What is a computer?
Without saying “something that computes”
I want to take a brief interlude from the usual schedule to introduce an important concept. Namely, what is a computer? The obvious answer is “something that computes” but this will hardly do. For one, this answer is tautological — that is, true by definition. It’s like answering “what is a car?” with “something that does car things”. Let’s try to find a more substantial, useful answer.
One way to define computers would be to list every single computer in existence. When someone asks what a computer is, just point to the list. Even ignoring the logistical nightmare this would entail, this approach still has a major flaw: who decides what counts as a computer and what doesn’t? We’ll all agree on some examples. If you downloaded this with a browser and are now reading it on a screen, then you’re using a computer; no one reasonable would contest that. But what about a microwave oven? Most microwaves have circuitry of some kind and buttons to set various options. There’s usually a primitive screen that includes a timer and clock. Is this a computer, if only a primitive one?
I think so and you probably do as well. Anything with buttons, a screen, and circuitry just screams “I’m a computer!” If it has speakers and audio software, anyway. But what about a toaster? Most toasters consist of, from a user’s perspective, two slots in which to place bread and a lever to lower the bread into a toast-producing inferno. Is there a fundamental difference between a microwave’s computer-ness and a toaster’s? Does it make sense to call a microwave a computer but claim a toaster is not one?
The thought process is probably more elucidating than the answer. If you’re going to draw a line between computers and non-computers, where should it be? Perhaps there shouldn’t be a line in the first place. But if there’s no line, what is it? A spectrum of computer-ness with a supercomputer at one end and a protractor at the other?
It may even be more complicated than that. Perhaps what comprises “computer” is not one indivisible property but a collection of independent properties, only some of which must be satisfied for an object to be deemed a computer. Sorry, that was a very computer sciencey way of putting it; let me rephrase that into more natural language. Take soda, for instance. The criteria for being called “soda” is something like this:
- Edible
- Liquid
- Water based
- Sugary
- Carbonated
- Nonalcoholic
But some drinks, like Grown-up Soda, aren’t very sugary at all. We still call them soda, though, because they’re close enough to the definition to be included. For the same reason, noncarbonated soda is still called soda because it meets the rest of the requirements. From this perspective, a definition is not an indivisible, monolithic requirement. Rather, it is a loose collection of requirements that need only be approximately satisfied by an object for that object to match the definition.
So getting back to the definition of computers, perhaps we should break down the definition into individual parts. If an object matches some of the parts then it may be considered a computer, from some perspectives anyway. Here’s an initial suggestion for the definition of computer:
- Able to receive information
- Able to produce information
- Able to communicate information
This may sound extremely broad and vague — and it is — but anything more specific risks excluding perfectly valid computers. To clarify, I would not consider this the official definition of a computer nor do I think it’s a perfect match. But it’s a good start for at least one reason. Specifically, it focuses not on what the computer is made of (because that’s really not relevant) but instead on what the computer does (which is very relevant). Notice that I didn’t mention having circuitry as one of the requirements. This gets back to that strange image at the top of this post. That image is a closeup of the Antikythera mechanism, an device found in an ancient shipwreck off of the island of Antikythera, just northwest of Crete.
This device, uncovered at the turn of the 1900's but not analyzed and understood until the 1970's, is an analog computer that displays various astronomical data. By turning a crank on the device, a user could set which date the machine would display. The machine had a dial that indicated the locations of the sun and moon at the date specified. There were even indicators for eclipses and moon phases. It’s pretty amazing, especially since it was built in the first century B.C.E.! We don’t know of any computers that rivaled it until the Renaissance.
Regarding its description, it’s called “analog” in opposition to “digital”. Digital computers work with integers (whole numbers). The integers are the numbers you count with. 0, 1, -1, 2, -2, all the way to infinity (literally ad infinitum). They can’t have decimals values, like 3.4 or 90.1, or fractional values, like 5/6 or 10/4. On the contrary, analog computers work with reals (continuous numbers). Reals can be decimals, fractions, and exotic numbers like √2. Most computers nowadays are digital, though it hasn’t always been this way. Actually, most computers are a very specific kind of digital: binary. Binary computers work entirely with 0's and 1's; binary indicates a basis of 2 numbers. Everything from numbers to words to images to sounds is stored as a series of 0's and 1's. Needless to say, we’ll talk about this a lot more!
You can see how different the Antikythera mechanism is from a modern binary computer. Rather than working digitally in 0's and 1's, the Antikythera mechanism operates with cranks and gears and dials — all analog machinery. The machine’s date that the user sets with the crank is not a series of distinct days or weeks but rather a continuous line; just as there is no one point at which day turns to night, there is no one point at which one day turns to the next on the machine; each tiny spin of the crank advances the date ever so slightly. Reals, not integers, are at play here.
Just as there are no binary divisions in the Antikythera device, perhaps there are no binary divisions in the definition of computer. The definition could be a list of requirements that objects more or less adhere to. In other words, rather than being entirely a computer or not at all, most objects would fall somewhere in between and it would depend on the context. The term “computer” becomes an ideal some information-related machines closely approach and others vaguely resemble. A laptop is quite close to the ideal while a ruler is far away. And it depends on context: if a laptop is being used as a footrest, it’s not being very computer-like. Conversely, if a belt is being used to measure distance (maybe measuring in belt-holes), then it is acting like a ruler — a simple computer.
This is just the beginning of a much longer discussion. We need to talk about ways to define information, ways to measure it, ways to store it, and so so so so many other things. Don’t worry if the Wikipedia pages I just linked to don’t make much sense — they will, eventually. If those links intrigue you, please keep with it. If they don’t — well, you’re probably reading the wrong blog (but keep with it anyway if you want!).
