Burningbird

I'm in the process of rolling out some change proposals and bug reports for HTML5. I had volunteered to help with reviewing MathML during Last Call, and submitting comments for the HTML WG. Unfortunately, the process did not go smoothly.

In the meantime, this week was the W3C's TPAC meeting, where all the boys and girls from all the W3C working groups get together for a face to face. Interesting stuff happened, including the TAG (TAG is the overall W3C architecture group) recommendation that HTML WG split Microdata from HTML5. We'll see where that goes.

Twitter was very useful for those of us who were not at TAG. Those at TAG pointed out the IRC channels associated with each meeting, and where links to reports and presentations could be found. It was an example of good Twitter use.

What was not an example of good Twitter use last week were the "live" Twitter messages that came from a soldier in a hospital within Fort Hood during the recent tragic events. The inappropriate and less than helpful use of Twitter was detailed in an exceptionally good post at Techcrunch, written by Paul Carr.

In the writing, Paul makes the point that rather than help, or at least get out of the way, during a crises, we grab our cellphones and become mini-journalists—macabrely excited about being "live" at the event. We post photos of people hurt in accidents, or shot by a crazy man, regardless of who we might harm, including family members or the victims themselves. We exaggerate the event until one gunman becomes three, and an act of insanity becomes one of terrorism.

More importantly, we jam necessary cellphone lines in order to get that last tweet out, cause confusion, and aid and abet chaos.

Even outside a crises, we don't seem to know when to turn off the spigot. How many of us woke up this morning to be met with the ultimate of absurdities: hundreds of messages from folks "live tweeting" a Congressional vote. My god, it's just a bloody vote. There is nothing exciting about a vote until the vote is finished and the tabulation made.

Frankly, I would rather hear what people had for breakfast.

Anyway, more on HTML5 later, and do read Paul Carr's writing.

update Suw Charman-Anderson has a detailed rebuttal. She has some good points, especially about the Iranians feeling reassured that people were listening.

What she misses, though, is the past tense: people were listening. People listened during the Iranian election, dyed their avatars green, and filled Twitter trends with the topic. And then...it all just went away. And that's the point I think that Paul was making: social media's ability to influence events is directly proportional to the attention of the participants, and the participants are being subjected to a continuous barrage of new events, and new outrages.

The green avatars are gone. Do the Iranian people still feel assured that people are listening?

It's unfortunate that the WhatWG made a unilateral decision to go Last Call with HTML5 at WhatWG, as HTML5 is far from ready for Last Call at the W3C.

Supposedly the reason it went to Last Call at the WhatWG is no bugs in the WhatWG database. Considering that there are significant bugs and issues at the W3C, one can only assume that concern about quality is not as strong at the WhatWG as it is at the W3C.

There are also more people questioning decisions, issuing proposals for change, and submitting bugs at the W3C. If one were suspicious, one could imagine the HTML5 editor made this move to assert some form of control.

Things too good to pass up:

• The Ken Burns special, The National Parks: America's Best Idea begins on PBS tonight. The libertarian forces are already out, accusing Burns of socialist propaganda.
• Karl Martino from Paradox1x links three stories related to the healthcare debate. Included is the story of a 22 year old woman who died because she didn't seek medical care early enough—she didn't have health insurance, and was worried about the cost. There was a rumor that she died of H1N1, but later reports state that she died of pneumonia. Regardless, if she had received medical care in time, she'd probably still be alive.
• Via 3QuarksDaily, an interview with the author of "Age of Empathy: Nature's Lessons for a Kinder Society". Women are more empathetic than men, dogs yawn when their owners yawn, and we're not the selfish prigs we think we are—it's an interesting sounding book, and have placed it on my wish list.
• Another from 3QuarksDaily: Harvard is publishing Michael J. Sandel's Justice class, online. Like the National Parks, these videos will be telecast on PBS via Boston's WGBH, and are also hosted at Harvard. I watched the first class in the series, and it's excellent. I can't recommend strongly enough. (Speaking of which, if you haven't made your annual donation to PBS, now would be a good time.)
• Oh hey, the Harvard web site is hosted on Drupal.

