Remove ads

Introducing the LemonFools Personal Finance Calculators

Is anything random?

A virtual pub for off topic, light hearted pub related banter and discussion. No trainers
OLTB
Lemon Slice
Posts: 657
Joined: November 4th, 2016, 9:55 am
Has thanked: 518 times
Been thanked: 175 times

Is anything random?

#166233

Postby OLTB » September 14th, 2018, 9:33 am

I listen to an enjoyable podcast called, 'The Curious Cases of Rutherford & Fry' which is hosted by Dr Adam Rutherford and Dr Hannah Fry. A recent episode discussed the subject of randomness and I immediately thought that rolling a dice would be a random outcome. Apparently not. If one has knowledge of the weight of the dice, the angle of throw, velocity etc. you could make a decent enough attempt at suggesting the outcome. Even asking a computer to generate a set of random numbers isn't totally random (can't remember the reason why though...).

The only area in which randomness has been found to be genuine is in the quantum world - now forgive me, I am not a scientist and get lost very easily. There are quantum particles that just appear apparently and you can't predict where or when these particles appear. Jim Al-Khalili who was explaining this is very good, but I do tend to drift when I'm driving and don't quite understand what I'm hearing.

I would have thought that the travel of smoke from a fire would also be random.. perhaps not.. :?

Cheers, OLTB.

UncleIan
Lemon Slice
Posts: 553
Joined: November 4th, 2016, 3:35 pm
Has thanked: 260 times
Been thanked: 181 times

Re: Is anything random?

#166238

Postby UncleIan » September 14th, 2018, 9:45 am

OLTB wrote:Even asking a computer to generate a set of random numbers isn't totally random (can't remember the reason why though...).


I vaguely remember it's something to do with having to start somewhere to generate numbers from, so it's within the bounds of possibility that the starting position is the same, and therefore the random numbers generated from that point would be the same. In fact, I used to work on a system that used random numbers for testing, but they had to be the same set of random numbers, so you used to be able to seed the random number generator with a number, then the random numbers it would generate would always be the same, they'd be random, but each time you ran it, the same numbers were generated.

OLTB
Lemon Slice
Posts: 657
Joined: November 4th, 2016, 9:55 am
Has thanked: 518 times
Been thanked: 175 times

Re: Is anything random?

#166242

Postby OLTB » September 14th, 2018, 10:03 am

How strange! The episode also talked about the shuffle mode on iPods - originally, there were many occasions of songs being played in succession from the same album, and customers complained that the shuffle mode wasn't working. Apple had to then 'force' randomness by ensuring that songs from the same album were not played after each other when the shuffle mode was on.

Cheers, OLTB.

AleisterCrowley
Lemon Quarter
Posts: 2689
Joined: November 4th, 2016, 11:35 am
Has thanked: 510 times
Been thanked: 535 times

Re: Is anything random?

#166244

Postby AleisterCrowley » September 14th, 2018, 10:06 am

UncleIan wrote:
OLTB wrote:Even asking a computer to generate a set of random numbers isn't totally random (can't remember the reason why though...).


.. I used to work on a system that used random numbers for testing, but they had to be the same set of random numbers, so you used to be able to seed the random number generator with a number, then the random numbers it would generate would always be the same, they'd be random, but each time you ran it, the same numbers were generated.

Surely they'd be pseudo random ? - as in PRNGs, which are deterministic with 'random qualities'(rather than a stochastic processes - which are truly random)

UncleEbenezer
Lemon Quarter
Posts: 2456
Joined: November 4th, 2016, 8:17 pm
Has thanked: 227 times
Been thanked: 312 times

Re: Is anything random?

#166250

Postby UncleEbenezer » September 14th, 2018, 10:39 am

Yes, random is entirely possible, and indeed widespread.

