True Random

If you insist on restoring this page...one would prefer that you do it in its entirety.

TrueRandom is (1) n. - any event or process that occurs without a cause or source. (2) adj. - not having a cause or previous pattern. synonyms - PerfectRandom?


It seems like this definition equates "True Random" with "True Random Event." A more appropriate definition might be worded "The quality of..." In my own opinion, it would be the quality of being completely unpredictable, even when given all information about the qualified except its outcome.

This of course gives us a definition in terms of predictability. By it, a dice isn't truly random. However (as mentioned below) the abstract concept still applies, since the players at the table don't know enough to figure what the roll will be.


It really didn't need the "and could not be predicted."


The above definition was added after much back-and-forth on this page. However, I find it is still lacking, for two reasons. Firstly, the useful parts of the definition are *still* not different from 'random'. Secondly, it conflates causality with determinism, but causality is not useful here: this assertion of a lack of any cause is a bit strange. If what you really want to do is assert the existence of supernatural events, why not do this? This assertion may have philosophical utility for you. It cannot have physical utility (by definition). It has nothing useful to add to the concept of randomness (in other words, anything useful it adds is completely separable from the concept of randomness). So in effect, you are looking for a definition which mixes up two separate, and separably useful ideas. Why do you wish to do this? In common usage we have random: synonymous with stochastic originates from Greek, for 'pertaining to chance', and is the opposite of 'deterministic' (meaning a process whose working can in principle be calculated).

...

(philosophical debate about "events without cause" moved to WithoutCause.)


6) What's the difference between a supernatural event and a random event?

(confusion about whether radioactive decay does, or does not, have a cause, moved to WithoutCause.)


Random taken to an extreme. Not defined by lack of predictability but by lack of source/cause because anything with cause is truly predictable by at least an abstract omniscient being.

With these restrictions, the idea is pretty meaningless. If atomic decay is not 'random enough' for you, you will have to resort to concepts like an 'omniscient being'; but these do not have much utility.

Take 0 and 1 and pick one (no pun intended) based on absolutely nothing.

This is an abstract concept that is currently believed to not actually exist in life.

[Surely (one interpretation of) quantum theory says it does, for everything, hence the atomic decay example.]


In computers, hardware, software, anything to do with it, it is impossible to mimic this TrueRandom action. [Proof? Does the assertion include a hypothetical computer with an unlimited amount of memory?]

As below, this all has to do with human logic also. If one is willing to leap out of human logic for an indeterminate amount of time, one might find that TrueRandom and random and whatever else you've got exist perfectly well.


In life, it is doubly impossible to mimic this random action. The reasons you pick one (no pun) are always more than nothing, regardless of your lack of knowledge of the reasons.

There is no TrueRandomness in life, everything is cause and effect as far as human logic is concerned.

This is written to explain an inconsistency in a particular belief of free will. The belief that FreeWill means not being influenced or directed by a higher power when making one's choice, thus the higher power can not predict or know the future if FreeWill exists. A higher power could know the future through deductive reasoning by watching all of the events that led up to the choice and that would still allow for FreeWill unless TrueRandom really exists. Unless the higher power is also held responsible for creating the physical universe, since everything in that would then be entirely due to such influence.

I have a hard time believing in TrueRandom because even with one theory that explains the possibility of FreeWill and the omniscience and omnipotence of God (that of there being a part of God (who is outside the 4 dimensions of our universe, yet interacts with it while obeying its "laws") in all of us that allows us to be able to make value decisions, to choose bad or good, or whatever), I find it hard to believe that there is something random about those choices... if it's all just random, what would be the point? If the reason I didn't choose the bad decision is purely one of luck, being at the right place at the right time, if it was random... I don't see a point, with or without God. Then again, I also don't see a point in having all of my decisions made based on my past experiences. There's got to be something inbetween or totally away from those possibilities. Anyway, I figured I'd argue against and define what I considered to be the bigger evil - I dislike the concepts of infinity and TrueRandom (in material reality).

[Nature isn't required to have a point, nor to conform to your arbitrary likes. I quite like the idea that what happens is partly random and pointless. Even if nature is entirely deterministic, it can still be pointless.]

