6 ms·

Mechanical computing systems using only links and rotary joints

100 points by ColinWright 5y ago

zhamisen 5y ago
Nice plenary talk from Ralph Merkle about the topic: https://youtu.be/yVX9Ob4SjGA https://youtu.be/yVX9Ob4SjGA
- f00zz 5y ago
  This is really cool, would love to see this sugar cube running Doom. But unless I'm missing something he doesn't say anything about how to go about building this thing?
euroderf 5y ago
Isn't this basically what's required on Venus ?
yrgulation 5y ago
Somehow i think this field should be researched more thoroughly. Imagine a plan B for complete societal collapse, or war like situations where digital systems are not available, yet some basic agricultural computations or means of communication can be achieved purely by mechanical computing systems. For instance typing on a typewriter (or predefined message buttons) while spinning a basic motor could create radio waves; similarly something could receive such messages and decode them. Perhaps even for space com such a device could be used as backup. If anything it might be interesting to have knowledge about.
- akeck 5y ago
  Go back to artillery, etc. mechanical targeting in WWII, just before electric computers came onto the scene. Feynman talks about those mechanical computers in Surely You're Joking Mr. Feynman. Apparently, you can do mechanical integrals and derivatives with specially-shaped non-round gears.
  - neltnerb 5y ago
    Not sure about derivatives, but an integrator is just a rotating disk with a ball on it, both spherical. The ball rolls over the disk from friction, and the spinning is the magnitude of the output integral. The input function is the position of the ball relative to the center of the disk. If the ball is at the center of the disk it doesn't move, if it moves positive the ball spins up, if it moves negative the ball spins down. The overall rotation of the ball is essentially the integral. You can also do almost this to make multipliers by varying the speed of the disk. These mechanical computers were used until the mid 1900s.
- WJW 5y ago
  You might be interested in the analog computers that were used to compute tide tables back in the day. It's quite fascinating how they even managed to do basic Fourier transforms without electricity: https://www.youtube.com/watch?v=IgF3OX8nT0w https://www.youtube.com/watch?v=IgF3OX8nT0w
- Koshkin 5y ago
  > complete societal collapse, or war like situations where digital systems are not available If something like that happens all you will want is clean water, some basic food, and a weapon to defend yourself, your family, and small, closely-knit community you will not have a chance of survival without. And yes, you will quickly forget about "space com."
  - waste_monk 5y ago
    Putting aside communications with other survivors / between settlements, there's a very good reason to want to receive satellite communications: weather satellites. You probably won't have a weather radar at your post-apocalyptic settlement, but if you can receive NOAA imagery (or whatever suits e.g. Himawari-8 if you're in the APAC region like me) you can at least see if there's any nasty storms coming up and hopefully do some basic agricultural forecasting. I do agree that it's something you should worry about once your basic survival, shelter, and community needs are secure. However given you would need a reasonable technology base to manufacture mechanical computers, I think the "dependency management" would naturally sort itself out. This also assumes that the satellites haven't been shot down during whatever disaster caused the complete failure of electrical computing, and that you can actually build a mechanical computer that can receive the broadcast and usefully render the images.
    - Someone 5y ago
      Do these satellites automatically correct for orbital decay (https://en.wikipedia.org/wiki/Orbital_decay https://en.wikipedia.org/wiki/Orbital_decay) and collision avoidance or does that require commands from the ground? Reading https://www.swpc.noaa.gov/impacts/satellite-drag https://www.swpc.noaa.gov/impacts/satellite-drag, I get the impression that involves ground control (because of those complicated models that get updated all the time) So, how long will these satellites stay up there without ground control?