However, that can be misleading: peoples understanding of randomness can be less than complete. Randomness doesn't imply that all outcomes are equally likely (nor even that there is more than one outcome, though the degenerate case isn't exactly interesting). It says nothing about correlation: there are many mathematical models that describe different kinds of relationships between random events.

Historically it's been the subject of "angels on pinheads" arguments: the determinist argument says that given sufficient data (the mind of God) you should be able to predict everything in the world, whereupon you conclude there's no such thing as free will because all is predestined. Or you make it something mystical: c.f. Paradise Lost. And trying to rationalise it can lead to manifest nonsense like the Calvinist notion of the Elect.

When you toss a fair coin in the normal sense, you have a genuinely random event, notwithstanding that it would be possible to work out the outcome with physics (and, with sufficient practice, might conceivably be possible to cheat). That's fine for determining such things as who goes first in a game.

The issue with computers is that it's not good enough for at least one key purpose: where computer security depends on randomness. On the one hand, a computer being much smaller than the physical world has hugely less entropy on which to draw. On the other hand, it has all the computing power necessary to reverse-engineer the physics of your coin-toss. So using randomness as a basis for security leaves open the possibility that an enemy will break your security by reconstructing your random events. Most likely they can only do that partially, but if they can reduce randomness to - say - a million possibilities, it's the work of a moment for a computer attack to try them all and see which one works. A lot of work has gone into the mathematics of modern cryptography that uses properties of (very large) prime numbers to ensure that you can only reduce randomness to an infeasibly huge set of possibilities that would take longer than your lifetime to try. And of course computers use a range of other defences.

There are other issues using them on computers, where the properties of a random number generator might affect the outcome of scientific studies like a stochastic simulation. A lot of work has gone into those, too. The issue there isn't randomness, but how it's used.

Damn, I've gone on far too long. Need to head over to Lidl for some random food.

SalvorHardin
Lemon Slice
Posts: 341
Joined: November 4th, 2016, 10:32 am
Has thanked: 404 times
Been thanked: 203 times

Re: Is anything random?

#166254

Postby SalvorHardin » September 14th, 2018, 10:52 am

OLTB wrote:I listen to an enjoyable podcast called, 'The Curious Cases of Rutherford & Fry' which is hosted by Dr Adam Rutherford and Dr Hannah Fry. A recent episode discussed the subject of randomness and I immediately thought that rolling a dice would be a random outcome. Apparently not. If one has knowledge of the weight of the dice, the angle of throw, velocity etc. you could make a decent enough attempt at suggesting the outcome. Even asking a computer to generate a set of random numbers isn't totally random (can't remember the reason why though...).

If you know the initial conditions and the physical laws under which the system operates then the outcome can be predicted with certainty as long as the laws which govern the system are not probabilistic (e.g. Quantum Mechanics). This is determinism, often called Laplace's Demon. Anything that can be modelled solely by classical physics is deterministic.

https://en.wikipedia.org/wiki/Laplace%27s_demon

The problem is determining all of the initial conditions. In the example of rolling a dice you need to measure many things which are difficult and/or expensive to measure; examples are the velocity of the dice when it is released (velocity in three dimensions), the position of the dice and each of its faces, the physical characteristics of the dice (mass, density, imperfections, coefficients of friction), those of the surface(s) it lands upon and those of the objects with which the dice will collide) as well as the gravity, temperature, humidity, air pressure, etc. of the environment. Plus any changes which will be introduced into the system after the dice have been rolled but before they come to a halt (e.g. change in air pressure and temperature because someone has opened the door to the room).

A truly random system system is one that can only be accurately modelled using quantum mechanics (i.e. atomic and sub-atomic particles). This is because the laws of quantum mechanics are probabilistic not deterministic. Consider two radioactive atoms of the same element (same isotope); we do not know when they will will decay, only the probability that each will have decayed by a particular time. If these atoms behaved according to the laws of classical physics then you would be able to state the exact time when each would decay.

