The idea of a self-replicating machines has long been a dream — and nightmare — of artificial-life science. It’s a dream for NASA dudes, for example, insofar as self-replicating machines would be useful for colonizing other planets: You just send a couple of ‘em to Mars in a flashlight-sized rocket, give ‘em 10 years, and they’ll eventually build a suburb. The nightmare scenario, on the other hand, is the old “grey goo” worry, wherein teensy nanomachines run amok and eat up the entire planet while trying to build infinitely versions of themselves, turning Earth into a slurry of unformed entropic crap, kind of like L.A.

Good times, either way you slice it! So I was intrigued to read the latest report in Nature about some Cornell scientists who’ve created genuinely self-replicating machines. They’re little cubes that can rotate internally, in a vaguely Rubik-like way; they can also attach to one another — and detach — via magnetism. The result is a set of borg-like cells that can form new copies of individual robotic “organisms.” The BBC reported on it, writing …

Their long-term plan is to design robots made from hundreds or thousands of identical basic modules.

These could repair themselves if parts fail, reconfigure themselves to better perform the task they have been set, or even to make extra helpers.

Do not fail to check out the super-odd video of the robots in action — duplicating themselves.

(Thanks to Erik and Bill for this one!)