      - waste_monk 5y ago
        My understanding is that (at least for Himawari-8) it is geostationary at a high enough altitude (apogee and perigee) that it should be good for at least a few decades - apparently you actively have to boost geostationary satellites out to a retirement orbit so they don't crowd up the prime real estate. That said, the imaging sensors have a limited lifetime (the satellite platform is designed to be good for ~15 years but the sensor package limits the mission lifetime to 8 years).
- PeterisP 5y ago
  Digital processors are not needed for "basic agricultural computations" nor for communication. A ham radio can work and transmit/receive information without any typewriter-like automation (both microphones and manual Morse code don't require computation) and you can do all the calculations needed for ordinary agriculture on an abacus if needed. In any case, the extremely precise machining in large quantities required for any reasonable mechanical computer means that it's not a low-tech or high-reliability replacement for electronic computers, it's a very high-tech future solution to the limitations of current electronic devices. We can't make such devices now, we'll hopefully be barely able to make them in the future, but in the case of infrastructure collapse we'll lose the capability to make these advanced mechanical devices before losing digital devices. And nowadays computing is so ubiquitous, that every horrible scenario would still leave us with huge amounts of computing power lying around. Soldering together devices out of scrap electronic components is a much more reasonable way of doing post-apocalyptic computation than any of this mechanical computing tech; If there was a real need, nowadays you could extract a dozen usable processors from a crashed car or modern kitchen - even excluding computers and smartphones, so many modern devices have random processor(s) that are more powerful than what people used as their primary computers some decades ago. Things like wireless headphones and washing machines have chips that could be sufficient for so many computing needs. Apple's Lightning to HDMI cable has a powerful ARM processor inside. etc, etc.
nynx 5y ago
It’s a shame Merkle and Freitas never made their nanotech dream happen. There are some hints that a Canadian company called CBN Nanotechnologies is working with them (they bought all of Merkle and Freitas’ patents), but who knows.
Animats 5y ago
This reminds me of Drexler's early nanotechnology work. They need to fabricate something, anything, to validate that they can get those devices to work at that scale. Things are not Newtonian down there.
- R0b0t1 5y ago
  It reminds me of Neil Stephenson's "rod logic" (which is not really expanded upon) in the Diamond Age.
  - kragen 5y ago
    That's Drexler's. He proposed it in Nanosystems and at NanoCon at about the same time; see http://www.nanoindustries.com/nanojbl/NanoConProc/nanocon2.html http://www.nanoindustries.com/nanojbl/NanoConProc/nanocon2.h... and https://hackaday.com/2015/10/19/rod-logic-and-graphene-elusive-molecule-scale-computers/ https://hackaday.com/2015/10/19/rod-logic-and-graphene-elusi....
- jcranmer 5y ago
  They do cite Drexler's early nanotechnology work... as more or less the newest citation for "hey, something like this can be built." It's really disappointing that rather key aspects of this are more or less ignored. There's not much of an analysis of things like what force propagation looks like, gate delay issues, mechanical energy losses due to heat, etc. I'm surprised there isn't anything like an analysis of macro-scale linkages to try to estimate what these things looked like if they were shrunk down to scale (even though, as you say, things don't work at that scale like they do at the macroscale).
  - kurthr 5y ago
