What's the most primitive computer possible?
September 29, 2010 4:05 PM Subscribe
Assuming knowledge of the theory behind logic gates and computation, when was the earliest people could have built a working working general purpose computer? Or to put it another way, imagine a modern day chip designer is trapped on a deserted island with no access to modern technology, but access to any necessary natural resources that he could acquire by himself, how could he go about building a general purpose computer to pass the time?
xkcd dealt with this
Anything that can simulate any kind of cellular automata will work as a computer.
posted by hobgadling at 4:32 PM on September 29, 2010
Anything that can simulate any kind of cellular automata will work as a computer.
posted by hobgadling at 4:32 PM on September 29, 2010
If you haven't already, you may be interesting in looking into Turing Machines.
A Turing machine is any (not necessarily digital) machine that can do all the functions of a modern computer chip. Turing worked out the math in the 1930s.
posted by auto-correct at 4:37 PM on September 29, 2010 [1 favorite]
A Turing machine is any (not necessarily digital) machine that can do all the functions of a modern computer chip. Turing worked out the math in the 1930s.
posted by auto-correct at 4:37 PM on September 29, 2010 [1 favorite]
I would think that an abacus doesn't count because computation with an abacus is in the method of using the abacus, not the way the abacus itself works.
A cellular automata could simulate general computation, but it would be an absurd amount of work to get anyway. I suspect the OP is asking the minimal materials for implementing the basic logic gates we now have in silicon.
Bruce Sterling wrote some steampunk where general computation was accomplished mechanically with gears, with motion the analogue to electricity (computer scientists bragged about how many miles of gears were in operation). I suppose a chip designer stuck on an island with sufficient tools could carve gears and rods and whatnot out of wood or bone--hook it up to a waterwheel and you've got something resembling a general computing device, including memory flip-flops.
posted by fatbird at 4:40 PM on September 29, 2010
A cellular automata could simulate general computation, but it would be an absurd amount of work to get anyway. I suspect the OP is asking the minimal materials for implementing the basic logic gates we now have in silicon.
Bruce Sterling wrote some steampunk where general computation was accomplished mechanically with gears, with motion the analogue to electricity (computer scientists bragged about how many miles of gears were in operation). I suppose a chip designer stuck on an island with sufficient tools could carve gears and rods and whatnot out of wood or bone--hook it up to a waterwheel and you've got something resembling a general computing device, including memory flip-flops.
posted by fatbird at 4:40 PM on September 29, 2010
Not all computers use logic gates. There's an entire category called analog computers. These may be easier to realize than a digital computer.
The Antikythera mechanism is an analog computer, dating from 400 BC and made of bronze. It isn't general purpose, but it does calculate. If the mechanism were generalized, it might qualify.
Your question is a tad on the vague side.
It would be a lot easier to make a 1-bit machine than a 64-bit machine and it would still qualify, though in utility, it might stink.
Back in the 1960's, there was a little plastic computer called Digicomp. I think it was 3-bits and was programmed via little pins. A handle was wiggled to 'clock' it, and it would do basic operations, very, very, basic operations.
end of man-splaining. see ya.
posted by FauxScot at 4:42 PM on September 29, 2010 [1 favorite]
The Antikythera mechanism is an analog computer, dating from 400 BC and made of bronze. It isn't general purpose, but it does calculate. If the mechanism were generalized, it might qualify.
Your question is a tad on the vague side.
It would be a lot easier to make a 1-bit machine than a 64-bit machine and it would still qualify, though in utility, it might stink.
Back in the 1960's, there was a little plastic computer called Digicomp. I think it was 3-bits and was programmed via little pins. A handle was wiggled to 'clock' it, and it would do basic operations, very, very, basic operations.
end of man-splaining. see ya.
posted by FauxScot at 4:42 PM on September 29, 2010 [1 favorite]
The difference engine was conceived of in the late 18th century... getting gearing that precise would be hard on a desert island, but I think it fits the idea of the question :)
http://en.wikipedia.org/wiki/Difference_engine
posted by wooh at 4:53 PM on September 29, 2010
http://en.wikipedia.org/wiki/Difference_engine
posted by wooh at 4:53 PM on September 29, 2010
Response by poster: Sorry, I meant specifically a general purpose programmable computer, not just an abacus or calculator.
posted by empath at 4:55 PM on September 29, 2010
posted by empath at 4:55 PM on September 29, 2010
Or to put it another way, imagine a modern day chip designer is trapped on a deserted island with no access to modern technology, but access to any necessary natural resources that he could acquire by himself, how could he go about building a general purpose computer to pass the time?
