It's a pretty broad term, with a lot of possibilities, but generally I mean a machine which, given a design for a machine in the form of digital data, can build a working machine of that design without further human interaction. Some imperfect approximations of the idea already exist:
- living things such as bacteria, humans, and plants: DNA is digital, true, but it's an unsolved problem to design the DNA to make a plant grow into the shape of even a chair, much less a bicycle, a lathe, or a computer.
- 3-D printers such as RepRap: they generally can only produce very limited machines without further assembly, with only very limited palettes of materials, falling far short of being able to print a whole printer without human intervention.
- ordinary compilers such as GCC: though they have already had a significant economic impact, they require a feedstock of magnetic disks or erased Flash from which to manufacture their output program, which still leaves their users vulnerable to TSMC and Western Digital. Worse, the output program depends on a CPU to run, leaving the users vulnerable to Intel and Qualcomm.
I like the idea of the lowest technology hurdle required to build one such device. Any ideas? Seems like we must be there in terms of inventability now. A good project to work on!
Your response comment saying "I have now read all of that" was posted 11 hours after I linked you to a bit over 120,000 words of notes on the topic you were asking about (according to 150 * [...document.body.innerText.matchAll(/(\d+)\s+minutes/g)].map(m => Number(m[1])).reduce((a, b) => a + b, 0) on a non-overlapping subset of the topic pages).
Some of that material is a little difficult. How did you manage to read it all so fast? I'd like to hear what you found most interesting.
It's a pretty broad term, with a lot of possibilities, but generally I mean a machine which, given a design for a machine in the form of digital data, can build a working machine of that design without further human interaction. Some imperfect approximations of the idea already exist:
- living things such as bacteria, humans, and plants: DNA is digital, true, but it's an unsolved problem to design the DNA to make a plant grow into the shape of even a chair, much less a bicycle, a lathe, or a computer.
- 3-D printers such as RepRap: they generally can only produce very limited machines without further assembly, with only very limited palettes of materials, falling far short of being able to print a whole printer without human intervention.
- ordinary compilers such as GCC: though they have already had a significant economic impact, they require a feedstock of magnetic disks or erased Flash from which to manufacture their output program, which still leaves their users vulnerable to TSMC and Western Digital. Worse, the output program depends on a CPU to run, leaving the users vulnerable to Intel and Qualcomm.