Robots are replacing humans in all sorts of jobs. But when it comes to creative fields, human beings still seem to have the upper hand. Take music, for example. It isn't just that robots don't exude the same sex appeal as rock stars—people actually prefer songs with human error. The reason for this, scientists say, is because in every man-made performance, there are tiny flaws that follow an appealing fractal pattern.

A 2011 study led by Holger Hennig, a physicist at the Max Planck Institute for Dynamics and Self-Organization in Göttingen, Germany, looked at the differences between man-made music and computer-generated tunes. First, they found that people prefer man-made music, with all its inherent imperfections, to computer-generated, technically-perfect beats. However, they also determined that when songs were digitally altered to include imperfections, listeners still preferred the organically "flawed" songs.

The so-called "humanizing" aspect available in some professional audio software applications adds random miscues in an attempt to make the sound more organic. But when humans make music, their natural divergence from perfection doesn't occur randomly, the research showed. Rather, these deviations create a fractal—a never-ending, self-similar pattern seen throughout nature, such as in the spirals of a seashell or the veins of a leaf.

More recently, a team of scientists, including Hennig, put their theory to the test by examining the drum beat of a popular song. (Granted, this means the study is incredibly limited in its scope.) The team chose pro drummer Jeff Porcaro for its focus—as a session musician, he was responsible for keeping time for the likes of Pink Floyd, Steely Dan, Michael Jackson, and Madonna. From there, they honed in on one particular song—the 1982 hit "I Keep Forgettin'" by Michael McDonald.

The study measured the deviations in the four-per-beat tink-tink-tink-tink on the hi-hat throughout the song, both in timing and in volume. The imperfect intervals between the sixteenth notes, as well as their volume, varied throughout. And when the researchers considered just these deviations, either in pacing or volume, they noticed the pattern was the same, whether they were examining just a few seconds or the song's full 3:40. As Hennig described it to Science Magazine, "It seems that the timekeeper in the brain not only produces fractal timing, but likely also fractal intensity or, in this case, loudness." Although both the intervals and the volume exhibited fractal variations, the patterns didn't correspond—meaning Porcaro wasn't just playing louder when he was also playing faster. 

This tendency towards fractals seems to be what gives music its artful, human quality. Although now that the curtain has been pulled back on the magic happening in a talented musician's brain, Hennig says this should allow artificial humanizing software to become even more, well, human-like. Better watch out, rock stars.