As we all know, the internet is a far from equitable place as far as accessibility is concerned. Despite general legislation enshrined in the Equality Act and the Disability Equality Duty, specific recommendations in the form of the Web Content Accessibility Guidelines (WCAG) 2.0 and an enforcing body – the Equality and Human Rights Commission – cyberspace is still far from being a level playing field for those of us ‘exceptional’ enough to be using adaptive or assistive technology. Ironically, one of the biggest bugbears for accessibility campaigners is a scheme conceived to protect users and websites from the kind of security breaches and identity abuse for which the internet is becoming increasingly infamous.
The CAPTCHA (or Completely Automated Public Turing Test to Tell Computers and Humans Apart) is a code which humans can crack but which, currently, computers cannot. The distorted image of a character sequence is the most commonly encountered form and usually appears as part of the registration process to enter or use a website or on-line service. If you can’t accurately interpret the code and correctly type it into a text box, you are unable to register - and the vast majority of sites provide no alternative. Unless you can craletters saying CAPTCHAck the code you are effectively locked out.
Humans vs robots
The critical problem with CAPTCHA is that many humans cannot decipher the code either, and nor can the technologies they commonly use (screen reading software for example). If you have a vision impairment, dyslexia or learning disability you may be unable to crack the code. The accessibility arm-lock that CAPTCHA represents is an ever-present problem for many millions of users world-wide.
Common CAPTCHA alternatives
The most common alternative is an audio version of the graphical image, providing a second chance at deciphering the code using an audio version that speaks out the characters or sometimes a series of numbers or words instead. However, the audio distortion needed to obscure the code from malicious programs, can make it almost impossible to hear – even for people with no hearing impairment. And even if when you manage to hear the content, the short-term memory recall demanded by the exercise may defeat the user anyway.
Other alternatives involve the selection of descriptions of images from drop-down lists and asking the user to decide if a randomly rotated image is upright; but these still require the user to have some useful vision and the ability to deduce the subject of the images.
One route of course is to ‘get round’ the problem altogether with solutions that demand either the cooperation of the website provider or the ability of the end user to enhance the functionality of his/her browser. One approach adopted by Google is to have a hidden link visible only to screen reader users (who are most adversely affected) that leads to a form where they can request to be registered for an account by someone in customer support, but this is slow and resource-intensive.
Empowering ‘community driven’ innovations such as ‘Webvisum’ fulfil a vital need to enhance accessibility, whilst arguably easing pressure on site owners to raise their game. A browser-based ‘add-on’, available only for Firefox, Webvisum offers dozens of features which make life easier for blind and vision impaired surfers such as high contrast page viewing, link and focus highlighting, and, perhaps most importantly, it provides automated and instant CAPTCHA image solving.
Ask different types of questions
So far, so good, but even WebVisum cannot handle all CAPTCHA images. The introduction of a series of random logic-based questions (such as the service provided by http://textcaptcha.com) is a far more accessible alternative to the graphical option. For example:
Which word starts with "k" from the list: demanded, knead, triplet?
The number of body parts in the list: dress, house, elbow and shirt is?
Although logic-based CAPTCHA offers more independence to those with sight problems, it has been subjected to criticism for demanding greater cognitive ability than the image-based variety. Whilst not disputing this fact, it is also true to say that users with learning difficulties struggling with these kinds of questions, would be unlikely to be surfing the web without assistance anyway.
Intelligent bots are on their way!
The truth of the matter is that however hard we try to provide inclusive CAPTCHA, the ‘bots’ are always one small step behind and improving all the time – necessitating an arms race that leads to ever more distorted codes to crack. Even those who generate these more accessible options, warn that it is only a matter of time before their security is breached by the evolution of bot intelligence such as natural language interpretation technology.
CAPTCHAs are here to stay – at least for the foreseeable future. The only failsafe solution if you require assistance, is to have an obliging human on standby to wrestle with the CAPTCHA test on your behalf.
Robin Christopherson, October 2012
This article first appeared in Robin's blog for PublicTechnology.net