In a technological world ready primed for recognition, add computer keyboards to the list of surveillance tools that can figure out exactly who you are.
Computer models that track your typing habits can predict whether you're a man or woman with as much as 95.6 percent accuracy, according to a new Digital Investigation study.
I mean, if Facebook can recognize your face why shouldn’t your keyboard recognize your keystrokes?
It’s a technique called “keystroke dynamics”, which measures behavioral biometrics – simply meaning “something you do”.
It works like Facebook’s facial recognition feature, but instead of creating a “template” based on your tagged photos, the program tracks your keystrokes.
Each time a person types, tens of thousands of features store and share information about who that person is. It measures things you might have never even realized you were doing. Flight time, for example, tracks the time between releasing one key and pressing the next. Dwell time, meanwhile, is how long a key is pressed down. Computer programs like TypingDNA and ID Control track other factors too, like how often mistakes are made in typing, how many times certain keys (like ALT, CTRL, and SHIFT) are hit, or even how often words are deleted.
Using a program called “ISqueezeU”, researchers spent 10 months collecting data from 75 volunteers. They found the computer system accurately predicted a typist’s gender over 78 percent of the time. Time between pressing the "N" and "O" keys was found to be the most helpful clue, although the program doesn't specify which gender uses those keys faster or more often.
The technology has existed since the Second World War when it first became apparent that telegraph operators could be identified as either an ally or an enemy based on their style of tapping in Morse Code messages.
Introduced in the 1970s, the field of forensic behavioral biometrics extends beyond the keyboard and is used to identify personal traits like a person's gait, voice, signature, and even the way they use their computer mouse.
Researchers say they want to expand the small sample size to determine whether education or handedness further affects accuracy.
5