In practice dice rolling is close enough to being a random process (for all practical purposes) because the roller doesn't possess enough information to be able to make such a calculation in a short amount of time or the skill to be able to fake a random roll whilst knowing the position of the faces when the dice is in their hand.

Most of the claims that Laplace's Demon doesn't apply in certain systems (e.g. thermodynamics) fail when you look closely at them because determinism only requires that you have complete knowledge of the initial conditions and the physical laws which apply. Chaos theory (e.g. the butterfly effect, behaviour of smoke) is one example; systems where chaos theory applies nevertheless follow well-defined deterministic laws and Laplace's Demon only applies when you have complete knowledge of the system (under chaos theory small changes in a system can produce massive changes in the outcomes).

scotia
Lemon Slice
Posts: 653
Joined: November 4th, 2016, 8:43 pm
Has thanked: 53 times
Been thanked: 163 times

Re: Is anything random?

#166289

Postby scotia » September 14th, 2018, 12:35 pm

ERNIE - "Prize draw uses heat for random numbers"
https://www.newscientist.com/article/dn ... m-numbers/

Back to pseudo random number generators which start from a seed. Although most software random number generators used back in my early life (a long time ago) employed a (comparatively fast) linear congruential generator technique, there were good, bad and indifferent implementations. To quote (from Numerical Recipes https://en.wikipedia.org/wiki/Numerical_Recipes ) "If all scientific papers whose results are in doubt because of bad RANS were to disappear from library shelves, there would be a gap on each shelf as big as your fist."

Gengulphus
Lemon Quarter
Posts: 2238
Joined: November 4th, 2016, 1:17 am
Been thanked: 1009 times

Re: Is anything random?

#166674

Postby Gengulphus » September 16th, 2018, 11:12 am

SalvorHardin wrote:A truly random system system is one that can only be accurately modelled using quantum mechanics (i.e. atomic and sub-atomic particles). This is because the laws of quantum mechanics are probabilistic not deterministic. ...

Not quite. Our stated laws of quantum mechanics are probabilistic, not deterministic, but that doesn't necessarily prove that the actual underlying way that the universe works isn't deterministic.

SalvorHardin wrote:... Consider two radioactive atoms of the same element (same isotope); we do not know when they will will decay, only the probability that each will have decayed by a particular time. If these atoms behaved according to the laws of classical physics then you would be able to state the exact time when each would decay.

Only if you knew every relevant aspect of the state of those atoms, sufficiently accurately to ensure that the inaccuracies in your predicted outcome were still sufficiently small by the time that the atom is predicted to decay. A key insight of chaos theory is that even if a universe were to work in a completely deterministic way, predicting what will happen in its future from its laws depends on knowing its initial state absolutely and completely accurately and performing your calculations absolutely and completely accurately. Otherwise, it's perfectly possible that the way the laws apply to the state you're considering causes inaccuracies to build up each time you advance your calculations forward in time, and if it does, the inaccuracies will eventually build up to the point where they overwhelm your results - i.e. your results become useless. Working more accurately and with more accurate initial data will mean that the calculations can be taken further ahead in time before that happens and so delay the point at which your results become useless, but it will still happen: only absolute and complete accuracy will delay it indefinitely.

As a thought experiment, imagine each atom had a tiny dice-rolling mechanism behind it, with the sequence of outcomes of its dice rolls determining when the atom decays. That dice-rolling mechanism cannot be seen directly by anything we possess, so only its indirect effects are things we can observe - i.e. when the atoms decay. And the way it operates is completely deterministic, but it does roll the dice in quite a vigorous, bouncing fashion that lets uncertainties build up chaotically (in particular, if a die bounces, quite small differences about exactly how it hits the ground on one bounce can cause differences in how fast it is rotating that lead to much bigger differences in how it hits the ground on the next, a recipe for chaotic effects). So we would see the times that each atom decays as random, even though they were deterministic and could in principle be determined exactly if we knew enough...

