The idea is simple: use a rapid prototyping machine to build duplicate machines, driving down the cost, and making rapid prototyping widely available.
Using manufacturing systems to produce more manufacturing systems is a major goal of molecular manufacturing, the advanced form of nanotechnology. But a group at Bath University is working toward that goal today, and they are not even using nanotechnology.
RepRap stands for "Replicating Rapid-Prototyper." Rapid prototyping (RP) machines have been under development for the last decade or so. They work a bit like a 3D version of inkjet printers, squeezing out plastic in three-dimensional shapes under direct computer control. (Other RP technologies use other materials and processes.)
For the most part, RP machines have only been used to make inert objects, because they can handle only one kind of material. The advance that's put RepRap on the map is the ability to include metal--wires--in a plastic product. If wires can be built in by the computer rather than connected by hand, then more complex products become easier to build.
The RepRap plan requires that the motors, bearings, and computer chips be purchased separately, and a person would need to snap the parts together. Their design is far from finished. But there's no obvious reason why a few hundred dollars of motors, chips, and plastic goop couldn't be turned into an RP machine costing many thousands of dollars today--and the new machine, being inexpensive, could be used to build a wide range of consumer goods that today need large factories to produce them economically.
Molecular manufacturing has several important advantages over RepRap or any other large-scale manufacturing system. Scaling laws mean that small machines can be many times more powerful than large machines. And building with atomic precision lets you take advantage of several physics tricks including superlubricity [PDF paper], zero-wear sliding surfaces, and automatic maintenance of precision in manufacturing operations.
Several advantages are shared between RepRap and molecular manufacturing. The RepRap web site says it very well:
The three most important aspects of such a self-copying rapid-prototyping machine are that:
1. The number of them in existence and the wealth they produce can grow exponentially,
2. The machine becomes subject to [development] by artificial selection, and
3. The machine creates wealth with a minimal need for industrial manufacturing.
The word I replaced in their second point was "evolution." This is because molecular manufacturing has become incorrectly associated with runaway evolution and uncontrolled self-replication. Let's be clear: both RepRap and nanofactories will only build duplicates on command, and only according to the blueprints that are supplied.
Even if RepRap succeeds in making the structural and electrical components of an RP machine, it will have lower performance than a molecular manufacturing system simply because droplets of plastic are far larger and less precise than molecules. But RepRap is a good start. If nothing else, a familiar human-scale machine creating useful duplicate machines will help to mitigate the unreasoning resistance to the idea that a nanoscale machine could do the same.