• All that is good is not video, sometimes there's still room for static images and writing. Sheila Lennon points out that all of Life Magazine, text, ads, and photos, are now online at Google Books. I decided to check out the issue closest to my birth.

  The Life Issue I checked out featured some fun ads, including one for a portable TV that looked like a tank, and one in the back by the American Petroleum Industry about the cheap cost of gas. The issue also featured stories of the time, including the recovery of a kitten washed out by a ocean wave, and the explosion of the world's fastest seaplane.

  The main focus of the issue, though, was on a story and photos of the beautiful Gina Lollobrigida, still considered one of the sexiest women of all time. The magazine covered her extensive wardrobe, and described how she created a catalog of our her outfits, to make it simpler for her maid to fetch the correct one. She even drew pencil sketches of the outfits, annotating them with numbers so she could call and ask for the outfit by number.

  The magazine described Ms. Lollobrigida as a perfect size 12. I choked at that, because in a recent discussion about "plus-size" models, several people accused a size 12 model of being "obese". Something for James Fallows to consider in his ongoing series about obesity and America.

Cynthia Shelly released an alternative proposed HTML5 draft that addresses the table summary attribute. The responses to her draft have been less than edifying, and demonstrate rather succinctly most things wrong with the HTML WG.

If you follow along in the thread, you'll see Apple's Maciej and IBM's Sam Ruby go back and forth on protocol, a discussion Maciej ends with a suggestion to focus on salvaging a "proposal" from the work. The thing is, providing alternative text in a specification is the proposal, the only that is deemed acceptable to the HTML WG. At least, that's what we've been told in the past.

Not that Cynthia is demanding that the text be used as is. This was a suggested text, addressing how summary could be discussed in the HTML5 specification, in order to ensure proper use. The proposal also removes summary from the obsolete list. Cynthia proposed this alternative text in order to generate discussion, leading to its refinement; to encourage team effort. Simple enough to understand, but then we're subjected to the typical Ian Hickson disingenuous approach to anything he disagrees with: pretend he doesn't understand what the proposal is all about.

I couldn't find any description of what problem this proposal is trying to solve. Could you point me to the description of the issue that is being resolved here? Why is the text currently in the HTML5 spec not considered acceptable middle ground?

It is difficult to evaluate proposals without understanding what problems they are trying to solve.

Incidentally, I believe the process that we are supposed to be following these days is that when there is a problem in the spec, a bug should be filed describing the problem, so that the issue can be tracked. If you could file a bug (or point me to the relevant bug if one is already filed), that would be very helpful.

(At this point I would like to inform my readers: everyone can file a bug, you don't have to be a member of any W3C organization to do so.)

So, HTML WG team members are told by one of the HTML WG Chairs to provide alternative specification text, while the HTML5 author countermands such a recommendation, with a note that we file bugs, instead. Seriously, I keep expecting the third stooge to enter the scene, stage left.

And he does. The author of validator.nu, worker extraordinaire for Mozilla, Henri Sivonen, puts on his court jester cap to derail even the potential for worthwhile discussion:

Further quotes are from the proposed text--not from Maciej:
> Summary is one way to provide explanatory information about tables  
> that consist of more than just a grid of cells with headers in the  
> first row and headers in the first column.
>
Does this intend to say that using @summary is categorically  
unnecessary when headers appear in the first column and/or first row?  
If so, it would be good to make this clear.
> Such explanatory information should introduce the purpose of the  
> table,
>
Shouldn't the purpose be stated to all readers?
> outline its basic cell structure,
>
Shouldn't this be generated by the AT from the table model?
> The information provided by the summary is needed by users who  
> cannot see the table, but would usually be redundant for those who  
> can.
>
This sentence sticks out as non-spec-like. It doesn't state a  
requirement, so it looks odd in the middle of a paragraph that states  
requirements.
> This must be done in a way that is associated with the table via  
> markup, such that user agents and assistive technology can  
> programmatically determine the relationship.
>
This sentence could make sense in WCAG-like contexts where things are  
defined in terms of what available software happens to support. It  
doesn't make sense in a spec that defines what software must support.  
(Furthermore, "programmatically determine" is a special term from  
other specs but isn't defined as part of the special vocabulary of the  
HTML5 spec.)

