First, lets look about that "how far out" your prediction is determined. You have to account for not knowing in advance the conditions which will exist. Is a car factory randomly producing cars because you cannot predict what the designers will design for next year's model? Is the car factory randomly producing cars because there could be a change in demand which will require a change in production volume? The owner could sell and the new owner could make changes. Does that make the car factory randomly producing cars?
I feel compelled to comment, because I was thinking about using this very example as an illustration. As you might guess, I was going to make the opposite point.
You see, I work in car factories. I go around and help them try and make more cars, faster. In my case, I write software for some of the machines that put together the cars.
One of the most common errors you see in car factory design is that the designers failed to account for the randomness in the car assembly process. Since the process is (allegedly) deterministic, they build it a certain way with those assumptions. It turns out, of course, that it is not deterministic. Machines break, at random intervals. Most car factories produce more than one sort of car, and they anticipate a certain volume of sales of cars, and then they design there plant to build cars at that rate, but the orders don't arrive at a uniform rate. Randomness causes clusters of orders, and lines end up slowing down because there aren't enough parts to deal with all the orders for one specific type of car.
Well designed car factories account for all the randomness. They predict in advance such things as spare parts inventory, overtime budgets, floor space needed for work in progress. All of these things must be calculated using random numbers, because if the stats are wrong, the predictions are wrong, and you don't make cars efficiently. I have been in car plants and listened to people blame our equipment for all sorts of line slowdowns when the real problem was that their calculations were based on assumptions of uniformity, without randomness in the system. It's a huge, huge, topic in the auto industry, in part because it is ofen done so poorly.
The car factory is, in fact, randomly assembling cars, and I've seen some really poorly built car factories that didn't work because the builders failed to design that in.
If the paint sprayer in the car factory is almost but not quite completely exact from car to car, are those cars randomly painted? If the cars' weights differ by some tiny fraction of a gram does that mean the cars have random weights? If a machine part gets slightly more worn with each car making each car slightly different, and every now and then the machine part gets replaced, think of the random variation that would cause in the cars.
Well, I can say that if you studied probability and statistics in college, and can't find a job in the insurance industry, the automobile industry is a good next best bet. I said that a lot of car factories failed to take into account the randomness, but even the bad ones know it's there. We hire lots of mathemeticians. The primary purpose of most of my software is to determine how much randomness is found in these sorts of processes.
I assure you that I spend a great deal of time thinking about "exactly" how much random variation that causes in cars, and I assure you that "random" is the "exact" term that I use.
But how many people would say the cars were randomly produced because of the minute randomness of the paint application? It would be absurd to do so.
But, I assure you, the pay is not bad. I wish it were a bit more, but it does put bread on my table.
I'm being a bit flippant, but I am not exaggerating. I really do make my living by saying that cars are randomly produced, and then quantifying that randomness. I work in the body shop these days, but when I moved to Detroit, my first job was in a paint booth, doing exactly what you are describing as absurd. I'm not rich, but I can afford private school tuition for my son. (Barely)
To call the manufacturing processes random because of molecular level differences would be ludicrous and useless.
But, as I said, the pay is not so bad.
Seriously, we do call manufacturing processes random, and we do use random numbers to describe them, and a great deal more time is spent on the "random" parts of car manufacturing than on the "deterministic" parts. What you describe as "ludicrous and useless" is how I make my living.
How would you decide what best described that car production as random or nonrandom? What criteria would you use?
The size of the paycheck. Random is bigger.