You are right. Nature isn't required to have a point or confirm to my opinions. Listen carefully: The above paragraph never said it did. The above paragraph never said that an entirely deterministic nature would have a point. It actually argued for middle ground. Nature isn't required to have a point. It would just make more sense to a number of people if it did.

Of course, if we take the human mind to be a chaotic system, or if we take even part of it to be a chaotic system (I'm going to go with large clouds of chemicals, aka hormones, which definitely influence the thought process), proving free will reduces to proving that some input to that system (e.g. quantum bubbles) is truly random.


The "based on absolutely nothing" criterion is not really necessary. Whether the results are generated by coin flips, die rolls, radioactive half-lives, the current value of a rapidly changing bit, or God's whim doesn't matter. What is important for "truly random results" is that (a) there is no way for the observer to predict the outcome, and (b) there is no rational reason for the observer to believe that one result is more likely than any other. (Note that (a) is really implied by (b).)


Maybe the abstract concept should be renamed. Any suggestions?


The most workable definition of "random" is the unpredictability of [an event]. TrueRandom may actually be an attempt at PerfectRandom?.

A thing is truly random if no-one relevant can predict it. A thing is perfectly random if no-one in any context can predict it. An event need not be PerfectRandom? to be TrueRandom.

If the observer(s) in the relevant context can't predict it, that's the threshold needed for TrueRandom. The more computational power is brought to bear on the context, the more obscure the "random" selection needs to be.

Are we trying to determine the framework for "random enough" or "random beyond all computation?"

If nobody at the table can tell which way the dice will stand, that's random enough for that game. It is uninteresting to the context that someone with a computer and six weeks to burn could "predict" all the outcomes of four hours of play. You can't bring the results to bear in time enough to matter.

It will always be possible to postulate enough computational power to predict an event. Whether such power can be brought to bear is the crux of TrueRandom.

How could you postulate enough computational power to predict quantum events with arbitrary entanglement allowed? In the limiting case, you would require all state in the universe, so postulating enough computation power is nonsensical.

[I cheat. I never said the computational power had to exist in the universe being computed.] Not good enough. You have to explain some method you can apply that computing power without interacting with the universe. Of course, you can posit a passive connection to every bit of matter in this universe, being fed into some vast computer in a much larger universe. Since you can't feed any of the information from that simulation back into this universe though, the concept is pretty useless.

Note, however, I'm messing with the definition of some else's words, so we'll call this an "opinion."

The point of defining TrueRandom is to define the random events that no omniscient being could predict. Impossible. An omniscient being with enough power to know everything should also have enough power to compute the result. It could look at the surface of the table, the wind molecules, the gravitational pull, the moron on the left who's going to hit the dice because he's a jerk, and the weight of each molecule of the dice to see where its going to land and how on the table.

There are assumptions implicit in these statements, that the omniscient being is somehow able to view this universe without being a part of it, etc. If you want to posit such a being, go right ahead, but the concept doesn't have much utility.

Actually, I didn't assume that the being couldn't be part of the universe. It seems that you're assuming that you must be apart from something to know all of it.

That isn't an assumption, it is basic information theory. If your being is part of the universe, it interacts with it and therefore complete state of the being must be taken into the calculations as well, but this is recursive so it can't logically hold. If the being is outside (meaning non-interacting), then it is at least logically sensible to talk about it's calculating the state of everything, however implausible...

[Isn't theory just an unproved assumption?] No.

[Original poster did not write the following questions up to the line.] You mean the same way it's not possible for a program to output a listing of itself?

No, that is completely different. If you think about it for a minute you will see why: the program needs to use extra storage to do this. The universe can't do this for itself, since by definition there is no extra storage. In your example above, it is not the program that outputs a listing of itself, it is the program+runtime that outputs a listing of the programs. Apples and oranges.

So the program can't also output the contents of whatever "extra storage" it is using?

No, because it isn't the program we are talking about. I would have thought this was fairly obvious, but apparently not... so to make things very explicit: The equivalent question is not 'can this program reproduce it's own listing completely', but 'can this *computer* reproduce it's own *state* completely. So tell me, can you show me a computer system that can model its own state completely, not counting the fact that it is in that state. In other words, you need a program that will tell you the state of every bit in memory, on the hard drive, in registers, etc, including the state that represents the program ... and therein lies the rub.