Basically, there's a sort of analogue of Clarke's 3rd law ("Any sufficiently advanced technology is indistinguishable from magic") that applies here, which is "Any sufficiently hidden determinism is indistinguishable from true randomness". (I should possibly add that various scientists have conducted experiments to try to reveal hidden determinism behind quantum effects, and they have had negative results that they have interpreted as saying that there are no "hidden variables" - i.e. that the effects really aren't deterministic. But those I've looked at seem to me only to say that any deterministic mechanism behind quantum effects must involve things that violate generally-accepted scientific principles, such as faster-than-light communication between the deterministic mechanisms underlying quantum effects in different locations. I.e. they certainly raise the "sufficiently hidden" bar that a deterministic explanation of quantum effects has to clear, but they don't make one entirely inconceivable...)

Anyway, what all that means is that as an answer to the question "Is anything truly random?", "Yes, quantum effects are truly random" is an excellent working hypothesis and it looks very unlikely that it will cease to be anytime soon, but it's not absolutely 100% certain. And that's basically as good as we ever get for any "scientific law" - so don't expect an absolutely certain answer to the question!

Gengulphus

melonfool
Lemon Quarter
Posts: 2635
Joined: November 4th, 2016, 11:18 am
Has thanked: 861 times
Been thanked: 469 times

Re: Is anything random?

#166690

Postby melonfool » September 16th, 2018, 12:06 pm

OLTB wrote:How strange! The episode also talked about the shuffle mode on iPods - originally, there were many occasions of songs being played in succession from the same album, and customers complained that the shuffle mode wasn't working. Apple had to then 'force' randomness by ensuring that songs from the same album were not played after each other when the shuffle mode was on.

Cheers, OLTB.


Yes, I read an article about this a few years ago - called something like 'How Many Tracks til Steely Dan', it was really interesting. Computers don't do true random because you have to tell them what to do and as soon as you tell them, it's not random.

It is a version of random though - my Amazon music has 11 James albums on it, and one Paul Simon, in the car the other day, I got through loads of tracks, of which 3 were Paul Simon, before it even played a James track.....

It's not just random - it's evil!

Mel

gryffron
Lemon Quarter
Posts: 1104
Joined: November 4th, 2016, 10:00 am
Has thanked: 56 times
Been thanked: 206 times

Re: Is anything random?

#166718

Postby gryffron » September 16th, 2018, 1:58 pm

melonfool wrote:my Amazon music has 11 James albums on it, and one Paul Simon, in the car the other day, I got through loads of tracks, of which 3 were Paul Simon, before it even played a James track.

Random does not mean equally or evenly distributed.

Most MP3 players are deliberately not random. Not even computer pseudo-random. Because people get upset if they play the same track twice in succession. Or a whole load of tracks from the same album ( :shock: ). Which could of course easily happen if it were truly random. It's actually quite common for the algorithm to deliberately play 2-3 tracks from the same album/artist/genre, and then move on. Makes it sound more like the radio!

Gryff

Melanie
Lemon Slice
Posts: 429
Joined: March 7th, 2018, 8:14 pm
Has thanked: 382 times
Been thanked: 38 times

Re: Is anything random?

#166731

Postby Melanie » September 16th, 2018, 3:57 pm

SalvorHardin wrote:A truly random system system is one that can only be accurately modelled using quantum mechanics (i.e. atomic and sub-atomic particles). This is because the laws of quantum mechanics are probabilistic not deterministic. Consider two radioactive atoms of the same element (same isotope); we do not know when they will will decay, only the probability that each will have decayed by a particular time. If these atoms behaved according to the laws of classical physics then you would be able to state the exact time when each would decay.

My contribution to this, is that any system at an atomic/sub-atomic level can *never* be accurately measured, because the things with which science uses to measure stuff, i.e. photons of light (for position and distance)/vibrations of caesium atoms (for time), are similar in size/energy to what you want to measure, and hence the measuring device and thing you are measuring will always perturb each, thus all the scientist is left with is the probabilities of things being where they are. Or something.

