Picture Your Password

Text-based passwords have become infamous for their ability to be cracked. But what if we used graphical representations that were easier for users to remember and tougher for bad guys to guess?

Researchers in Ottawa recently conducted a field study of graphical-based passwords as a possible replacement for traditional passwords, and they found that while graphical passwords are more easily recalled, many users don't like them as much.

"To make [text] passwords memorable makes them insecure," says Sonia Chiasson, a PhD student in computer science at Carleton University in Ottawa, Ontario, and one of the authors of the study on "click-based" graphical passwords. "[Users] have to use insecure passwords to cope with the memory load. In those cases, passwords become insecure."

Chiasson and a team of Carleton University researchers last week presented their findings at a usability and security conference hosted by Carnegie Mellon University. They found that click-based graphical passwords -- where users pre-select and then click a set of five points in a graphical element such as a photo of a parking-lot full of cars -- can work in the "real" world outside the lab.

But the study's results also suggest that if users have multiple graphical passwords, they end up struggling to recall them -- much like they do with multiple text-based passwords, notes Lorrie Cranor, director of Carnegie Mellon University's Symposium On Usable Privacy and Security (SOUPS) and associate research professor for Computer Science and Engineering & Public Policy.

"More research is needed," Cranor said in an email response. "At this point, it is not clear to me that graphical passwords really offer much in the way of usability or security benefits over text passwords. In general, I'm not that optimistic that graphical passwords are going to be workable as a widespread replacement for text passwords, as there is little evidence that they offer much -- if any -- memorability improvement for people who have dozens of passwords, which is most of us."

The study did show promise in how users could easily learn and deploy the technique. But there were some security red flags with the graphical passwords: The so-called "hot spots" (click spots) in the image were sometimes easily guessed, and users in the study say they often used a common pattern for their hot spot areas.

"Some parts of the image were more likely to be clicked on," Chiasson says, which left some of these passwords open to attackers.

Chiasson says the next phase of the study will attempt to avoid letting users pick their click-points in any predictable pattern, as they did in the recent study. "We're investigating different ways in helping people create better passwords."

So why did the users in the study still lean toward those easily-guessed, easily-lost, and easily–hacked text passwords? Chiasson says she thinks the users in the study favored text-based passwords mostly because that's what they're used to, and because the graphical approach is less familiar to them.

At least one vendor, Passfaces, already sells graphical password software. Unlike the click-based technology used in the Carleton study, Passfaces specifically uses facial recognition and cognition only. (See Passfaces Expands Reach.)

Paul Barrett, chairman and CEO of Passfaces, says his company's graphical technology is a natural evolution from the text-based password. Still, he says, most Passface customers still use it as a second layer of authentication, along with text-based passwords.

— Kelly Jackson Higgins, Senior Editor, Dark Reading