 I made sure to look up the pronunciation of von Neumann before I started recording this. After all, there's no reason to make myself look like a fool. At my new job, I spent a lot of time working with 3D printers to prototype mechanisms and parts quickly and easily. 3D printing technology has existed for decades, but in the past five years or so it's become readily accessible by hobbyists and makers of all stripes, and has exploded in popularity. For a few hundred dollars, you can probably afford a machine that will spit out little plastic models of just about anything, maybe one of the thousands of models available for free online. The process is pretty straightforward. The printer melts raw plastic and extrudes it out of a heated nozzle, very much the way a hot glue gun operates. It uses motors to drive the nozzle in two axes, up and down and left and right, and draws shapes on a surface with the plastic. By dividing three-dimensional objects into very thin layers, the machine can build three-dimensional parts by drawing them slice by slice up from the bottom, until, when everything's said and done, you've got a little triceratops skull. Pretty frickin' cool. The process is called additive manufacturing because you're slowly adding material to build up to the final product, as opposed to subtractive manufacturing, where you take a raw hunk of something, like marble, and remove the bits you don't want until you have David. Many modern 3D printers are built with components that are themselves 3D printed, a recursive loop of manufacturing that has all sorts of benefits. If some part of your printer gets damaged, print a new one. If someone figures out a better design for some component, you can print one out and replace it without issue. Unfortunately, many of the components of the printer aren't made of plastic and can't be made this way. Belts, motors, wires, circuit boards, there's no real way to make these sorts of objects using the layers of plastic process. But there's nothing in-principle preventing them from being fabricated using an additive process. Any material with a melting point, copper, steel, rubber, aluminum, could theoretically be heated to that melting point and built up by a print head using the same method. Some underpricing engineers and inventors have demonstrated that it's possible to print all sorts of different things. Circuit boards, houses made of concrete, an incredible variety of useful stuff. Nobody's made a printer that can replicate itself entirely out of raw materials yet, but it's definitely within the realm of possibility. And, as it turns out, people have been thinking about exactly that for decades. In 1966, mathematician and all-around smart dude John von Neumann published this paper based on one of his lectures, Theory of Self-Replicating Automata, which is one of the most awesome and nerdiest lectures ever. He investigates all sorts of cool stuff from the energy required to run a human brain to the average number of errors our neurons must make in the course of a lifetime, just a phenomenal number of interesting observations and calculations. But the climax of the lecture involves the idea of a universal Turing machine, that is, a computer, which allows an automaton, that is, a robot, to make copies of itself. Immediately this brings all sorts of interesting questions to mind. When does a machine cross over to being considered alive? Runaway robots, paperclip optimizers, it's fertile ground for speculation. But one of the coolest ideas that came from his thoughts on self-replicating automatons bears his name explicitly, the von Neumann probe. See, getting humans into space is actually a huge pain in the neck, because space is decidedly inhospitable to life as we know it. Radiation, extreme temperatures, micrometeorites, not to mention the astounding lack of oxygen or seamless delivery options. We've been to the moon just a few hundred thousand miles away, and even that was one of the most astounding and audacious acts of human ingenuity and effort that we've ever achieved as a species. As anyone who's played Kerbal Space Program can tell you, space is hard enough when you don't have to worry about annoying things like life support or extreme G-forces. It's much, much easier to use robots for space exploration, which is why we do exactly that. The Voyager probe, the Curiosity rover, we've littered practically every nearby astrological object with scores of machines that send us details about our immediate neighborhood. Of course, as the dinosaurs might tell us if they were still around, there's still plenty of good reasons to get humans onto other planets, but to get to that point it would still probably be much easier to send robots ahead of them to do some of the heavy lifting. Although it's much easier than using people, getting robots out there is still pretty freaking hard. Some of the most advanced scientific and technological organizations on earth still hold their breath every time they send something into space, and at ten thousand dollars a pound to launch something into orbit it's easy to see why. One loose connector, one snapped screw, one misplaced decimal point, one inconvenient rocky outcropping, and you've spent a huge amount of money to deliver a useless hunk of metal somewhere UPS and FedEx won't go. Even if everything works once it gets there, the speed of light makes controlling remote robots extraordinarily difficult. Imagine trying to play a driving game online with a ping of about three seconds. But what if you could deliver more than one robot at a time? I mean, for each extra robot you launch you're going to double the cost, but what if you could launch one self-replicating robot, something that could build a copy of itself out of raw materials when it got to its destination? This is the idea behind von Neumann probes. If you can get a single machine out of your planet's gravity well that's capable of autonomously gathering resources and duplicating itself, suddenly you have two machines at your destination, one of which you got more or less for free. Not only do you get all the benefits of self-made parts, like with 3D printers, but you also have exponential growth working for you now. Every time your robots copy themselves you get to double the number of workers available for harvesting resources, manufacturing, repairing, and also whatever you originally wanted your robots to do. The automated nature of the probe means that it doesn't require input from humans back home, and doesn't suffer from the radio lag problems of current probes. We could simply check in from time to time to see how far the process had gotten, how many copies our robots had been able to make. At some point after so many thousand robots had self-assembled out of asteroids or regolith or whatever, we could signal or reprogram them to switch over and start making whatever we cared about, space stations, mining equipment, maybe a Dyson Sphere? In 1980, NASA started seriously investigating this approach to design future missions. Their publication, Advanced Automation for Space Missions, available to the public, detailed several strategies for using von Neumann probes for a number of useful tasks, like creating a sensor network around Earth or exploring Titan under the moons of Saturn. It's a fairly simple idea, at least theoretically, and it has the potential to be extraordinarily powerful, which is why it worries some people. Consider this. If we're not the only intelligent life in the universe, we might well expect alien civilizations to build these sorts of self-replicating probes for their own purposes. We can imagine a few such goals, exploration of nearby planets, collection of resources for construction projects back home, maybe building habitats or even terraforming entire worlds to be suitable for colonization. Sounds pretty cool, unless the civilization in question doesn't really care about preserving alien life, that is to say, us. They wouldn't need to be particularly cruel or sadistic to exterminate entire ecosystems with self-assembling robots. They could simply not bother to differentiate between carbon and carbon. Of course, we haven't seen anything of the sort flitting through the universe yet, which might also be cause for some concern. We've talked before about the Fermi Paradox, the idea that, by the numbers, we might expect to find dozens of intelligent alien species on the trillions of nearby planets in our galactic neighborhood, but the universe remains eerily silent. A number of explanations have been advanced for the peculiar absence of messages from alien civilizations, but the simplicity of the von Neumann idea leaves us with yet another problem to explain. No radio signals and no robots. In any case, aliens are no. The potential of self-replicating automata is truly impressive, and unlike many other science-fiction-ish ideas, we might be on the cusp of being able to send a single 3D printer to the asteroid belt and waiting for it to make some friends. Let's just make sure that it knows not to use Earth for friend-making, yeah? Do you think that we're anywhere close to being able to manufacture our first von Neumann probes? Please leave a comment below and let me know what you think. Thank you very much for watching. Don't forget to bubble, subscribe, blah, share, and don't stop thunking.