It's not at all difficult. The computer outputs any particular given total state, and then puts itself into that state. The objection that the computer did this in the wrong order doesn't work because you don't know that the computer in fact did it in the "wrong" order, just that the computer said "here's my total state" and was correct about that.

This is still missing the point. There is nowhere to output *to* without changing the state of the computer (there is nothing outside the computer, remember). There is nowhere to hold the state describing the state, at least in the general case. You may be able to find, like a QuineProgram, particular configurations of the state that are self-describing, but this certainly doesn't allow the machine to model itself exactly. You can try all the contortions you want, it is simply not possible for the machine to hold an image of itself (other than the trivial one), which is the whole problem. Note also that your scenario breaks down on another level: This whole analogy, weak as it is, is about modeling the *evolution of a system to make predictions. So in your case, even if you could describe a snapshot of the state (which in general you can't) you can't evolve that state - your program will step on state whatever it tries to do.


(more stuff about atomic decay moved to WithoutCause)


[Note that even if radioactive decay is truerandom, we can't use it unless we have perfect detection of non-random events which significantly affect our observation apparatus.]


Do scientists believe they will never discover the pattern behind the rate of atomic decay?

Depends on the scientist. The standard QM model suggests that there is no pattern to discover.

... which would suggest that it's ... truly (perfectly?) random ...

Right. But not in any sense that is different from 'random'. This 'truly random' seems to be a definition looking for a purpose. That is, there are random things, and then there are 'truly random things', that only differ from random for some hypothetical being (i.e. for humans there is no difference). As far as I can see, random is a perfectly good term and 'truly random' is essentially meaningless, but I may have missed something...

The difference between "random" and "truly random" as I see it would be that events that are random could be in principle both predicted and influenced whereas a "truly random" event with no cause could be neither predicted nor influenced. To use atomic decay as an example, if this is merely "random" (as opposed to "truly random") then it may be possible to induce the decay to occur, which would be a fantastically useful thing to be able to accomplish.

But this is useless, because you are describing the difference between random and pseudo-random. If an event is random you cannot, in principle or otherwise, predict it or influence it. It is a perfectly good term with a perfectly good definition; why dilute it? Going back to the atomic decay example - under the current best understanding of science, there is no reason to believe that atomic decay is in principle predictable. If our understanding of the physics changes so that this is not so, atomic decay will no longer be considered random (but probably would be considered one heck of a good pseudo-random source).

But the definition of random isn't one of influence, only predictability. It's possible to influence and do a whole host of other things to random things without affecting its unpredictability. Isn't it possible to be unable to predict something, and use it as random, but it not actually be TrueRandom? I think so - just because we haven't found a way to predict atomic decay, it doesn't mean it's truerandom. It's random, and may always remain random to us, but to someone else, an alien, God, a dog, it may not be random, they can predict it.

I don't see what your point is. I said above you cannot *even in principle* *predict* the decay, and only added 'influence' because you had it in the previous paragraph. Perhaps all of this sidetracking into a particular example of a physical phenomenon which is considered random is the problem, so let's step back. My position is this: You (or anyone else on this page) have not provided a useful or logical differentiation from the perfectly good term 'random'. I can see that you are trying to make a distinction, but I can't see *why*. All of this vague rambling about hypothetical beings (some of which has been logically inconsistent) has not created a useful distinction between the terms. The term 'true random', so far as this page as presented it, is useless. It has exactly the same value as if I said: I want to define pink-unicorn-random - it is just the same as random, except that the pink-unicorn can predict it.

Well, I can't explain what everyone's point is, I've only inputted a little here and there. Let me try... truerandom defines an event with NO cause, NO influence. Usually, this means it wouldn't be predictable. Using the atom decay as only an example, and a rough one since I haven't studied it, it would be like heating it up, cooling it down, shaking it around, nothing could cause it to change the rate it decays. Almost like a, 'I think therefore I am,' instead, 'I do therefore I do.' This definition was started to try to wrap an implied definition (not held by me) of free will, that there must be no influence, no possible predictability of someone's actions for there to exist free will. Plus, isn't it fun to define new concepts? But seriously, random doesn't overrule the possibility of cause, just the predictability of its effects. At least according to Webster.

