The three most imposing technologies today, according to many who should know, are genetic engineering, nanotechnology, and robotics. What is most startling about these technologies is that they are all on (or over) the brink of being capable of self-replication. Witness the recent Cornell University announcement of a self-replicating robot. The robot consists of a vertical stack of four four-inch cubes that are held together magnetically. Each cube has identical machinery and a computer program that allows for replication. The magnets within each cube allow it to attach and detach from others, and each cube can "bend" along its diagonal, allowing the stack to reach down to the table and pick up another module. A robot in this case is made up of four cubes linked together; one robot can assemble an identical robot when fed the ingredient cubes by researchers. Its two most limiting factors are that it secures its electric power from the table upon which it replicates and that it must be fed already-assembled modules by the experimenters. The Cornell replicating robot is of a low order, but clearly we are moving in the direction of higher forms of replication, and this is a big step. The groundwork of replicating automata traces in large part to the seminal paper by John von Neumann, "The Theory of Self-Reproducing Automata," which describes his 1950 Universal Constructor. Almost all replication studies and experiments build on von Neumann's shoulders -- as do present-day computers. His Universal Constructor work envisions a machine capable of building any other machine as long as it is able to access its description. The machine theoretically consists of a memory tape that holds the description, a constructor that can read the tape and interpret it, and an arm that is under the direction of the constructor and builds the replicated machine. It is a clear extension of the Turing Machine, which Alan Turing conceived in the late 1930s. Decades later, the Cornell scientists and others have taken up von Neumann's trail of crumbs leading to artificial replication. A group referring to itself as BioWall has confronted the huge problem resident in von Neumann's cellular automaton -- a prodigious memory requirement -- and settled for a reduced set of elements in building a hardware model of one "organ" of von Neumann's machine. This is no small feat. Another legatee of this kind of vision is Chris Langton, who investigates artificial life. His unusual work and perspective are reviewed rather extensively in Kevin Kelly's book Out of Control. This book is a good source for those who wish to get started in this direction. Is it possible for us to conceive of a time, not so distant, when products are replicated rather than manufactured? That is to say, products planned and designed to reproduce themselves? The thought is both numbing and to a degree shocking. It wrenches our traditional thinking to allow such a notion to have a chance. What do you mean "fly"? Robots on Mars; are you kidding? Thousands of songs in a tiny little box; you must be mad! The next time we are presented with such supposed improbability, we might want to step back and take another look. Could an iPod make itself? Can we make the leap into the unknown and accept the idea that this is in our future?