Matt

melonfool
Lemon Quarter
Posts: 2635
Joined: November 4th, 2016, 11:18 am
Has thanked: 861 times
Been thanked: 469 times

Re: Is anything random?

#166744

Postby melonfool » September 16th, 2018, 5:37 pm

gryffron wrote:
melonfool wrote:my Amazon music has 11 James albums on it, and one Paul Simon, in the car the other day, I got through loads of tracks, of which 3 were Paul Simon, before it even played a James track.

Random does not mean equally or evenly distributed.

Most MP3 players are deliberately not random. Not even computer pseudo-random. Because people get upset if they play the same track twice in succession. Or a whole load of tracks from the same album ( :shock: ). Which could of course easily happen if it were truly random. It's actually quite common for the algorithm to deliberately play 2-3 tracks from the same album/artist/genre, and then move on. Makes it sound more like the radio!

Gryff


God, I really get fed up of people telling me stuff that, if they read what I wrote (and don't just select one quote), I VERY OBVIOUSLY KNOW.

Sigh.

Mel

Gengulphus
Lemon Quarter
Posts: 2238
Joined: November 4th, 2016, 1:17 am
Been thanked: 1009 times

Re: Is anything random?

#166883

Postby Gengulphus » September 17th, 2018, 11:05 am

melonfool wrote:Yes, I read an article about this a few years ago - called something like 'How Many Tracks til Steely Dan', it was really interesting. Computers don't do true random because you have to tell them what to do and as soon as you tell them, it's not random.

Actually, some computers do true random, at least as far as we know (for the reasons in my last post, we cannot be entirely certain that anything is truly random). They do that because the instructions about what to do that they contain are to read the state of a physical device that is part of the computer and that generates an output that depends on quantum effects. For example, such a device could consist of a small radioactive source and a radiation detector, set up such that the chances of at least one atom in the radioactive source will decay and have that decay detected are about 50% each millisecond. To get a random bit, the computer reads the device in two consecutive milliseconds. If it sees a decay followed by a non-decay, it produces a 0; if it sees a non-decay followed by a decay, it produces a 1; if it sees two non-decays in a row or two decays in a row, it doesn't produce anything immediately, but repeats the whole procedure until it does produce a 0 or a 1. (The reason for that somewhat elaborate procedure is to make it equally likely to produce 0s and 1s even if the chance of decaying in a millisecond are not exactly 50%.)

That's not standard in computers because hardware randomness generators do tend to add to the cost of the computer, be slow or both. But they do exist and do get used when true randomness is sufficiently desirable.

I should incidentally say that my example above of a hardware randomness generator using radioactive decay is chosen for using a source of true randomness (as far as we are aware) described earlier in the thread, not for being particularly cheap, fast or state-of-the-art!

Gengulphus

StepOne
Lemon Slice
Posts: 396
Joined: November 4th, 2016, 9:17 am
Has thanked: 50 times
Been thanked: 65 times

Re: Is anything random?

#166939

Postby StepOne » September 17th, 2018, 2:23 pm

For an example of actual randomness, there is the location of my car key after my wife has borrowed it.

If there is any pattern to it, it's not yet become apparent as we have only been together 15 years.

Cheers,
StepOn

Melanie
Lemon Slice
Posts: 429
Joined: March 7th, 2018, 8:14 pm
Has thanked: 382 times
Been thanked: 38 times

Re: Is anything random?

#167154

Postby Melanie » September 18th, 2018, 12:25 pm

melonfool wrote:3 were Paul Simon, before it even played a James track.....

It's not just random - it's evil!

If it missed out "She's a Star" or "Laid", I'd be inclined to demand a refund!


Return to “Beerpig's Snug”

Who is online

Users browsing this forum: Google [Bot] and 0 guests