What broke down in the cases we've examined here wasn't just HTML code. What broke down was a process, and that's one of our key points here. Even with the best intentions in the world, text-only variants are almost inevitably second tier; they're poor cousins to the (media-) rich folks at the fancy-dress ball. They receive far less attention from designers, developers, and managers than the media-rich primary pages do, and it shows. Separate is not equal.
The Amazon Access site raises similar issues. There's no evil there, either. Just a misunderstanding, perhaps even an inevitable one. Those of us in the accessibility community have been "selling" accessibility, in part, by making the point that many of the techniques that support enhanced accessibility for people with disabilities especially people who are blind or visually impaired are also essential to support wireless access. Perhaps what we see on Amazon Access, then, is evidence of what happens when these claims are interpreted too literally, when they're understood to mean that designing for wireless access is the same thing as designing for accessibility to people with disabilities. In Section II of this book, we'll show that what's really necessary is to turn the telescope around. Maximum accessibility means starting with the needs of users who have disabilities and building from there.
So let's try thinking about this problem from a different standpoint. It's important to approach ALT text and the notion of "equivalent alternatives" as much more than something we need to provide to help out those who are unfortunately unable to see the images or hear the sounds on our Web sites. Instead, we can recognize that equivalent alternatives like ALT text, for example like other well-written and meaningful text, like well-crafted images are important resources for excellent design, elements that we can take advantage of as we work to create the best possible experience for everyone who visits our sites. That's maximum accessibility!