    I agree, and I know (and like) Eric and Ralph... but this is philosopher scientists and computer scientists looking at mechanical engineering (and material science) problems. Similarly, there was a lot of work at MIT on doing micro scale motors and turbines, which just work very different due to Reynolds # scaling, heat dissipation, and other issues. Realistically, you need a demand side on this sort of work for functional and realistic solutions to take hold (iterate and improve). Right now Moore's Law has driven semiconductor logic to the point where very little else can economically compete. Thus there's little investment of resources... or even accurate determination of markets. At the same time many people view nano (and micro) machining through Moore's lens, when in reality the low dimensionality, isolation, and fault tolerance of electrical/digital solutions allows for much more design independence/flexibility and complex system development. Imagine the vibratory harmonics of a mechanical computing system. There would be literally millions or billions of coupled oscillators. What about stiction/frictional losses due to surface chemistry/tension? What about limiting thermal noise effects? It's complicated and there's no money in it :^)
Koshkin 5y ago
Imagine the noise!
thedrbrian 5y ago
Sounds a bit like analogue computing OmegaTau has a podcast on them [ as well as plenty of other things ] https://omegataupodcast.net/159-analog-computers/ https://omegataupodcast.net/159-analog-computers/
klysm 5y ago
I want to be able to use this as a complication target.
mLuby 5y ago
A comment on the article's video had two interesting points: > There is linear stress problems that can occur, a single "activation" line (clock signal through a train of locks) can eventually reach a point that the mechanism will be unable to move at all. You can see an example of that long single activation line in this gif http://www.merkle.com/images/4_phase_shift_register_v4.gif http://www.merkle.com/images/4_phase_shift_register_v4.gif and described in more detail in this part of the video: https://youtu.be/yVX9Ob4SjGA?t=769 https://youtu.be/yVX9Ob4SjGA?t=769. I was surprised this wasn't explicitly addressed in the talk. Transistors aren't just powerful because they can be off or on (switch) but also because they can amplify signals (a lever). Edit: The paper does address this in section 4.5, saying that "The method for dealing with scaling is to isolate individual cells in a shift register (or more generally, individual clocked logic elements, such as Fredkin gates) by using the four-phase clock. That is, forces along signal lines cannot move beyond two cells before encountering a locked lock, preventing any accumulation of forces over long distances. At any point in time, a cell is either blocked from transmitting forces to adjacent cells, or can transmit forces only to one other cell (either the predecessor or the successor)" and "Force accumulation in link logic systems can be mitigated because each lock acts as a mechanical amplifier, with a small force controlling a much larger one." > I'm really loving the idea of mechanical logic gates and computers... it's far easier to produce in 3D print, which will be vital in space exploration and colonization. Question is, what additional primitives are needed to construct all the primitives (locks and balances) from a bin of links and rotors? Maybe a "lifter" that raises a hinge pin out of 2D plane so it can be placed in a rotor.
ffhhj 5y ago
This is fascinating. I always dreamed of building a circuit of light instead of electricity, controlled by micro-mechanical elements. An idea I got from reading an old book about building analog computers from wires, bolts, magnets, wood, etc. (forgot the name of the book, any ideas?)
chalcolithic 5y ago
I wonder why (it seems) no one has designed computer that would run on steam. Wouldn't it be easier to manufacture because of lower precision and wear resistance requirements?
- nokita 5y ago
  Babbage's analytical engine comes to mind.
  - chalcolithic 5y ago
    My bad, I wasn't clear enough. I didn't mean mechanical computer powered by steam engine. I meant a computer that used steam to propagate signals, akin to hydraulic computers

