W3C home > Mailing lists > Public > public-html@w3.org > June 2009

What makes a failure? (@summary, et al)

From: Charles McCathieNevile <chaals@opera.com>
Date: Mon, 08 Jun 2009 11:48:47 +0200
To: "public-html@w3.org" <public-html@w3.org>
Message-ID: <op.uu69jlnqwxe0ny@widsith.local>
There has been a lot of discussion about how various accessibility  
features have, for various reasons, failed on teh web.

The most recent such argument I read was this morning, and stating what I  
understood it runs:

- the summary attribute (in this case) more often than not contains stuff  
that isn't a valid and useful value.
- this means that in most cases of its use, accessibility is not served.
- this means that accessibility is harmed by the summary attribute.
- this means it is a failure and we should use something else.

It is specifically the third step (which is often the most implicit part  
of the reasoning) that I think needs to be examined more carefully.

 From my understanding and experience, accessibility of the web overall is  
not very good. There are lots and lots of things that cause problems.  
However, in various specific cases it has improved.

For example, when working on WCAG 1 a decade ago people said it is simply  
outrageous to demand alt attributes in order to ensure accessibility -  
despite the fact that without them, it is completely obvious to everyone I  
have talked to that there will be major accessibility issues. Alt  
attributes are still used badly or not at all in many cases (and despite  
our best efforts, even here at Opera we have a few such cases pop up from  
time to time), but I believe there are far more good uses of alt text than  
there were 10 years ago.

Further, I believe that this represents an improvement in accessibility.  
This is despite the common use of alt="" or some other default meaningless  
text where it is inappropriate, which actually reduces the accessibility  
of the particular page even in comparison to simply leaving off alt, by  
actively misleading the user. (Leaving off alt is still going to break  
accessibility in such cases, it is just a slightly lesser among available  
evils, in a case where it is possible to do good).

In other words, my experience in accessibility leads me to believe that in  
general, the presence of lots of broken and harmful content is not  
necessarily a failure of a feature. Otherwise I would conclude that the  
effort to achieve accessibility has failed - and while I believe that  
unfortunately accessibility is still woefully inadequate (and have some  
blame to share around...) I think that there has actually been real  
improvement, measured in terms of people's increased ability to do things.

A clear assumption here is that failure is a relative, rather than  
absolute term. And this invites the further reflection that using  
something which is less good may be blocking us from implementing  
something that would be used better - i.e. would lead to a higher  
incidence of people being able to do things, as a measure of accessibility.

Indeed, finding a better solution is why we don't simply make something  
and then stop. Promoting a better solution over a worse one is valuable.  
But measuring better is based on guesswork. I have argued (for many years,  
in particular with my friend John Foliot) that the accesskey attribute in  
HTML has suffered from terrible advice leading to unworkable browser  
implementations (a generally recognised statement), but that this is  
actually very easy to fix, only requiring a few browsers to change their  
implementation in relatively simple ways. And I have argued that this  
would be better than implementing the new access element proposed by the  

I cannot prove that argument, since it is based on predicting the future.  
I can point to the large amount of data that will not need to be changed,  
although in practice that argument will lose its value over time as less  
of that data remains important (or even available) and it becomes an  
increasingly small fraction of the web. I can explain that as well as the  
data, the tools and knowledge that lead to its production are going to  
hang around - and will be used to generate new content the same as old  
content for some time (looking at the state of HTML on the web, one would  
suspect that it will be a long time before poor old tools are replaced),  
which delays the previous replacement effect. Some people upgrade faster  
than others - corporate environments, university laboratories, private  
users with specific requirements, poor schools and rich start-ups are all  
different environments. And so on...

It has been proposed to remove a number of accessibility features of HTML  
on the basis that they don't work well - and on the basis that this is  
therefore harmful, so just removing them is better than doing nothing. I  
argue that this is not true.

There are other factors. Accessibility attributes can be used for Search  
Engine spamming, and this is obviously counter to the specific interests  
of some WG members (and not a great thing for the web as a whole). The  
failure of accessibility frustrates users, and leads to criticism of the  
technology. The fact that upgrading systems and knowledge of those systems  
is extra difficult for many people who require accessibility features,  
leading to people not adopting things that are "better" when they are  
available, can complicate the picture by misleading casual observers as to  
the relative value of various features (and by making it unclear even to  
experts). And so on...

It seems to me that we should be far less hasty to remove things from HTML  
which were designed to support accessibility - and far more hasty to work  
on adding an alternative which is better, in order to actually field test  
the two side by side for long enough to understand which of our  
assumptions were brilliant insights and which were foolishness... because  
I am certain that we will all find things we backed on both sides of that  



Charles McCathieNevile  Opera Software, Standards Group
     je parle français -- hablo español -- jeg lærer norsk
http://my.opera.com/chaals       Try Opera: http://www.opera.com
Received on Monday, 8 June 2009 09:49:32 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 10:15:46 UTC