More CAPTCHA follies
We’ve posted a good bit about CAPTCHAs, the automated techniques that try to distinguish humans from vile software bots when visiting websites or registering for accounts, such as the blurry letters you have to type into a box. It’s an arms race between CAPTCHA designers and bot designers, with visually impaired users as collateral damage.
The latest design calls for users to click near the geometric center of any image in a composite set of wall-to-wall images drawn from a database. That’s only step one; step two shows you another image, which you must identify from a list of options. They’re gonna have to get really creative to figure out a non-visual approach to this task. And hey, webmaster, is your site’s porn late medieval Italian poetry really worth all this effort?
Ars Technica: Researchers stay step ahead of bots with image-based CAPTCHA
Inflated hopes
Designed for motorcycle riders, this wearable air bag system will protect against falls and other high impact events. Many more people could continue to live independently if they felt safer moving around their homes and neighborhoods. Just be sure to de-activate it before doin’ the Bump at the Senior Surgical Sock Hop.
Camera-pedia
Just point your camera phone at any building or object (or person?) and a team of AI, GPS, and image processing software back in serverland delivers information to you: what it is, what it costs, your chances for a date…. Nokia just purchased a research company with this vision and some early results. No bar codes needed — just point and think. Security and privacy concerns aside (and we can say that because we have nothing to hide. Really.), this could provide a cognitive support system as well as an alternative input mechanism.
Hand-y passwords
Researchers in Glasgow are looking at password strategies that don’t require memorization. Options in their prototype system, Dynahand, include being able to pick out your own handwriting from a “lineup,” and using drawings rather than alphanumeric symbols. Convenient for anyone without visual disabilities who might have difficulty remembering passwords: children, seniors, people with cognitive disabilities…and most of the rest of us, for that matter.
Pop a CAPTCHA
OK, so we’re now at a point with CAPTCHA tests where either they’re too easy for spammers to crack or too hard for humans to interpret. New mainstream strategies are highly dependent on image recognition–impossible for blind people to use unless ALT attributes are implemented properly, but these attributes could also make spamming easier than ever. Meanwhile, Gareth Hayes has been developing various models of accessible CAPTCHAs with variable success–the latest requires users to hold down keys for specific lengths of time, but will that work with alternative keyboards, let alone screen readers?
Good to the bone
In this Popular Science review of seven products under development for the future office, we particularly like Image 4, a security system whose biometrics are based on facial bone structure. This is a human characteristic that is about as universal as these things get and, worst case, presumably could be adjusted after an accident or plastic surgery.
Flyometrics
San Jose Airport is implementing a new system that will let flyers pay $100 and go through a one-time background screening, then ever afterwards be able to pass standard security lines, zip through a biometric scanner, and continue on their merry way. Whatever else we might be feeling about this, we were buoyed by one fact: the system designers provided the option of using either iris or fingerprint scanning.