Cut down bamboo stalks, halve them, tie them to vines and run water through them under the force of gravity. Mr. Chip pulls vines to flip switches. Water runs downhill to other logic gates to perform some larger computation.
posted by Blazecock Pileon at 5:02 PM on September 29, 2010
Cut down bamboo stalks, halve them, tie them to vines and run water through them under the force of gravity. Mr. Chip pulls vines to flip switches. Water runs downhill to other logic gates to perform some larger computation.
posted by Blazecock Pileon at 5:02 PM on September 29, 2010
As mentioned above, you can look at what Turing worked out in the 1930s to understand the underlying theory (e.g. Turing machines and such). Basically, you can can build computing machines using just about anything -- transistors, electromechanical relays, tinker toys, pressurized air, marbles, etc, etc. Speaking very broadly, all of these machines are functionally equivalent. They can all compute the same things. But if you move away from theory into the practical world, these devices might compute things veeeery slooooowly. Or they may be very, very large. Or require huge amounts of energy. Etc. You can look at a marbles-and-wood adder here to get an idea of what I'm talking about. To build something that can compute something useful, and that is reasonably sized, etc, would require some significant capabilities (maybe being able to make small-ish metal parts accurately/consistently, maybe having electricity, etc.). Wouldn't be able to do much with desert-island kinds of capabilities (you'd be down around the marbles-and-wood adder :).
posted by madmethods at 5:12 PM on September 29, 2010
posted by madmethods at 5:12 PM on September 29, 2010
Could probably do it with a giant marble track thousands of years ago. Marbles of two different sizes could be used for 0 and 1 bits for the output stream and the input "tape". The state logic would just be some massively complicated track that you would trigger by dropping a single marble into at each step; it could toggle switches, release other marbles, etc. The marbles used to perform the state transition would be returned to the top by a water wheel. The state logic would be hard-coded, but that's fine, just make it a UTM. Takes forever, inconvenient to program, but it would work.
posted by equalpants at 5:12 PM on September 29, 2010
posted by equalpants at 5:12 PM on September 29, 2010
It would be pretty much impossible. It took rooms full of tubes just to make a *barely* general purpose computer work. Can you imagine what the size and energy requirements would be using twigs and rocks? You'd need a really good waterfall, or a shitload of horses.
posted by gjc at 5:40 PM on September 29, 2010
posted by gjc at 5:40 PM on September 29, 2010
Wasn't Babbage's Difference Engine a general-purpose computer? Not that it actually got made.
posted by AmbroseChapel at 5:52 PM on September 29, 2010
posted by AmbroseChapel at 5:52 PM on September 29, 2010
Now I remember what this question reminded me of, and why I immediately thought of an abacus. The Arthur C Clarke short story "Into the Comet," which I read probably 15 years ago.
As I recall, the computers went out on a spaceship. This left the crew stranded, as they needed the computers to compute the way home, or something like that, and everyone figured they were doomed. But the mathematicians figured out how to break the problem down into thousands of smaller chunks, the engineers cobbled together abacuses (abaci?) from spare parts, and a Good Time Was Had by All.
posted by Alaska Jack at 5:55 PM on September 29, 2010 [1 favorite]
As I recall, the computers went out on a spaceship. This left the crew stranded, as they needed the computers to compute the way home, or something like that, and everyone figured they were doomed. But the mathematicians figured out how to break the problem down into thousands of smaller chunks, the engineers cobbled together abacuses (abaci?) from spare parts, and a Good Time Was Had by All.
posted by Alaska Jack at 5:55 PM on September 29, 2010 [1 favorite]
Are you aware that the word "computer" used to have a different meaning? Now it means "a machine that performs calculations". It used to mean "a person who performs calculations".
