[accessibleimage] Re: Please support web access to visual verification systems

Hi Chris et al,

Reply inline...
----- Original Message ----- From: "Chris Hofstader" <chris.hofstader@xxxxxxxxxxx>
To: <accessibleimage@xxxxxxxxxxxxx>
Sent: Sunday, January 15, 2006 12:03 PM
Subject: RE: [accessibleimage] Please support web access to visual verification systems


I will happily sign this but I will also send this along to a friend of mine who is working on image analysis software > to solve this specific problem. His thesis is that the
problem is less the Turing tests themselves but, rather,
the lack of innovation or will by the screen reader
manufacturers to
even try to create an intelligent agent that can figure out > the text in these somewhat obfuscated graphics. His
work has shown a 95-97% level of accuracy with these > bitmaps and, once they have it converted to a to a plug > in for IE, they'll announce and make it available to
people who want it.

I think the problem really lies with the screen reader vendors, but it also has other aspects to it that compound the screen reader vendors position. The screen reader vendors are really now starting to fail their users. Screen scraping has reached it's limits, and to gain access to more content alternative techniques will have to be developed that superceed screen scraping. However, the screen reader vendors still continue to push screen scraping with no visible signs that they are even considering an alternative, let alone working on one. This problem is somewhat compounded by socio-cultural and political aspects. Over the past few years, at least in terms of computing accessibility, accessibility has really become synonomous with making things work with a screen reader. I get the impression that people have lost sight of what accessibility is really all about, and that is enabling people to do things, not making sure something works with a particular technology. This false association between making things accessible and making them work with a screen reader has meant that the spin doctors at the screen reader vendors can usually make the claim that it's not their fault that product x is inaccessible, it is the fault of vendor y for not making product x work with a screen reader. A cottage industry has sprung up around making things work with a screen reader, who have a vested interest in maintaining this false relationship, and legislation, in the form of Section 508, demands that some software work with a screen reader.

All this has led to the focus being switched from the screen readers failiure to innovate to the software vendors failiure to make something work with a screen reader. Therefore, if the screen reader vendors are going to be made to innovate the focus will have to be switched back to their failings. Personally, I feel that the focus should be on the screen reader vendors, as they hold the power to make almost everything accessible and in an efficient manner.

If the screen reader vendors continue to believe that it is not possible to innovate beyond screen scraping and accessing a DOM, then I'm quite happy to show them my latest creation, a system to render automatic audio descriptions for videos on the fly. It's not a system I expect the average user to use, at least not yet any way, due to the system requirements, but a lot of the techniques I used to do this can be applied to shape identification for static images, such as those found on most software user interfaces, and should be able to be run on an uniprocessor system for static images, due to lesser requirements for concurrency. I'd be quite happy to show this to the screen reader vendors providing they didn't charge the end user for using any of my work.

After fighting the document accessibility and screen reader innovation wars for six years, I left FS to work on > more future oriented projects. I truly believe that the
major screen reader companies have given up trying to > do any real innovation and are ignoring problems like
these graphics and complaining that they present
problems that cannot be solved.

As a researcher I'm a professional sceptic. I haven't yet seen an overwhelming body of proven evidence that convinces me that this sort of thing is impossible to solve. However, I've seen some evidence that has started to convince me that it is possible to solve these problems. Therefore, I think the screen reader vendors are trying to get out of doing some work, which wouldn't surprise me, it's something they're rather good at.

I think that the "impossible" excuse has been worn out > so badly that its use is nearly laughable. One screen
reader company has an overwhelming dominant
position so doesn't need to innovate to make money
and the others are too small to have the dollars or the
ability to take risks on moving the art forward.
The result is that we blinks are screwed until something changes in the approach by the established companies or > a new player comes along and, to break into the
market, must do something radically new and exciting
to distinguish themselves and offer a reason for users,
agencies, governments
etc. to switch.

With the current state of screen readers, it's not terribly hard to set a value proposition that exceeds that offered by, say JAWS, especially if you're a researcher *g*.

I think we're seeing the start of the trend to the new companies pushing the innovation bar up. Serotek with
Freedom Box System Access and Code Factory
with Mobile Speak Pocket are doing some extremely interesting things and taking a lot of risks by playing
against the establishment players. It's
yet to be seen if they can cause a tectonic shift in the
AT landscape though.

I'm just waiting until I can pick up 11800 31st Court North for a few dollars in the bargain basement sale of Freedom Scientific's assets. I truely believe, that unless the screen reader vendors start to innovate, it will not be long before a new entrant comes to market with an offering who's value proposition dwarves that of existing screen readers. When that happens, it will likely not bring good news for Dick C's bank balance.

To fit this message in with the topic of this forum, one of the things I believe it technologically possible fore screen readers to do, is to make images accessible. I've already done it, if only on the scale of a research prototype, for the W3C's Scaleable Vector Graphics format.

Will


Other related posts: