I've had an urge to discuss this topic for a long time. Hopefully there are some other people on here who also find it as interesting as I do.

I'll start with the last paragraph of this article: http://engineering.mit.edu/live/news/1753-can-a-computer-generate-a-truly-random-number



The article claims that thermal and atmospheric noise are "random". It sounds like they are saying that because it is not human-defined and not able to be predicted by humans.

What I question is this: What defines if something is random?

- If it's defined purely based on lack of human predictability, does that mean, for example, the lunar cycles used to be "random". But now that humans understand them, they are not random anymore? The cycles themselves never changed of course.

- What about earthquakes, are they random? We can partially predict them now. Maybe some time in the future with improved technology we will be able to predict exactly when, where and with how much force they will strike years ahead of time. Will that semi-random event become non-random?

- What about the simple act of a person flipping a coin into the air and letting it land on the ground? We can't predict what side the coin will land on, but isn't the movement of the coin following the laws of physics exactly? What's random about it?

If a hundred computer programmers spend a hundred years writing an algorithm with as many complicated variables as exist in the physics of a coin flip, would the results be any less random than a man flipping a coin?

This post was edited by kayeto on May 16 2013 11:56pm

why do you combine your question with computers?
imo you should have left computers out of discussion, because it's obvious that a algorithm can't produce a random numbers. there are only pseudo random number generators, no matter how long the algorithm.
that's why i voted "no" (i answered the question in the topic name)

but if you take some physical information and use them in the software, it gets impossible for reverse-engineers to predict numbers. "if the physical states are really random", thats a very philosophical question. you can even use the result of the "flipping coin" as the entropy source... keep that in mind.

I tried to tie it together with the question in the last line.

>Do you think me throwing a coin is 'truly random' (i.e more random than pseudo-random)?
>Isn't the physical motion of the coin flip nothing more than billions of atoms interacting in a scripted manner (the laws of physics)?
>How is that physical action any more 'truly random' than a huge program with just as many billions of variables?



I guess resolving that issue is where the conversation should begin then. I fail to see anything random about billions of atoms acting exactly as nature designed them to.

This post was edited by kayeto on May 17 2013 01:50am

NO

if you manage to get a truly random input into the computer the result could be a random number but then the randomness is not computer generated but comes from the input

as already mentioned, algorithm-based randomness is not real

a coin flip is more of a chaotic process, i would say it is deterministic nevertheless (the outcome is either or, so the chaotic nature can be interpreted as randomness)

flat out no

If you take some process and start it with certain initial conditions, and sometimes get different results, the results are random. But in almost every case where this seems to happen, if you look closely you will realize that in fact you failed to reproduce exactly the same initial conditions, so the different outcomes were the results of different initial conditions and not indicative of anything really random.

Let's take a real world example: Let's say you make machine/robot that is designed to thrust a pool cue (at a pool ball, on a pool table). So we will set it on the table, cock the arm/pool cue and have it hit a ball in a certain direction, with a set force and the ball goes flying around the table for a while and comes to a stop. I think you realize that if you were to try to do it again, the ball is not very likely come to the same location on the table the second time, no matter how careful you are. There will be all kinds of difficulties in reproducing the original shot. The spring in the arm of the robot will not quite have the same tension as the first time. How will you get the angle and initial placement of the ball exactly the same? The felt on the pool table will be slight different based on the friction from the previous shot. There might be slightly different air currents in the room. All kinds of nasty little problems. However the better job you do addressing these problems, the closer you will come to the original outcome.

There are however ways to address these nasty problems with extremely simple interactions to create real random numbers, but for most practical purposes you can get extremely good pseudo random numbers out of a computer much easier.

This post was edited by Azrad on May 17 2013 12:02pm

Yes. Quantum randomness has been proven to exist without hidden variables.

So you are saying that a man flipping a coin isn't random, right?

He won't exert the exact some force on the coin twice, so the results will be different but not due to randomness. Just due to different input.

yes