3 ms·

How much longer till a machine can produce an identical, working and complete copy using only raw materials and a blueprint?

marcosdumay 13y ago
After we discover a way to make microchips in factories that cost $1k instead of $1G, and deal with the revolution in economics and computing that follows it... Then we can think about machines that reproduce themselves.
- mikeash 13y ago
  As chip technology advances, does any of it make it substantially easier and cheaper to make old, slow chips? In other words, while massive factories are clearly the way to go for current technology and they let you pump out cheap chips in massive quantities for ridiculously low per-unit prices, what if you wanted to, say, fab an 8086 today? If you jump a few decades back, can you make those chips individually today through some reasonable process, or is the whole business just inherently large-scale even after all this time?
  - sbierwagen 13y ago
    An automated chip factory intended for use in a self-replicating machine is only useful if it can produce chips from unprocessed ore. You can make your job a lot easier if you design it to just run on silicon boules-- but those are almost as expensive per kilogram as finished packaged chips... So you need a furnace for refining sand, a chemical factory to produce the dopants, lithography equipment, clean rooms, ultraprecise steppers... Really, the problem here is the hidden complexity. It's not like you can look an obsolete chip foundry, and say, "Okay, so this facility is worth $100 million, and is about an acre of equipment, how do we miniaturize it?" That single foundry is the tip of a $100 billion pyramid. You can't just replicate the foundry, you have to replicate all the factories that built the stuff that went into the foundry.
    - rmason 13y ago
      How this usually works is someone comes up with an entirely new way of accomplishing the task. Someone young or from outside the industry who doesn't know what they're trying to accomplish is impossible. When I look at how fast the 3D printer industry is going you never know. If you can currently print a house using concrete and in the near future print out a kidney who is to say you won't be able to make an 8086? http://www.wimp.com/printerhouse/ http://www.wimp.com/printerhouse/ http://www.livescience.com/41480-3d-printed-kidneys-take-small-steps.html http://www.livescience.com/41480-3d-printed-kidneys-take-sma...
      - sbierwagen 13y ago
        Programmers typically don't have very good intuitions regarding the pace of change in other industries. Quoting myself (http://c1qfxugcgy0.tumblr.com/post/31187427192/the-enduring-tragedy-of-the-robotics-industry-is http://c1qfxugcgy0.tumblr.com/post/31187427192/the-enduring-...): Not every technology follows the price curve of computers— computing is the bizarre outlier. If you had predicted in 1970 that computers would be tens of millions of times cheaper per FLOP in the future, you would be a visionary, but if you predicted that jet airliners would cost $10, fit in your pocket, and fly at Mach 100, then you would be an idiot. Some things are just hard. Not everything gets 10X better per year for decades and decades. Most industries hit hard physical limitations long ago, and are improving either very slowly, or not at all. 3D printing is a particularly bad example of revolutionary change: it was invented by industry insiders, for decades was performed by exorbitantly expensive machines, and is mostly hyped based on the term "3D printing." Actual FDM machines are much less exciting, and aren't terribly useful, commercially.
sbierwagen 13y ago
Depends on what you'd call a "machine", and how "raw" your raw materials are. Bob Freitas wrote an open-sourced book on this a decade ago: http://www.molecularassembler.com/KSRM.htm http://www.molecularassembler.com/KSRM.htm For simple tinkertoy-ish machines, floating in a sea of parts, and sitting on a shaker table, (to provide power) it's already been done. For anything with a microchip in it, full closure would require machine-phase nanotechnology. Partial replication should be possible with conventional technology, (a "clanking replicator" http://en.wikipedia.org/wiki/Clanking_replicator http://en.wikipedia.org/wiki/Clanking_replicator ) if you supply the replicators with "vitamin" parts they can't produce themselves. (Chips, stepper motors, anything made of rare minerals) A lot of work was put into this for the Advanced Automation for Space Missions study in 1980, a proposal for a solar panel factory on the Moon that would build more solar panel factories: http://www.islandone.org/MMSG/aasm/ http://www.islandone.org/MMSG/aasm/

marcosdumay 13y ago

After we discover a way to make microchips in factories that cost $1k instead of $1G, and deal with the revolution in economics and computing that follows it... Then we can think about machines that reproduce themselves.

