Computer pioneer aids spam fight

Many web-based businesses are trying to thwart the spammers by using techniques that humans can cope with but which stump robots.

But the techniques rely on people deciphering distorted images, which can make it difficult for people with visual problems to sign up and use some websites.

To help these people researchers are now working on other systems that use sound to catch out the spammers.

Turing tested

Many companies that make a living sending unsolicited commercial e-mail, or spam, use software robots to scour the web looking for, and signing up to, free e-mail accounts.

The robots use the accounts as a launch pad for millions of spam e-mail messages and plunder directories to gather lists of people they can bombard with unwanted mail.

Now, some web-based e-mail operators, such as Yahoo and Hotmail, are using tricks that make it impossible for the robots to sign up for an account.

The technique presents someone signing up with a distorted image of a sequence of characters, usually a random mix of letters and numbers.

" Security and accessibility must co-exist, not conflict "
Julie Howell, RNIB

While humans can easily decipher the jumble, robots cannot.

These spam-stopping systems are known as "Captchas" which stands for Completely Automated Public Turing tests to tell Computers and Humans Apart.

Pioneering development on Captchas has been done by Luis von Ahn, Manuel Blum, Nick Hopper and John Langford from the School of Computer Science at Carnegie Mellon in Pittsburgh, US.

The robot-spotting system draws on the work of pioneering scientist Alan Turing.

He wrote that computers could be considered truly intelligent only when they could fool a human into thinking they were having a typed conversation with another person rather than a machine.

One of the most popular Carnegie Mellon Captchas is called Gimpy and it randomly grabs a word from a dictionary and then distorts it using different colours, by stretching letters and adding extra features such as dots and lines.

The systems are not foolproof because some researchers have written programs that can decipher the distorted images, albeit slowly.

Other robots are known to route sign-up pages that use Captchas to real people for completion.

Site unseen

Captchas are starting to cause problems for anyone who is blind or has impaired vision and cannot see the characters they are being asked to decipher.

"If this new way of presenting password information prevents visually impaired people from using a service then we have a serious problem on our hands," said Julie Howell, campaigns officer at the Royal National Institute for the Blind in the UK.

She said legislation in the Britain and US demands that companies make websites accessible to people with disabilities.

"Security and accessibility must co-exist, not conflict," she said.

She said industry should provide a means of blocking spammers that kept the bots out but let all human beings in.

Now, some scientists are working on Captchas that use sound, rather than images, to keep robots and spammers at bay.

Nancy Chan, a graduate student at the City University of Hong Kong, is working on one that overlays white noise and other distractions on top of a voice reading out random letters and numbers.