100 points by ColinWright 5y ago

zhamisen 5y ago

Nice plenary talk from Ralph Merkle about the topic: https://youtu.be/yVX9Ob4SjGA https://youtu.be/yVX9Ob4SjGA

f00zz 5y ago
This is really cool, would love to see this sugar cube running Doom. But unless I'm missing something he doesn't say anything about how to go about building this thing?

euroderf 5y ago

Isn't this basically what's required on Venus ?

yrgulation 5y ago

Somehow i think this field should be researched more thoroughly. Imagine a plan B for complete societal collapse, or war like situations where digital systems are not available, yet some basic agricultural computations or means of communication can be achieved purely by mechanical computing systems. For instance typing on a typewriter (or predefined message buttons) while spinning a basic motor could create radio waves; similarly something could receive such messages and decode them. Perhaps even for space com such a device could be used as backup. If anything it might be interesting to have knowledge about.

akeck 5y ago
Go back to artillery, etc. mechanical targeting in WWII, just before electric computers came onto the scene. Feynman talks about those mechanical computers in Surely You're Joking Mr. Feynman. Apparently, you can do mechanical integrals and derivatives with specially-shaped non-round gears.
- neltnerb 5y ago
  Not sure about derivatives, but an integrator is just a rotating disk with a ball on it, both spherical. The ball rolls over the disk from friction, and the spinning is the magnitude of the output integral. The input function is the position of the ball relative to the center of the disk. If the ball is at the center of the disk it doesn't move, if it moves positive the ball spins up, if it moves negative the ball spins down. The overall rotation of the ball is essentially the integral. You can also do almost this to make multipliers by varying the speed of the disk. These mechanical computers were used until the mid 1900s.
WJW 5y ago
You might be interested in the analog computers that were used to compute tide tables back in the day. It's quite fascinating how they even managed to do basic Fourier transforms without electricity: https://www.youtube.com/watch?v=IgF3OX8nT0w https://www.youtube.com/watch?v=IgF3OX8nT0w
Koshkin 5y ago
> complete societal collapse, or war like situations where digital systems are not available If something like that happens all you will want is clean water, some basic food, and a weapon to defend yourself, your family, and small, closely-knit community you will not have a chance of survival without. And yes, you will quickly forget about "space com."
- waste_monk 5y ago
  Putting aside communications with other survivors / between settlements, there's a very good reason to want to receive satellite communications: weather satellites. You probably won't have a weather radar at your post-apocalyptic settlement, but if you can receive NOAA imagery (or whatever suits e.g. Himawari-8 if you're in the APAC region like me) you can at least see if there's any nasty storms coming up and hopefully do some basic agricultural forecasting. I do agree that it's something you should worry about once your basic survival, shelter, and community needs are secure. However given you would need a reasonable technology base to manufacture mechanical computers, I think the "dependency management" would naturally sort itself out. This also assumes that the satellites haven't been shot down during whatever disaster caused the complete failure of electrical computing, and that you can actually build a mechanical computer that can receive the broadcast and usefully render the images.
  - Someone 5y ago
    Do these satellites automatically correct for orbital decay (https://en.wikipedia.org/wiki/Orbital_decay https://en.wikipedia.org/wiki/Orbital_decay) and collision avoidance or does that require commands from the ground? Reading https://www.swpc.noaa.gov/impacts/satellite-drag https://www.swpc.noaa.gov/impacts/satellite-drag, I get the impression that involves ground control (because of those complicated models that get updated all the time) So, how long will these satellites stay up there without ground control?
    - waste_monk 5y ago
      My understanding is that (at least for Himawari-8) it is geostationary at a high enough altitude (apogee and perigee) that it should be good for at least a few decades - apparently you actively have to boost geostationary satellites out to a retirement orbit so they don't crowd up the prime real estate. That said, the imaging sensors have a limited lifetime (the satellite platform is designed to be good for ~15 years but the sensor package limits the mission lifetime to 8 years).
PeterisP 5y ago
Digital processors are not needed for "basic agricultural computations" nor for communication. A ham radio can work and transmit/receive information without any typewriter-like automation (both microphones and manual Morse code don't require computation) and you can do all the calculations needed for ordinary agriculture on an abacus if needed. In any case, the extremely precise machining in large quantities required for any reasonable mechanical computer means that it's not a low-tech or high-reliability replacement for electronic computers, it's a very high-tech future solution to the limitations of current electronic devices. We can't make such devices now, we'll hopefully be barely able to make them in the future, but in the case of infrastructure collapse we'll lose the capability to make these advanced mechanical devices before losing digital devices. And nowadays computing is so ubiquitous, that every horrible scenario would still leave us with huge amounts of computing power lying around. Soldering together devices out of scrap electronic components is a much more reasonable way of doing post-apocalyptic computation than any of this mechanical computing tech; If there was a real need, nowadays you could extract a dozen usable processors from a crashed car or modern kitchen - even excluding computers and smartphones, so many modern devices have random processor(s) that are more powerful than what people used as their primary computers some decades ago. Things like wireless headphones and washing machines have chips that could be sufficient for so many computing needs. Apple's Lightning to HDMI cable has a powerful ARM processor inside. etc, etc.

nynx 5y ago

It’s a shame Merkle and Freitas never made their nanotech dream happen. There are some hints that a Canadian company called CBN Nanotechnologies is working with them (they bought all of Merkle and Freitas’ patents), but who knows.

Animats 5y ago

This reminds me of Drexler's early nanotechnology work. They need to fabricate something, anything, to validate that they can get those devices to work at that scale. Things are not Newtonian down there.

