Powered by Blogger

Friday, September 14, 2007

User Error or Screen Reader Deficiency?

In the item I wrote yesterday about the interfaces to a pair of google
online applications (gmail and blogger), I stated that JAWS required
one to switch to the basic html view on gmail, which is definitely the
case. I also stated that JAWS works deplorably on the blogger pages,
to which, Darrell Shandro, my friend and a guy I trust a lot posted a
comment saying that blogger, in fact, works quite well with both JAWS
and System Access. As Darrell has used the blogger interface for
quite some time and I have rarely used it, preferring instead to post
my Blind Confidential items via the blogger for Word button bar with
the old blogger interface and with its email feature in the new
interface, the problems I experienced with it are as likely to be user
error or a lack of familiarity on my part as deficiencies in JAWS 8.
I will defer to Darrell's assessment and, therefore, make the
assumption that the blogger interface can be used with JAWS in a
reasonably accessible manner and will go back to blogger to
investigate further.

One problem I have experienced with the google and, for that matter,
various Microsoft online features have nothing to do with screen
reader access but, rather, the distortion they add to their audio
alternatives to visual verification. I recently had my hearing
checked and the results said my hearing is above average. Thus, I
don't think the problems I experience result from my audio input
wetware.

So, I believe that in their attempts to provide an accessible
alternative to the visual CAPCHA while maintaining a high degree of
security, google and Microsoft have inadvertently created a system
that continues to stand in the way of my getting past a Turing test.
Yesterday, I had a Skype chat with Darrell about an unrelated issue
but he mentioned that he could successfully use the google audio
alternative but not the one that Microsoft offers, so your results may
vary from my own. In the google sounds, I can pick out some of the
characters it plays but I have yet to actually get all of them and, in
a system that requires one enter the information exactly as presented,
I find myself unable to do so. On the Microsoft site, I don't think I
heard a single thing that it required me to type in. Either way, I'm
SOL with either site.

For blogger, I will install Word 2007 and check out its blog
interface. As all three of the Windows screen readers I use with any
frequency perform reasonably well with Word, I'll predict that this
feature will work pretty well with all three. Of course, my
predictions tend to be wrong so, for all I really know, the actual
results may suck out loud.

I haven't spent more than an insignificant amount of time with any
Office 2007 applications other than Word and Outlook so I can't say
anything about the entire suite. When one gets the knack of using the
Word ribbon interface, I think they will generally like it primarily
do to its context sensitivity. In previous versions of Word (I've
been using it since version 3.0 for DOS back in 1986), I have found
myself searching around its interface to get to features that I do not
use often. The new Word interface, because of its contextual nature,
makes finding features related to what one is doing much simpler than
in the past.

I find the Office 2003 keystroke compatibility feature of Word 2007
quite confusing because it doesn't contain all of the keystrokes I've
grown accustom to using over the years. In general, when a program
claims keystroke compatibility with an earlier version of itself or
with a competitor's product, I find that such causes confusion if some
of the keystrokes with which one has familiarity do not work. I
believe that one will lose confidence in a compatibility layer when
they need to remember which keystrokes work and which do not.

Well, this post has wandered pretty aimlessly so I'll just end it here
so I can get on with my real work.

-- End

3 Comments:

Anonymous Will Pearson said...

There's no such thing as user error! User errors have been reclassified as design flaws in a product; flaws that make it difficult or impossible for a particular user to use that product. Claiming things are "user error" makes you sound like Eric, although he calls them "training issues", and we both know it was Eric's denial of the existance of design flaws in the JAWS UI that started the bad blood between Freedom Scientific and myself. User interfaces should be designed to fit the user, the context in which the user will use the product, and the task they will use the product to accomplish. Designing products around users helps to avoid so called user error; so, when users do produce errors it is often indicative that a product hasn't been designed properly.

User errors by screen reader users can indicate design flaws in the screen reader they are using. The current set of screen readers should be called semantic strippers. All of the modern screen readers strip out some of the semantic content that is contained in a user interface. Most of the semantics that are lost are communicated through the position of something, either absolute or relative, and the set of attributes that are based on the position of elements within a set, such as size, grouping, and style. The semantics conveyed by these attributes can sometimes help a user to understand the purpose of a user interface and the components that comprise it. So, by leaving these attributes out of their interpretation of the user interface screen readers actually make it harder for users to use certain pieces of software or web pages. These are flaws in the design of modern screen readers that seem to have crept in to screen readers because the people building those products don't understand the needs that their products fill.

Current implementations of the audio capcha concept are another instance of poor design. One design flaw that is often found in audio capcha implementations is that they force people to listen to the whole character string and they don't allow people to listen to just a single character. This differs from the visual capcha that does allow people to focus their attention on just a single character. The ability for people to scope their attention to just a single character can be quite important if people are having difficulty recognising a particular character or characters in the string. Repeatedly listening to the entire character string just to recognise a single character likely raises a person's cognitive workload and may cause problems with memory storage and retrieval.

One factor running through current approaches to accessibility is that they seem to only provide functional equality. Functional equality is great to some extent as it provides blinks with the ability to do things but it isn't real equality. Even though things may be accessible it is often more difficult and more time consuming for a blink to do something than it is for someone else. Real equality would be a situation where these differences didn't exist. It's quite likely that if someone produces something that is accessible they are still breaking disability discrimination laws due to a lack of real equality.

The sad thing is that in the world of software it's really only a few companies that are responsible for this lack of real equality and they are the screen reader vendors. A lot of the inequality stems from the differences in how information is presented to a user and the low level capabilities that a form of presentation affords a user. In the current approach to accessibility, the one where things are made to work with a screen reader, it is the screen reader vendors who have responsibility for presenting the information to a user. Therefore, the screen reader vendors are the real source of this lack of real equality. If people want to minimise the chance of breaking disability discrimination laws then maybe they should build their own accessibility solutions instead of making things work with screen readers.

5:02 PM  
Blogger Philip. said...

On a completely separate note, I always wondered how screen readers managed with smilies - is there a standardisation, or if I were to post colon, dash, bracket does it just say colon, dash, bracket?

3:31 AM  
Anonymous Jake said...

I agree with Will Pearson when he says that there's no such thing as "user error." I have tried Google's audio CAPTCHA before, but I could never get it right. What are the characters supposed to be separated by, i.e., spaces or some sort of punctuation marks? BTW, I have created a journal over on Tabulas. I had originally thought about using LiveJournal, but again I couldn't correctly do their audio CAPTCHA. So then I registered on Diaryland, which at the time at least, had none of that CAPTCHA stuff but the comments feature didn't work properly. Someone alerted me to this when she tried posting a comment, so I tried numerous times to contact the people in charge at diaryland.com but nobody would respond. I saw Tabulas mentioned in an email I received some time ago, so I decided to register and see if it worked. It does, and the Comment's feature even works properly. There is no CAPTCHA of any sort, registration is free, and their website is quite accessible. Just click on my name and you'll be taken directly to my journal.

10:51 AM  

Post a Comment

<< Home