The proposed text seems to imply (in the edits done on examples) that  
having the explanation in a paragraph preceding the table isn't  
sufficient without an explicit aria-describedby link (misspelled in  
the proposed text as aria-described-by). Why is that not sufficient?
> When using summary in combination with another technique, authors  
> must not use the duplicate text, but instead use summary for the  
> parts of the description that are only useful to users who cannot  
> see the table.
>
What about duplicating information that AT should be able to voice  
based on the table model?
> <table summary="The table is divided into six columns: Map number,  
> Date, Area or stream with flooding, Reported deaths, Approximate  
> costs (uninflated), and Comments. The rows are grouped by flood  
> types into six subcategories: Regional flood, Flash flood, Ice-jam  
> flood, Storm-surge flood, Dam-failure flood and Mudflow flood." 
>
In this case, the first sentence clearly duplicates information that  
are trivially programmatically determinable by the AT from the table  
model (given proper <th> markup). As for the second sentence, I think  
it would be worth investigating if the salient content of the second  
sentence is also realistically programmatically determinable from the  
table model. On the face of it, discovering the content of the second  
sentence from the table model doesn't seem like an overly hard  
software problem.

So, the text of the proposal that Cynthia provides is addressed to humans, which Henri rejects, because Cynthia's text should be addressed to machines. She discusses declarative markup, and addresses this discussion to people, in order to ensure that the summary attribute is properly used by web page authors and designers. Henri reduces the whole to algorithms, care and feeding of.

This is a perfect lead in to another discussion about HTML5 taking place elsewhere, in the W3C TAG, which has ultimate responsibility for ensuring the many specifications such as HTML5 work in a complementary manner for the web. The focus of the TAG at this time is detailing issues this group has with the current HTML5 draft, a discussion generating a typically mature level of discussion in the WhatWG IRC channel.

One such issue, as I have noted, as others have noted, is the fact that the specification is given in algorithmic terms, rather than as declarative text—based on discussions of a Document Object Model (DOM) with HTML markup given as a distant secondary item (barely covered, and leaving ripples of confusion in its wake).

The current rendering of the specification is considered more precise for the browser companies, for Mozilla, Google, Microsoft, Opera, and Apple, but the precision completely obfuscates the information needed by thousands, perhaps millions of web page authors and designers.

In the past, the main specification would be about the markup, with a secondary document describing the DOM. And oddly enough, this has worked, if we can believe the evidence of our eyes. Evidently, this wasn't to the taste of the browser companies, who believe that it is more important that their needs be met, rather than the needs of the thousands, perhaps millions of web page authors and designers.

In addition, rather than leave many decisions up to the implementors of the specification, the editor's draft seeks to detail, in minute detail, how everything is to be handled by implementors. Precise, very precise. Good luck with the 50,000 or so test cases.

So far, I have submitted three HTML5 bugs:

1. When Web Workers was removed from the spec, orphan references were left - clean up is needed
2. To remove the Microdata section, as it isn't necessary, nor widely supported
3. To allow other namespaced elements in SVG, since the use of these elements is valid within SVG

And I just submitted a bug for table summary. There will be others. Too bad I'm not one of the elite.

Ian Hickson, sole author of the HTML 5 specification, on standards in a recent #whatwg IRC discussion:

i think standards bodies are an outdated concept

Using the Firefox nightly, I was able to open an HTML5 document that contains a SVG graphic, and see it rendered, though the page was served as HTML. Wonderful! Unfortunately, running it through the HTML 5 conformance validator was less than joyful, which I'll cover in more detail later.

For now, returning to Ian's quote at the beginning of this post, I find it ironic that not only do we welcome, we demand that the browser companies be given free and unfettered reign to define the future of web standards, as they will.

Ironic, because if the web were any other industry—petroleum, pharmaceutical, airline, auto, electrical utility, and so on— allowing the companies who produce products in the industry, free and unfettered reign to define the standards for their industries, would draw howls of protests, and a demand for accountability.