Sunday, April 25, 2010

Omniscient Machines

I'm avoiding studying for exams, so I decided to discuss something I've been thinking about. I'll try to make it as short as possible so I can effectively convey my ideas.

We can certainly agree that the concept of randomness exists; in fact, the concept has many practical applications, particularly in computer science when random number generators are employed for some purpose. If we define a random number generator as one which outputs numbers with no predefined conditions, then it does not follow any pattern, and each time it outputs a new number, every possible output is allotted equal chance of appearing. Real-life random number generators mimic this definition, but they aren't perfect; every digital random number generator in existence must be seeded by some constant flow of input, whether this be the current system time, the digits of pi, or otherwise. Because of this, there is always some bias (however small) towards certain numbers, and thus the machine doesn't meet our qualifications to be considered a true random number generator.

From an analog perspective, we can consider the rolling of a six-sided die; theoretically, each time the die is rolled, it has an equal chance of 1/6 of rolling any of the six numbers. However, this isn't truly random either; it's impossible to make the die perfectly symmetric, so the geometry will always apply a small bias; on top of this, the number rolled is largely dependent on the initial force applied to it. If you could throw a die and know its exact position, speed, and trajectory, as well as the effects of air currents, restitution, and miscellaneous forces, you could predict with certainty which number it will land on. And if it's possible to predict the output, the die certainly can't be considered random.

It would seem that in any situation where we use the term "randomness", we're referring simply to a near-perfect simulation of randomness. From this claim, it would seem that true randomness is purely theoretical and doesn't actually exist in the real world.

If you accept this claim, then imagine the entire universe frozen at one instant. Suppose you had a machine capable of collecting every bit of data available at that instant, down to the position and velocity of every particle. Say this machine were capable also of carrying out calculations using the laws of physics. In this case, the machine would be able to eventually predict the exact state of the entire universe a moment (say, one millisecond) later. Through recursion, the machine would therefore be able to predict the state of the universe after two, three, fifty, or one million milliseconds into the future. Using this same principle, the machine could also work backwards, predicting the state one millisecond before, etc. Given enough processing power, the machine we have is effectively capable of telling us everything about the universe at any arbitrary point in time. Thus, looking forward into the future or backwards in time is possible.

I'm curious to see what everyone else thinks about this.

2 comments:

  1. Quantum mechanics says that this cannot happen, since it is impossible to predict the behavior of particles at a quantum scale. That is the whole principle behind Schrodinger's cat; it is impossible to predict whether the cat is alive or dead until you open the box and collapse the wave function.
    - Trevor

    ReplyDelete
  2. While Quantum Mechanics does effectively nullify my ideas regarding the collection, storage, and analysis of all the universe's data, it doesn't necessarily disprove determinism, which is only possible in the case where the future and past COULD be predicted. If you consider Quantum Mechanics to be simply another variable to take into consideration on the path between the initial conditions and the end result, true randomness still doesn't exist in that situation.

    ReplyDelete