It is fun, for me at least, to construct new concepts. For them to be new, they have to be differentiated from existing concepts in a meaningful way. If something is just a different name for the same thing, it makes communication worse, not better, no?. For this reason, I assumed that this page had a useful definition in mind. You (and perhaps others) still seem to be confused about the atomic decay example. The rate of atomic decay is extremely well predicted. Nobody has claimed *that* was random. What *is* random is the time at which a particular decay event occurs. It doesn't matter *what* you do to the atom, this does not become more predictable. Your last comment captures part of the problem of this page: predictability is key to randomness. Random events are by definition unpredictable, and this 'true random' does not offer any useful differentiation from that idea.

Pseudorandom is another definition that people use to reference things that pretend to be random. Those things are random to people not in the know, but aren't considered random by those in the know.

The atom decay example in the last few paragraphs was only meant to give a general idea of the kind of thing we're talking about, not the actual facts. I don't believe that predictability is key to randomness. Predictability is purely perception. I wanted something based on less perception, which could be true or perfect random. If regular random is enough, than dice are enough, I don't think you currently have enough technology to predict how a person is going to throw the dice, how the air is going to affect the dice once in the air, and how the table surface will affect the impact of the dice. But dice really aren't truerandom, we know they are affected by this that or another, and if we could measure that stuff, we know we can predict it. We know that the movement of a person's arm isn't random, its dictated by the person's feelings, thoughts, limited by the muscles and the training they have gone through. Lets make it simpler. Put the dice in a cup operated by a finely tuned machine. We can get it to toss the dice the same way every time. We can put the dice in the cup in a specific way, and we can put it in a vacuum that is affected by a consistent gravity and we can have the machine toss the dice. All we would have to do is record the first rolls, put everything back the way it was and do it again to get the same rolls. Would that satisfactorily nullify its randomness to you? I'm looking for the dice that roll differently after setting everything up again. Impossible unless the dots or numbers 'magically' move. Find me those dice and you've found me TrueRandom. I vaguely recall a quote about the dice being loaded...

No. This is the point that keeps getting missed here. Dice are *not* random. They are a fairly good pseudo-random source. Predictability is *not* perception. Imprecise use of language may lead you to that conclusion, but it is simply incorrect usage. The idea of a random source is a useful theoretical tool that is well defined. Physics tells us that to the best of current knowledge, such things exist physically as well. Nothing you have said on this page differentiates 'true random' from this, the true 'random' .

Physical dice, coins, guesses, whathaveyou are not random. It is that simple. There are two problems here: we could posit with enough computational power that we could predict your rolling of dice (but the argument could be made that this is not possible in practice). However, as best science can tell today, the atomic decay example is fundamentally different. You simply cannot know enough to predict it, even if you know the mechanisms by which it occurs. Now people may be sloppy when talking about random when they really mean 'random enough for this'. That doesn't change the fact that a random source is not even in theory predictable, let alone in practice. Got it?

I hold that dice ARE actually random - TO SOME DEGREE. It's just a matter of magnitude. The most perfectly controlled reproducible die-rolling machine would still not be able to roll the die the same way each time, because at each subsequent roll the universe will have changed, gravitational and other forces will be different, etc. And no power in the universe can comprehend all of the variables (someone else made the point that it would take a computational system at least twice as large as the universe to comprehend every variable in the universe). So it is not possible to construct the ideal non-random dice-rolling machine. It MAY be possible to construct a machine which is very closely reproducible, so that the changing external variables affect it minimally, or even immeasurably, but the fact is, it will never be PERFECTLY reproducible. And this is on a macro scale. If one considers that there is a similar degree of unpredictability on the micro scale, down to a quantum level, then there is "even more randomness" in the picture. So I think all events are to some degree deterministic, with a "fuzzy border" of randomness around them; sometimes the border is sharp enough for us to make accurate-enough predictions or uses of a phenomenon or machine, but there is always that margin of error, of unpredictability. And whether this unpredictability is a product of idealized true randomness, or of a pseudo-randomness that is beyond computational power to predict, is arguably an unanswerable question, and may remain one of the universe's mysteries. -- gk

Nope. Haven't got it. Imprecise use of language? On Earth, language is imprecise at best, ignore the usage. A more precise communication would most likely be telepathy, a direct transference of all the feelings and thoughts wanting to be conveyed. There seems to be a belief on this page that scientists determine the exact meanings of words. I'm sad to say, the entire populace comes up with the definitions ... and they don't do it by sitting down and agreeing, they do it through usage. Thus, I think it would be fair to say that my concept of regular random as defined as an adjective describing a process or outcome that is not predictable (which I had looked up somewhere, I can't say its Webster) would say that dice are actually random, assuming that I take the dice in my hand and toss them to the floor right now. Now, when you say not even in theory, help me out here ... is there not a theory or law of cause and effect? That every action has a reaction and that every reaction has an action?