It used to be that when complex calculations were needed, there was a room full of people using whatever calculating system was available (at various times they used abacusi, or slide rules, or books full of logarithms and trig and stat, or mechanical adding machines, or even just pencil and paper) to carry out a series of calculations needed to get the final result.
That's how they did structural design calculations for buildings and ships about a hundred years ago. It wasn't fast (especially by modern standards) but it was fast enough to be practical.
Your hypothetical castaway engineer could do exactly the same if there was a native tribe who could be bribed (or enslaved) into learning how to do such calculations for him. There's no real material requirement, especially if you're at the level of doing multiplication and division manually. A stick and a sandy beach would suffice.
posted by Chocolate Pickle at 6:10 PM on September 29, 2010
It used to be that when complex calculations were needed, there was a room full of people using whatever calculating system was available (at various times they used abacusi, or slide rules, or books full of logarithms and trig and stat, or mechanical adding machines, or even just pencil and paper) to carry out a series of calculations needed to get the final result.
That's how they did structural design calculations for buildings and ships about a hundred years ago. It wasn't fast (especially by modern standards) but it was fast enough to be practical.
Your hypothetical castaway engineer could do exactly the same if there was a native tribe who could be bribed (or enslaved) into learning how to do such calculations for him. There's no real material requirement, especially if you're at the level of doing multiplication and division manually. A stick and a sandy beach would suffice.
posted by Chocolate Pickle at 6:10 PM on September 29, 2010
As others point out, depends big time on what your requirements are to be considered a computer. If you're willing to accept very, very, very simple examples, then it can be done given some time. But beyond that, it becomes near impossible for a single person. Modern computers rely heavily on layers upon layers of abstraction to become a reality.
posted by Diplodocus at 9:10 PM on September 29, 2010
posted by Diplodocus at 9:10 PM on September 29, 2010
Alaska Jack: Does an abacus count as a computer?
No; an abacus is not programmable. It is merely a static register which allows the user to add and subtract numbers.
It would be pretty much impossible. It took rooms full of tubes just to make a *barely* general purpose computer work. Can you imagine what the size and energy requirements would be using twigs and rocks? You'd need a really good waterfall, or a shitload of horses.
gjc, it's not only possible, the Antikytheros machine could be held in your hand. A more general computer needn't be much (any?) larger.
posted by IAmBroom at 11:48 PM on September 29, 2010
No; an abacus is not programmable. It is merely a static register which allows the user to add and subtract numbers.
It would be pretty much impossible. It took rooms full of tubes just to make a *barely* general purpose computer work. Can you imagine what the size and energy requirements would be using twigs and rocks? You'd need a really good waterfall, or a shitload of horses.
gjc, it's not only possible, the Antikytheros machine could be held in your hand. A more general computer needn't be much (any?) larger.
posted by IAmBroom at 11:48 PM on September 29, 2010
The Antikythera mechanism is not a "general computing device" as the OP specified; it can only compute certain fixed functions such as the position of the sun and moon and planets on a given date.
You could definitely build something like a 4 bit CPU completely out of wood and vine using only hand tools. Whether it would be useful is another matter entirely, but it could be done.
posted by Rhomboid at 1:10 AM on September 30, 2010
You could definitely build something like a 4 bit CPU completely out of wood and vine using only hand tools. Whether it would be useful is another matter entirely, but it could be done.
posted by Rhomboid at 1:10 AM on September 30, 2010
I'm not the OP. And I didn't say the Antikythera mechanism was a general purpose machine. I said it was an analog computer, which it is. I suggested that it could be 'generalized', meaning that instead of being constructed with gearing that related to celestial cycles, it could be designed with gearing related to decades (or some similar numerical period) and then be used for if nothing else, modular arithmetic.
posted by FauxScot at 7:19 AM on September 30, 2010
posted by FauxScot at 7:19 AM on September 30, 2010
This thread is closed to new comments.
posted by Geppp at 4:16 PM on September 29, 2010