3A%2F%2Fwww.google.com%2F

Knowing that Google has intelligent and experienced professionals that 
produce that page, ones that are highly motivated to produce pages that 
are exceedingly interoperable, I ask the following questions:  How many 
of the over three dozen "errors" and one warning can be corrected 
without increasing bandwidth or decreasing interoperability?  How 
effective has labeling useful and often times widely interoperable 
behavior as conformance errors been in widely changing authoring practices?

My preference is that the current draft be updated to reflect that 
input.  But to date, no amount of data provided and no amount of 
reasoned debate has proven to be effective in making that outcome a reality.

If in fall of 2009 we have two draft specifications, one that is 
acknowledged as technically superior but seen as one that few will 
follow, and one that is technically inferior that describes 
interoperable behavior and is widely viewed as one that people will 
follow, my (one) vote will be that latter spec be the one that proceeds 
to Last Call.

And before you ask, yes, I do view the number of people who will 
willingly follow a given specification a valid factor to base decisions 
on.  Even for decisions that have a large technical component.

- Sam Ruby

Received on Tuesday, 7 July 2009 09:05:35 UTC