Philosophy of randomness

[zoomboom]

It's always possible a random number generator can be flawed, but it's usually quite difficult to prove. There are services that use atmospheric noise and other such things to produce random numbers, which I guess are supposed to help it be even more random. I don't know if this site employs any method like that.

One could philosophize and ask God or the universe or other [wannabe] philosophers if anything is ever 100% truly random. I have a hard time personally understanding what 100% random should really mean. I don't think it can ever be proven that any method can produce 100% randomness- even if one can clearly communicate a precise enough definition of that... in fact, the definition of random that I find online is "made, done, happening, or chosen without method or conscious decision" -- and so it seems by that definition that it is quite impossible for any method to produce randomness, because the definition says "without method". Side note- in my understanding, "without A or B" should be interpreted as "(without A) and (without B)". One could argue however that this definition could be wrong, or imprecise, according to general understanding of randomness, or at least according to their own personal understanding of it.

I might have a problem with the word "chosen" in this definition, for instance. Or perhaps my problem is with the phrase "without method"... or perhaps with my own cognitive dissonance between those two things... Is it really possible for something to be chosen randomly? Even more particularly, without method? To choose something, whether it be by a conscious entity (ie, human) or by a non-conscious entity (ie, computer), I think it would have to require a method to make that choice...

Is there a better definition? Or other thoughts?

[robinzig]

It's not as if this isn't a topic into which a lot of thought has been put by a lot of clever people over the years...

As ever, the wikipedia article is a good introduction with many links to follow. In particular, given the things you ask, it may be useful to follow up the links to various definitions of randomness (in the sense of tests which can be done on a sequence of numbers which, while they can't "prove" it's "truly random", can give high confidence in it being "random enough"), and to generation of psuedorandom numbers which is in practice how the vast majority of "random numbers" are produced by computers, including I assume by BGA.

[Jellby]

Throwing dice or shuffling cards is very definitely non-random. Dice and cards follow high-school physics.

To all practical purposes,however, the results are most of the time random because they cannot be reliably predicted, not with a higher success rate (on average) than pure chance. Any method that leads to such unpredictability is random enough for BGA.

[Meeplelowda]

Any method that leads to such unpredictability is random enough for BGA.

This is my go to. It doesn't have the cryptographically securely random, just random enough.

I've put myself on a self-imposed ban on entering any thread that talks about rigging because it just gets me in a foul mood because I have seen the same tropes for decades (I'm old and been online for a loooong time). I have even given up and blocked anyone that raises it. But as long as we keep the discussion to the theory behind random physical processes, I'm on board.

[Shobu]

God secretly registered as a premium for getting better rolls

[zoomboom]

If there's a God (as I believe there is) who know all things, who knows the end from the beginning, then that would surely mean that nothing is purely random. But I agree that for most intents and purposes, we can speak of things being random. I wonder how far AI will take us towards discovering currently unknown patterns, to upset our gambling industries and whatnot. I think there are many things that are or seem random to most or all people at the moment, but in the future, will no longer be. Should we define randomness according to our perception of it? Will it turn out in the end that there's no such thing as randomness?

[zoomboom]

It's not as if this isn't a topic into which a lot of thought has been put by a lot of clever people over the years...

As ever, the wikipedia article is a good introduction with many links to follow. In particular, given the things you ask, it may be useful to follow up the links to various definitions of randomness (in the sense of tests which can be done on a sequence of numbers which, while they can't "prove" it's "truly random", can give high confidence in it being "random enough"), and to generation of psuedorandom numbers which is in practice how the vast majority of "random numbers" are produced by computers, including I assume by BGA.

Thanks for these links - yes, wikipedia's a good source of other people's thoughts as well. One interesting part to me is :

Cristian S. Calude stated that "given the impossibility of true randomness, the effort is directed towards studying degrees of randomness".[6] It can be proven that there is infinite hierarchy (in terms of quality or strength) of forms of randomness.[6]

Might be interesting to try to look at how he believes that can be proven.

I didn't yet look too deep into the randomness tests, but testing for randomness seems difficult to me... I guess they won't give a 100% definitive answer, but can give a certain degree of confidence in whether a data set is random or not.

It's also interesting that as you take more and more steps backward to see results of many supposedly random events, there is an aspect of those collective results which is far less random- which is the distribution of the outcomes. That distribution can be predicted with a lot more accuracy than can individual outcomes. As the main randomness article says, in regard to possible outcomes of differing probabilities, "randomness is not haphazardness; it is a measure of uncertainty of an outcome." (I might say an individual outcome.)

But I don't how if anything regarding the order of the outcomes becomes any more predictable? I would guess not.

[Jellby]

That distribution can be predicted with a lot more accuracy than can individual outcomes. As the main randomness article says, in regard to possible outcomes of differing probabilities, "randomness is not haphazardness; it is a measure of uncertainty of an outcome." (I might say an individual outcome.)

Well, you could say that individual outcomes are (or may be) random, but a probability distribution most certainly isn't (even if it is unknown).

Something that is often mixed up with randomness is "fairness" or "uniformity". People say "results are not random because I got 6 doubles in a row". Well, even if that could prove anything, it would at most prove that the probability is not uniform: a die that rolls a 3 90% of the times, can still be random if you don't know when the other 10% will happen

[Kayvon]

Here's an analysis of half a million random numbers generated by BGA:
https://boardgamearena.com/forum/viewto ... and#p59122

[Meeplelowda]

Here's an analysis of half a million random numbers generated by BGA:
https://boardgamearena.com/forum/viewto ... and#p59122

I'd not seen this before. But that's because, as mentioned above, I'd banned myself from rigspiracy threads.

Doesn't matter though. Someone can be smacked in the face with the data and simply ignore it because something *feels* off.