For those who don’t remember (and my last two dozen or so posts have made it easy to forget) I am involved in the web development industry, and have been for more than ten years. I don’t often write about it here, first because it became hip to blog about web standards for a while, and then because I had nothing to say about it.
A couple of years ago I had a major wake-up call involving the way web sites are built. I was given the opportunity to watch several blind people navigate a web site I had built. I left feeling awful. I felt like I had been torturing these people. It took them hours to fill in a single form to request information. They all admitted to me that they would probably just call a sighted friend and have them enter the information instead of grinding through a form that was worse than dental surgery to them.
It opened my eyes, and frankly, it turned me into what my colleagues call an “accessibility nazi.” I began researching, and eventually implementing accessible web sites. I’ve learned a lot between then and now, but like web standards, accessibility became cool to write about, and I rarely follow the crowd.
That’s changing for today. I came up with an idea yesterday that hopefully solves one of the major accessibility nightmares on the Internet. You’ve all seen them. An image block containing a string of warped, obfuscated letters and numerals, and a box for you to type what you see. This is called a Captcha. It apparently stands for something (Completely Automated Public Turing test to tell Computers and Humans Apart). If that wasn’t made up after the fact, I don’t know what is.
Captchas seem to be a great solution for keeping spam-bots out of sites and letting people in. Except that people accessing the web with a screen reader, say blind people, cannot use them. The wonderful completely automated test cannot, in fact, tell computers and humans apart if the humans can’t see. Besides, clever spam-bot creators have brute-forced captchas and can get past them anyway.
None of this, despite legislation, depsite lawsuits, has stopped their proliferation on the ‘net. They’re everywhere. Gmail uses them, yahoo uses them, every forum in the universe uses them. And all these systems exclude blind people. Oh sure, some provide a phone number to call to circumvent the captcha (most usually having a sighted person on the other end of the line fill in the form for them) but that doesn’t make it right.
So I have been mulling this over for a long time, and I’ve come up with an idea. A solution, maybe. Unfortunately, it is only a partial solution, as it causes difficulty with users that might have cognitive disabilities, but they’re also having trouble with standard captchas, so maybe we’re one step closer to a perfect solution. Maybe my idea will inspire someone to come up with something better.
It is with this hope that I publish the idea, get it out into the world where it can be refined and debated, instead of keeping it close to me in an ego-driven effort to toot my own horn. If someone makes a go of it, they can name it after me or something. They can call it a RAIN test (Random question to defeat Artificial Intelligence agents and keep them out of the Network) or something. Stranger things have happened.
It goes like this. Instead of your standard, run of the mill captcha, a simple question is asked. Several answers are provided, but only one makes logical sense. These are not multiple choice. You still have to type these into a box, just like with a captcha. The big difference is that a screen reader can read all the answers to you. With care and thought given to the question and answers, it should be be an easy matter to lock out the spam-bots without locking out the blind. Add several dozen questions to a site (large volume sites would have to update them regularly), and have them randomly chosen at the time of page load, and we should accomplish what we set out to do without punishing those using screen readers simply because they can’t see the letters and numerals in a picture.
The idea is rough, most definitely, and I need time to build a test implementation, but it HAS to be better than the garbage we’re shoveling at screen readers right now.
And it doesn’t require a sighted friend.