4 ms·

So I've known about slide-rules for a while. What I didn't realize was that the mechanical principles of slide-rules were applied to fire-control computers on

So I've known about slide-rules for a while. What I didn't realize was that the mechanical principles of slide-rules were applied to fire-control computers on WW2-era Battleships. https://en.wikipedia.org/wiki/Mark_I_Fire_Control_Computer https://en.wikipedia.org/wiki/Mark_I_Fire_Control_Computer These computers had slide-rule like cylinders that would rotate. As you aimed the gun sights "at" a target, the gun's coordinates were being mapped to a cylinder, which would move the gun a different direction (you have to aim the gun "higher" the further away the target is, to account for gravity "drooping" the bullet). In effect; the weirdly shaped cylinder inside the fire-control computer is a pre-computed solution that maps x/y coordinate inputs to a z-coordinate output (where "z" would be fed into the gun's raw aiming system). The different parameters of the gun were programmed in with gears and cranks, moving the cylinder around physically. For more details, see: https://www.youtube.com/watch?v=gwf5mAlI7Ug https://www.youtube.com/watch?v=gwf5mAlI7Ug Its a 20-minute video. The cylinder mechanism (of 2x inputs mapped to 1x output) is specifically talked about here: https://youtu.be/gwf5mAlI7Ug?t=636 https://youtu.be/gwf5mAlI7Ug?t=636 . Carefully crafted cylinders could be any pre-computed 2-input / 1-output function. --------- Mechanical computers are fun. Slide rules are but the simplest kind. People were using more complicated "mechanical computers" for very serious combat / rangefinding / etc. systems just a few decades ago, before the advent of modern digital computers.

7thaccount 5y ago
Mechanical computers are really cool. Veritasium just did a 2 part video on them. The second goes over some ML applications for the future. You can make a really tiny and energy efficient mechanical chip to do some cool things.
- HideousKojima 5y ago
  Technically Veritasium's videos were about analog computers (of which some mechanical computers are a subset, it's entirely possible to make a digital mechanical computer though it would be extremely slow).
  - ddingus 5y ago
    Yes, but when reading the parent comment I was struck by the parallelism in that analog computers and mechanical ones run on the rules of the world. Take a cam and follower, perhaps with a mechanical or hydraulic means to hold or change or respond to a state. That is not really different from a periodic signal fed into a trigger and or a flop or gate. And somewhere on here we all had a discussion about math and someone ran the test and determined the "simulation" we are living in is a twos complement machine. (That's really an artifact of our math and set of axioms we are working on), but it was fun to think about. By default, these kinds of computation systems are robust because the world is robust. And they can respond as things in the world, running on the world "simulation" tend to do.[1] Another related thought are those PONG games lacking a CPU. First time I encountered a schematic, I realized the circuit was the game! Robustness can be high value and fast! Extending it a little, one could take lookup tables, simple ADD operations and software that runs mostly out of a ROM. The old Atari 2600 (VCS) was like that having only 128 bytes of RAM! The Apollo flight computer was not powerful, in the sense of having exotic math functions, or tons of speed to throw at a problem. But they ran in real time, able to control a lot of I/O and such. These kinds of systems are remarkable in that they tend to be able to perform their designed for tasks well, and will always do that in a lean way, and we could always package that up, provide interfaces to far more complex systems and get way more out of it than one might expect. The other way we tend to get at these benefits is by putting things into custom silicon. A look back at some powerful, low latency solutions shows just what really understanding a need or problem well enough to make a circuit, or mechanism to deliver can do. (A lot) A while back I was setting up some advanced CNC CAM software for an injection mold shop specializing in high volume production. For a while they had an absolutely huge demand for 3.5" floppy disk cases, for example! They developed top notch, high performance mold tooling and had the process very highly automated. The vast majority of that automation was done with cams, both linear and rotational types, simple electrical valves to operate the necessary effectors, grips, pushers, and the occasional PLC here and there, and those were mostly about space and sometimes ease of development by newer engineers who just did not think in that way more than it was a requirement of some kind the mechanical means could not address. Similar means were in the machines themselves, analog computers taking inputs from thermocouple, more cams and such would maintain process consistency in a robust, will run for a few decades with no worries kind of way. I find this particular line of reasoning compelling. It is, in some ways, much harder than the familiar run code on a CPU connected to electromechanical devices and sensors way we see today. But, it is also easier in ways too. One notable ease was simply being able to understand what the intended behavior was just by looking! I, being a lay person with only at that time a shallow understanding of the concepts, was brought up to speed quickly by the older engineer teaching all this stuff to new team members, got it quicker than I expected. And it all can be done using ordinary tools, means, methods. Again, I find it interesting and compelling to learn about this stuff. We can do a lot, and did amazing things on what we could easily view as "stones and chisels" type tech, given a modern perspective. And take something like an EMP or some serious event or other. People with these kinds of reasoning skills and means, tools, methods could build, repair, and or engineer solutions to tough problems without anywhere near the degree of bootstrapping needed to get going from a "reset" class event. [1] - Not that I believe we are in some sort of simulation or meta world space. I frankly do not know and tend to focus my thoughts toward things that matter more to me and also found all this to be convenient and lean such that I need a lot less poking about on my mobile touch screen to express my thoughts.
wlesieutre 5y ago
I don't have one because they go for $1000+, but speaking of cylinders, the Curta mechanical calculator is fascinating https://en.wikipedia.org/wiki/Curta https://en.wikipedia.org/wiki/Curta https://www.youtube.com/watch?v=loI1Kwed8Pk https://www.youtube.com/watch?v=loI1Kwed8Pk
pugworthy 5y ago
Another cylindrical rule is the Thacher Cylindrical Slide Rule. It's a cylinder with many rules marked out on it, that rotates in a set of fixed rules. All told it is the equivalent of a 66 foot long conventional slide rule. https://www2.humboldt.edu/scimus/NorthCoastInst/ThacherCalculator/ThacherCalc.htm https://www2.humboldt.edu/scimus/NorthCoastInst/ThacherCalcu... I actually picked one up out of the garbage bin at the university I worked at. It didn't have an inventory sticker so they had to toss it according to the rules.