R0b0t1 5y ago
It reminds me of Neil Stephenson's "rod logic" (which is not really expanded upon) in the Diamond Age.
- kragen 5y ago
  That's Drexler's. He proposed it in Nanosystems and at NanoCon at about the same time; see http://www.nanoindustries.com/nanojbl/NanoConProc/nanocon2.html http://www.nanoindustries.com/nanojbl/NanoConProc/nanocon2.h... and https://hackaday.com/2015/10/19/rod-logic-and-graphene-elusive-molecule-scale-computers/ https://hackaday.com/2015/10/19/rod-logic-and-graphene-elusi....
jcranmer 5y ago
They do cite Drexler's early nanotechnology work... as more or less the newest citation for "hey, something like this can be built." It's really disappointing that rather key aspects of this are more or less ignored. There's not much of an analysis of things like what force propagation looks like, gate delay issues, mechanical energy losses due to heat, etc. I'm surprised there isn't anything like an analysis of macro-scale linkages to try to estimate what these things looked like if they were shrunk down to scale (even though, as you say, things don't work at that scale like they do at the macroscale).
- kurthr 5y ago
  I agree, and I know (and like) Eric and Ralph... but this is philosopher scientists and computer scientists looking at mechanical engineering (and material science) problems. Similarly, there was a lot of work at MIT on doing micro scale motors and turbines, which just work very different due to Reynolds # scaling, heat dissipation, and other issues. Realistically, you need a demand side on this sort of work for functional and realistic solutions to take hold (iterate and improve). Right now Moore's Law has driven semiconductor logic to the point where very little else can economically compete. Thus there's little investment of resources... or even accurate determination of markets. At the same time many people view nano (and micro) machining through Moore's lens, when in reality the low dimensionality, isolation, and fault tolerance of electrical/digital solutions allows for much more design independence/flexibility and complex system development. Imagine the vibratory harmonics of a mechanical computing system. There would be literally millions or billions of coupled oscillators. What about stiction/frictional losses due to surface chemistry/tension? What about limiting thermal noise effects? It's complicated and there's no money in it :^)

Koshkin 5y ago

Imagine the noise!

thedrbrian 5y ago

Sounds a bit like analogue computing OmegaTau has a podcast on them [ as well as plenty of other things ] https://omegataupodcast.net/159-analog-computers/ https://omegataupodcast.net/159-analog-computers/

klysm 5y ago

I want to be able to use this as a complication target.

mLuby 5y ago

A comment on the article's video had two interesting points: > There is linear stress problems that can occur, a single "activation" line (clock signal through a train of locks) can eventually reach a point that the mechanism will be unable to move at all. You can see an example of that long single activation line in this gif http://www.merkle.com/images/4_phase_shift_register_v4.gif http://www.merkle.com/images/4_phase_shift_register_v4.gif and described in more detail in this part of the video: https://youtu.be/yVX9Ob4SjGA?t=769 https://youtu.be/yVX9Ob4SjGA?t=769. I was surprised this wasn't explicitly addressed in the talk. Transistors aren't just powerful because they can be off or on (switch) but also because they can amplify signals (a lever). Edit: The paper does address this in section 4.5, saying that "The method for dealing with scaling is to isolate individual cells in a shift register (or more generally, individual clocked logic elements, such as Fredkin gates) by using the four-phase clock. That is, forces along signal lines cannot move beyond two cells before encountering a locked lock, preventing any accumulation of forces over long distances. At any point in time, a cell is either blocked from transmitting forces to adjacent cells, or can transmit forces only to one other cell (either the predecessor or the successor)" and "Force accumulation in link logic systems can be mitigated because each lock acts as a mechanical amplifier, with a small force controlling a much larger one." > I'm really loving the idea of mechanical logic gates and computers... it's far easier to produce in 3D print, which will be vital in space exploration and colonization. Question is, what additional primitives are needed to construct all the primitives (locks and balances) from a bin of links and rotors? Maybe a "lifter" that raises a hinge pin out of 2D plane so it can be placed in a rotor.

ffhhj 5y ago

This is fascinating. I always dreamed of building a circuit of light instead of electricity, controlled by micro-mechanical elements. An idea I got from reading an old book about building analog computers from wires, bolts, magnets, wood, etc. (forgot the name of the book, any ideas?)

chalcolithic 5y ago

I wonder why (it seems) no one has designed computer that would run on steam. Wouldn't it be easier to manufacture because of lower precision and wear resistance requirements?

nokita 5y ago
Babbage's analytical engine comes to mind.
- chalcolithic 5y ago
  My bad, I wasn't clear enough. I didn't mean mechanical computer powered by steam engine. I meant a computer that used steam to propagate signals, akin to hydraulic computers