Patrick Traynor, assistant professor in the School of Computer Science, and colleagues have programmed smartphones to use their accelerometers to detect and decipher strokes on nearby keyboards with up to 80 percent accuracy.