- May 1st, 2013, 2:29 am
#133369
Greetings. I think there are a couple of aspects in the language that need to be defined before we can get too much further. In his original post Allinone talked of quantum mechanics and it being random and chaotic. These are not the same thing. Quantum phenomena are random, but not chaotic. The orbits of three massive objects subject to mutual gravitational attraction on the other hand is chaotic, but not random.
A chaotic system is one characterized by extreme sensitivity to initial conditions. Such systems often exhibit a unique forward limit set which is referred to as a chaotic attractor, which is most often a fractal image in N dimensional phase space.
A random event by contrast is one which is fundamentally described by a probability distribution, that is an event which is not deterministic.
In short, chaotic systems as classically defined are not random but deterministic, and by definition no truly random event can be chaotic since true randomness cannot be tied to initial conditions.
A roll of the die may be taken as random, though given enough information its outcome may be deterministic. We may accept that the result of the toss of a fair die can be taken a priori as probabilistic since the conditions of the air currents applied during fall are sufficiently complex to make all outcomes equally likely for the throw of any given die for which the path is not deterministicaly worked out ahead of time. I bring this point up only because I will continue to use dice in my post though we could substitute some truly random phenomena in its place, say the decay of a cesium atom.
As Allinone describes above the outcome of any roll of the die is random, but the aggregate results of 6 million tosses is not. This is a result of the law of large numbers, however, and does not display an interconnectedness between the events. The argument to the contrary is often referred to as "The Gambler's Fallacy" which results from assuming that results of one toss have some effect on the results of a future toss, this is not the case. In another post Allinone discussed predicting where the 6's were tossed if we knew where all the 1-5 were tossed. This is of course true because if we know where all of the 1-5 tosses are, the only other thing the remaining players could have tossed were 6s! But I think he was aiming for something a bit deeper.
It has been shown that our minds often fall into a trap of "observed probabilities" which do not bear out in reality. This comes from the post hoc view of events. If a person rolls a die he knows to be fair many times and does not get a 6 he begins to rate the likelihood of a 6 on the next roll higher, he begins to feel that a 6 is "overdue" and thus more likely. However, even if you have not rolled a 6 on 20 consecutive rolls, the chance of a 6 on the 21st roll is 1/6, exactly the same as on the first roll. The fallacy comes from our post hoc view of events. It is highly unlikely that one will not roll a 6 for 20 rolls (only 2.6%) and even less likely that one will not roll it for 21 rolls (2.1%). The mind knows this intuitively and thus begins to expect that a 6 is due, but the chances have not changed. It is only in retrospect that we see that the string of rolls is unlikely, but this is true of any string of rolls.
If you roll a die 20 times, the odds that you will come up with the result that you do is 4.56-15 %, but this does not demonstrate that you have beaten the odds in any way, it is only in hindsight that the result can be seen to be unlikely. That result was exactly as likely as all the others and thus no great stroke of luck. Had you written down that exact list of numbers, in order, before rolling the die, that would be a different matter. The probability of a string of random events cannot be thus determined in this post hoc fashion. No reasonable person would call the string of 20 random numbers rolled on a die a miracle because the fallacy is obvious. However, our minds are often tricked into exactly that fallacy when it is much subtler, as when a gambler begins to think his lucky roll is "due."
I was going to put another tidbit about probability and the mind here but I think I've made the point I wished to and don't want to travel too far down the rabbit hole.
The point is: the distribution of the decay of atoms in radioactive material does not display an interaction between the atoms, but rather is the result of there being so many of them. At any given moment there is a tiny statistical chance that a specific given atom will decay, however, since there is a huge number of atoms that all have a small chance to decay, the number of atoms in the substance decaying is a near certainty. Just like tossing trillions of dice. The odds that a specific die comes up 6 is exactly 1/6 regardless of what any other die comes up, if no other die in the universe came up 6 it would not increase the odds that your die comes up 6 by one iota. If your intuition suggests otherwise it is the well documented Gambler's Fallacy rearing up. However the odds that no die comes up 6 out of trillions thrown is astronomically high. Thus by the law of large numbers the overall count for 6s is a near certainty (within of course a margin of error), this is due, not to any interaction between the dice, but rather to the properties intrinsic to each individual die.
Perhaps one last point will clarify the matter. You speak of the fact that as we toss more and more dice the distribution becomes clearer and clearer. This is indeed true, with one very important caveat. Say we toss 600 dice and 6 trillion. We should expect that the 6 trillion dice conform much more closely to the distribution of 16.67% likelihood for the roll of a 6, however, we should also expect that number of 6s by which we are off by is much higher than that for the 600 dice.
That is: we may have rolled 88 die with a 6 out of 600 total tosses which gives us 14.67% 6s out of our 600. Rolling even as close as 16.65% for the 6 trillion rolls, however, gives us 990,000,000,000 die with a 6, thus we are off of the perfect value by 10 billion rolls! Thus as we roll ever greater numbers of dice we get closer and closer to the exact probability due to the large numbers involved; but only when we consider the number of dice as a percentage of the total rolled. We do not actually get closer to a perfect distribution in terms of the number of the individual die tosses, in actuality, in nominal terms, we get further away.
Thus the statistical convergence has everything to do with there being many many events which are taking place so that the results of any particular event influence the total outcome very little and this does not constitute any interaction between the events themselves.