Language is imprecise, yes. That is why we have to work at being precise in communication when precision is needed. Normal day to day conversation is almost never precise, but puts limits on what you can achieve conceptually for that reason. Sometimes you just have to bite the bullet a be rigorous :) A random source is a (theoretical) source with outcome that is due purely to chance, it is not predictable in any sense. This is very useful concept, for many reasons. Cause and effect has nothing to do with it, we are dealing with theoretical sources here. So there is your definition of random, that is not in any useful way different from 'true random' of this page. Perhaps it would help you to use the more technically oriented synonym: stochastic. Normal day to day 'random' things like coin tosses are clearly not really random. However, in the usual imprecision of day-to-day language, we often say they are when really me mean 'random enough'. Ok, so now you have a definition, which seems to supplant any need for 'true random' (whatever that is supposed to mean - it really would help to have a precise definition here). Further more, to the best of our knowledge today (of course this could change, but that is not relevant), physical sources exist such as the decay events in atomic decay. If your argument is that common usage of 'random' doesn't capture the precise meaning you wanted, well that is usually the case. Common usage is always sloppy, which is why there are specific usages (and language for that matter) by and for mathematicians, for example, that are not sloppy. If you want communicate about precise things, you need to use careful language...

[In your example, use a heavily loaded die. The result is highly predictable. Ordinary dice are just a less extreme case. They are predictable, but less so than the loaded die. BTW, 'regular random' wasn't defined previously on this page.]

Okay...I'd gladly rewrite the definition as soon as we come closer to agreeing. Cause and effect were fundamental components of the definition. TrueRandom is: any event or process with an outcome that has no cause, a reaction without an action. If that's already been defined, I apologize for the waste of your time. And, one more question. Dealing with your theoretical sources... if I'm clairvoyant and can tell you the precise time of decay every time, does that mean its no longer random?

[Are you saying that 'regular', as you used it before, was just a synonym for 'true'. If so, can't we just stick with one term? (I'll duck the clairvoyant trap for the time-being.)]

I've used heavily loaded dice before, you'd be surprised how many times I didn't roll an 18 for my character stats. :) Meanwhile, just because you can guess correctly 99% of the time doesn't make it predictable. In school, in math class, if you got the right answer, it didn't matter that it was right, what was important was getting it the right way. If the dice roll wrong just one time, its proof that you don't have the correct prediction model and we can consider them to be random. Yeah... figured RandomNumbers had enough of a definition, but it doesn't.

[Unpredictability can be related to randomness - that's standard theory.]

[Surely it's clear that dice rolls are not truerandom - that's not disputed, and needn't be justified here at length.]

[Person X performs certain actions at random times, and an hour later person Y copies those actions. I can't predict the actions of person X, but having seen them, I can predict the actions of person Y. The actions of person Y therefore pass all statistical tests and tests of predictability except those using prediction based on knowledge of the previous actions of person X.] I don't understand what relevance this statement has, can you expand?

[Just guarding against mistakes which might arise.]


It's not about physical or productive utility so much as the pursuit of all knowledge and the utility of using it in a philosophical argument. There's utility in almost anything, just got to look hard enough.

There is no apparent philosophical utility in this concept as written, but it is too vaguely specified to be certain about that. I think it is obvious that there was never an attempt at physical utility.

The original definition designer stands on the following grounds... overall culture is more important, The definition is for overall culture and general populace discussions and not scientific debate.


Part 2 of TheArtOfComputerProgramming discusses the question of what are truly random sequences of numbers. Various possible definitions are given. Very interesting from a philosophical point of view.


See also: RandomNumber UnitTestingRandomness WithoutCause Contrast: PseudoRandomNumberGenerator


EditText of this page (last edited December 26, 2005) or FindPage with title or text search