W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > October to December 2005

Re: Validity

From: Matt May <mcmay@bestkungfu.com>
Date: Fri, 4 Nov 2005 15:18:17 -0800
Message-Id: <916A31E0-E639-42D6-81C5-647FE1EF3CF2@bestkungfu.com>
Cc: w3c-wai-gl@w3.org
To: Jens Meiert <jens.meiert@erde3.com>

On Nov 4, 2005, at 1:27 AM, Jens Meiert wrote:
> No offense meant, but this issue is ridiculous, isn't it? Not  
> including
> validity undermines the WCAG (how is something supposed to be  
> automatically
> testable without required validity) as well as the W3C (by  
> disrespecting
> other key specifications), and it generally sets the wrong signal.

There is nothing ridiculous about pragmatism. I would love to hear  
someone say the same of extremism.

> Is there really /any/ advantage of not including it? Stating that  
> there are
> user-agents where you can actually improve (!?) accessibility by  
> invalid
> markup is like stating that you can cross the city faster by  
> ignoring all
> red traffic lights.

Ooh. I love traffic analogies. Here's one:

Cars pollute the atmosphere. So a government decides that in 2006,  
all cars on the road must produce a maximum of x parts per million in  
CO2, where x is a technically achievable number.

Now let's say that x is so low that no cars made before 2001 could  
achieve it, and only 5% of new cars meet it. Furthermore, there  
aren't enough of those cars to go around, and they start at 60% of  
the country's average salary; kits to fix existing cars are  
expensive, take mechanics lots of time to install, and are not  
available in adequate numbers to meet demand. Most cars cannot meet  
that standard, ever, and that means millions, including the most  
disadvantaged groups, risk legal trouble if they continue to drive  
them. The result would be economic disaster (and in the US, anyway,  
perhaps revolution), because people will have been forced to spend  
money to meet the letter of the law, without ever sharing the  
underlying goal.

This is an elaborate analogy, I know, but each part of it matches up  
with the Web today. We have ten years of legacy sites. Standards- 
based design didn't appear as a fully-formed methodology until 1998,  
didn't catch on until well after 2000, and is still not nearly an  
industry standard. Many college-level HTML classes still teach tag  
soup. And I can tell you who in the world I trust to produce valid  
content at all times because _I am on a first-name basis with each of  
them_. That is, maybe a hundred out of _millions_ of content producers.

Worse is the state of authoring tools. Now, remember when I say this  
that the last 3 years of my career have been spent working on this  
area. And of all of the WYSIWYG tools, content management systems,  
export engines, and so on, I've seen nearly all of them produce  
invalid content at some point. Those are the good ones: recently- 
released tools, many of which try to maintain validity. The old stuff  
is impossible, or nearly so, to make valid, and that's provided you  
have the skills in your organization to get it done, and lots and  
lots of time to do it. Again, this is work that is not primarily  
about accessibility: it is merely passing a validity test, and  
remediation would probably be done by people who won't apply the  
semantics or metadata needed to meet the rest of WCAG. They'll just  
push the button and hope to get a treat. These are the cargo-cult  
programmers I was talking about at the f2f, and they won't make  
things any better.

Then there are the third-party additions: many sites import chunks of  
code from here and there, and don't have any control over whether or  
not that fragment is valid. I'm in this situation right now: a client  
of mine uses a third-party control that contains bogus attributes. I  
could threaten my client or this other vendor until I'm blue in the  
face, but as it is, it won't break anyone's experience to keep those  
attributes there, and I can't change it short of re-architecting  
another company's system from scratch for free. So what value should  
I put in a document that tells me I'm inaccessible when I know for a  
fact that's not the case?

To sum up, the problem set: too few products that do what is required  
out of the box. Too much expense to bring the horribly broken stuff  
out there, which is to say, 95-99% of the web, up to conformance. No  
margin for error. And a world of developers who are under-equipped  
both in their own skill level and the universe of tools available to  
them today to reliably make it happen. This is the Web in November  
2005 that I'm talking about. Seriously, look around and try to tell  
me otherwise. I'm an idealist, like all of you. I do standards-based  
design as the minimum starting point for every project I work on. I  
evangelize it to every audience I speak to. I say that producing  
valid, semantic code at all times is the only sustainable way to  
produce accessible content day in and day out. I wish all of you  
would understand that I object to this not because it's more  
convenient for me, or because I stand to gain in some way from it,  
but simply because it is bad policy to advance to the world at large.

If a government did what I outlined for automobiles, they would be  
seen as out of touch with reality, and actively causing damage to the  
cause they were planning to help. This is the same situation:  
validity at level 1 turns this document from a specification  
outlining problems and minimum measurable solutions into a vehicle  
for advancing a religious war.

The next argument put forth, if I've been paying any attention to  
these dustups, is: well, then, the governments should be the ones to  
relax the requirements. But how many of us really want to see every  
government who evaluates WCAG 2 open it up and pick which things they  
want to do? We will not have advanced accessibility worldwide if we  
can't produce something that everyone can agree is the absolute  
minimum that is necessary and sufficient to raise access standards  
for users with disabilities, for all content, in all localities. To  
require validity without airtight evidence that it is necessary,  
sufficient, and the least restrictive means of achieving that goal,  
we're not shooting ourselves in the foot. We're shooting ourselves in  
the head.

If you want to create an environment where it is reasonable to expect  
valid code at all times, then you should be calling authoring tool  
vendors every day demanding that they produce valid code at all  
times. Better yet, get them to commit to conforming to ATAG. That's  
what it's there for. But until the full cycle of content development  
can be made valid in more than a small utopia of individual  
instances, it must not be a minimum requirement.

Received on Friday, 4 November 2005 23:18:24 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 16 January 2018 15:33:57 UTC