mikeash 13y ago
As chip technology advances, does any of it make it substantially easier and cheaper to make old, slow chips? In other words, while massive factories are clearly the way to go for current technology and they let you pump out cheap chips in massive quantities for ridiculously low per-unit prices, what if you wanted to, say, fab an 8086 today? If you jump a few decades back, can you make those chips individually today through some reasonable process, or is the whole business just inherently large-scale even after all this time?
- sbierwagen 13y ago
  An automated chip factory intended for use in a self-replicating machine is only useful if it can produce chips from unprocessed ore. You can make your job a lot easier if you design it to just run on silicon boules-- but those are almost as expensive per kilogram as finished packaged chips... So you need a furnace for refining sand, a chemical factory to produce the dopants, lithography equipment, clean rooms, ultraprecise steppers... Really, the problem here is the hidden complexity. It's not like you can look an obsolete chip foundry, and say, "Okay, so this facility is worth $100 million, and is about an acre of equipment, how do we miniaturize it?" That single foundry is the tip of a $100 billion pyramid. You can't just replicate the foundry, you have to replicate all the factories that built the stuff that went into the foundry.
  - rmason 13y ago
    How this usually works is someone comes up with an entirely new way of accomplishing the task. Someone young or from outside the industry who doesn't know what they're trying to accomplish is impossible. When I look at how fast the 3D printer industry is going you never know. If you can currently print a house using concrete and in the near future print out a kidney who is to say you won't be able to make an 8086? http://www.wimp.com/printerhouse/ http://www.wimp.com/printerhouse/ http://www.livescience.com/41480-3d-printed-kidneys-take-small-steps.html http://www.livescience.com/41480-3d-printed-kidneys-take-sma...
    - sbierwagen 13y ago
      Programmers typically don't have very good intuitions regarding the pace of change in other industries. Quoting myself (http://c1qfxugcgy0.tumblr.com/post/31187427192/the-enduring-tragedy-of-the-robotics-industry-is http://c1qfxugcgy0.tumblr.com/post/31187427192/the-enduring-...): Not every technology follows the price curve of computers— computing is the bizarre outlier. If you had predicted in 1970 that computers would be tens of millions of times cheaper per FLOP in the future, you would be a visionary, but if you predicted that jet airliners would cost $10, fit in your pocket, and fly at Mach 100, then you would be an idiot. Some things are just hard. Not everything gets 10X better per year for decades and decades. Most industries hit hard physical limitations long ago, and are improving either very slowly, or not at all. 3D printing is a particularly bad example of revolutionary change: it was invented by industry insiders, for decades was performed by exorbitantly expensive machines, and is mostly hyped based on the term "3D printing." Actual FDM machines are much less exciting, and aren't terribly useful, commercially.

sbierwagen 13y ago

Depends on what you'd call a "machine", and how "raw" your raw materials are. Bob Freitas wrote an open-sourced book on this a decade ago: http://www.molecularassembler.com/KSRM.htm http://www.molecularassembler.com/KSRM.htm For simple tinkertoy-ish machines, floating in a sea of parts, and sitting on a shaker table, (to provide power) it's already been done. For anything with a microchip in it, full closure would require machine-phase nanotechnology. Partial replication should be possible with conventional technology, (a "clanking replicator" http://en.wikipedia.org/wiki/Clanking_replicator http://en.wikipedia.org/wiki/Clanking_replicator ) if you supply the replicators with "vitamin" parts they can't produce themselves. (Chips, stepper motors, anything made of rare minerals) A lot of work was put into this for the Advanced Automation for Space Missions study in 1980, a proposal for a solar panel factory on the Moon that would build more solar panel factories: http://www.islandone.org/MMSG/aasm/ http://www.islandone.org/MMSG/aasm/