7thaccount 5y ago

Mechanical computers are really cool. Veritasium just did a 2 part video on them. The second goes over some ML applications for the future. You can make a really tiny and energy efficient mechanical chip to do some cool things.

HideousKojima 5y ago
Technically Veritasium's videos were about analog computers (of which some mechanical computers are a subset, it's entirely possible to make a digital mechanical computer though it would be extremely slow).
- ddingus 5y ago
  Yes, but when reading the parent comment I was struck by the parallelism in that analog computers and mechanical ones run on the rules of the world. Take a cam and follower, perhaps with a mechanical or hydraulic means to hold or change or respond to a state. That is not really different from a periodic signal fed into a trigger and or a flop or gate. And somewhere on here we all had a discussion about math and someone ran the test and determined the "simulation" we are living in is a twos complement machine. (That's really an artifact of our math and set of axioms we are working on), but it was fun to think about. By default, these kinds of computation systems are robust because the world is robust. And they can respond as things in the world, running on the world "simulation" tend to do.[1] Another related thought are those PONG games lacking a CPU. First time I encountered a schematic, I realized the circuit was the game! Robustness can be high value and fast! Extending it a little, one could take lookup tables, simple ADD operations and software that runs mostly out of a ROM. The old Atari 2600 (VCS) was like that having only 128 bytes of RAM! The Apollo flight computer was not powerful, in the sense of having exotic math functions, or tons of speed to throw at a problem. But they ran in real time, able to control a lot of I/O and such. These kinds of systems are remarkable in that they tend to be able to perform their designed for tasks well, and will always do that in a lean way, and we could always package that up, provide interfaces to far more complex systems and get way more out of it than one might expect. The other way we tend to get at these benefits is by putting things into custom silicon. A look back at some powerful, low latency solutions shows just what really understanding a need or problem well enough to make a circuit, or mechanism to deliver can do. (A lot) A while back I was setting up some advanced CNC CAM software for an injection mold shop specializing in high volume production. For a while they had an absolutely huge demand for 3.5" floppy disk cases, for example! They developed top notch, high performance mold tooling and had the process very highly automated. The vast majority of that automation was done with cams, both linear and rotational types, simple electrical valves to operate the necessary effectors, grips, pushers, and the occasional PLC here and there, and those were mostly about space and sometimes ease of development by newer engineers who just did not think in that way more than it was a requirement of some kind the mechanical means could not address. Similar means were in the machines themselves, analog computers taking inputs from thermocouple, more cams and such would maintain process consistency in a robust, will run for a few decades with no worries kind of way. I find this particular line of reasoning compelling. It is, in some ways, much harder than the familiar run code on a CPU connected to electromechanical devices and sensors way we see today. But, it is also easier in ways too. One notable ease was simply being able to understand what the intended behavior was just by looking! I, being a lay person with only at that time a shallow understanding of the concepts, was brought up to speed quickly by the older engineer teaching all this stuff to new team members, got it quicker than I expected. And it all can be done using ordinary tools, means, methods. Again, I find it interesting and compelling to learn about this stuff. We can do a lot, and did amazing things on what we could easily view as "stones and chisels" type tech, given a modern perspective. And take something like an EMP or some serious event or other. People with these kinds of reasoning skills and means, tools, methods could build, repair, and or engineer solutions to tough problems without anywhere near the degree of bootstrapping needed to get going from a "reset" class event. [1] - Not that I believe we are in some sort of simulation or meta world space. I frankly do not know and tend to focus my thoughts toward things that matter more to me and also found all this to be convenient and lean such that I need a lot less poking about on my mobile touch screen to express my thoughts.

wlesieutre 5y ago

I don't have one because they go for $1000+, but speaking of cylinders, the Curta mechanical calculator is fascinating https://en.wikipedia.org/wiki/Curta https://en.wikipedia.org/wiki/Curta https://www.youtube.com/watch?v=loI1Kwed8Pk https://www.youtube.com/watch?v=loI1Kwed8Pk

pugworthy 5y ago

Another cylindrical rule is the Thacher Cylindrical Slide Rule. It's a cylinder with many rules marked out on it, that rotates in a set of fixed rules. All told it is the equivalent of a 66 foot long conventional slide rule. https://www2.humboldt.edu/scimus/NorthCoastInst/ThacherCalculator/ThacherCalc.htm https://www2.humboldt.edu/scimus/NorthCoastInst/ThacherCalcu... I actually picked one up out of the garbage bin at the university I worked at. It didn't have an inventory sticker so they had to toss it according to the rules.