Now being published as a novel. Click the picture to find out more:

Monday, October 29, 2007

Did God play dice with the world?

Einstein famously mentioned that God did'nt play dice with this world. This was in spite of the fact that his own findings (further worked upon by others) suggested that our universe, capable of bearing intelligent life, belonged to an infinitesimal minority of all the universes possible. More recently, this was coined in the form of "Anthropic Principle", stated simply:

The universe must be such so as to admit the creation of us as observers within.

In my mind, such a principle overly simplifies things and undermines any deeper research. Consider some answers from the Anthopic principle:

Q: How can intelligent life exist in the Universe in spite of such heavy odds against it?
A: Heavy odds or not, the event has happened. We are here. That is the explanation.

Q: How could the condition on earth be ripe for our existence?
A: If the condition were'nt ripe, we would'nt be here to ask this question, now, would we?

This leads to the question whether the nature is truly random and we are here just as a statistical possibility. Put another way, if I throw coins for millions of years, will the results be equally split between heads and tails?

I have been intrigued by this question before and considered it to be just a matter of belief since I did'nt think it could be possible to prove one way or the other. But a program on National Geographic yesterday shook my thoughts again. The episode was about a Random Event Generator (REG) Experiment. Pioneered by Prof. Jahn in 1970s and furthered by Doctor Nelson at Princeton University, the experiment consists of REG black boxes (smaller than a cigerette box) containing a microchip which constantly throws a one or a naught randomly. 65 such REG units have been installed in 40 countries and have been throwing one/naught for over a decade now at extremely high speed. If nature is truly random then the results should be equally split.

And so is the case. However, not quite so. Every once in a while, when a global event happens, the results go out of whack. The first such event was in 1997, Princess Diana's funeral. The chart shot upwards recording a major shift in the results. Infact, since then, the REG has somehow sensed a whole lot of global events ranging from September 11 attacks to the Asian Tsunami. And behold for the spookiest detail - the charts sensed Sep 11 attacks four hours beforehand and the Tsunami 24 hours beforehand !!! An electronic oracle??

The Scientists are skeptic as well as flabbergasted. The project team states that the chances of such correlation between REG and global events is a million to 1 and ascribes the correlation to a "hypothesis" that human community shares a common sub-conscious mind and when they are all focussing on one thing, they tilt the odds of the nature. Phew !!!

I am not so much concerned about the power of mind, important as it might be. I am just happy that there is some hope that we are just not statistical freaks of nature but there could be more to us.



Anonymous said...

Scientist's skepticism is well warranted. However being a layman what would intrigue me is whether such correlation (be it million to 1 or billion to 1) which scientist claim is because of common sub conscious mind (boy - o - boy even scientists have become economists now everything is in hindsight with no ability to predict) works for both positive and negative news flows. What i mean is would the REG experiment be throwing more 1's if something good is expected to happen and more naughts if some negative news flow is around the corner. And if this truly is the case then this no longer is a hypothesis this is realty.

Am just wondering whether its possible to use this technique to predict the sensex (NASDAQ equivalent in India) and become a successful day trader.

Beta said...

Interesting point you made. As fas as I know, the spikes on the chart are directionless and just signify something extraordinary which is happening but not whether its good or bad.

I am not sure about sensex but it got me thinking about the collective focus of a billion people towards cricket. I would like to know if the twenty-20 cricket worldcup final was registered in the REG plots or not. I went back to check if they have REGs installed in India and bingo, yes and at three locations. So may be it is worth checking.

Mahogany said...

Actually, if you tossed a coin a million times, it is incredibly unlikely that you would get half a million heads. In fact, the odds of that happening are staggeringly small, since that is only one "toss-pattern" out of a massive number of other, equally likely "toss-patterns".
I think.
I better pull out my copy of Kreyszig.

Beta said...

Mahogany - The most likely scenario of an infinite number of tosses is 50% split and the probability of that happening is exactly 0.5 which is higher than any other possible scenario.Lets see for a case of two tosses:

P (1H)= 1- P(2T)-P(2H) = 1-0.25-0.25=0.5

Lets now see for a case of 4 tosses:

P(2H) = 1- P(4H) - P(4T) - P(1T) - P(1H) = 1-0.0625-0.0625-0.1875-0.1875 = 0.5

You can extrapolate the same over million times and the probability for equal split will remain 0.5 while the probability of all other likelihoods will keep on dropping.

Infact, the 65 REGs are tossing coins at the speed of 200 per second for the past decade which means that they have already tossed coins a staggering 2.9 trillion times. Hence, any spike away from the expected outcome is worth checking and even more so if it coincides with any global event.

Trishanku said...

Sidhuism: Statistics are like skirts....

Mahogany said...

Actually Beta. your approach breaks down after 4 tosses. I checked Kreyszig and confirmed. Basically what you are looking at is a binomial distribution. Formula is
P(x) = (n/x) * (p^x) * (q^(n-x))
P(x) is the probability of x heads in n tosses, q is the probability of a head, q is the probability of a tail.

For n = 4 and x = 2, you are right that P(x) = 0.5. But after that p(x) falls rapidly. The probability of exactly 500 heads in a 1000 tosses with a fair coin is ... 1.53E-148! With a million tosses the probability of half a million heads falls to essentially zero.

Check out this link

Beta said...
This comment has been removed by the author.
Beta said...

Mate - Thanks for pointing the flaw and I concur with you.

Going back to the topic of this post, while P(exact half heads) will keep going down, the bionomial distribution will become highly skewed at the center i.e. around having half-heads. See a probability distribution below:

No. toss P(half heads) ±20% ±10%
10 24.6% 89% 65%
50 11.23% 99% 88%
100 7.96% *100% 96%

Thus, you will not really be expecting exactly half-heads but you will get alarmed if the results are way out of the predicted bionomial distribution (which incidentally keeps getting highly skewed at the center, what with 2.9 trillion tosses).

Makes sense?

Mahogany said...

Yep, totally makes sense.
Phew, glad we sorted that one out :-)

shrenik said...


your blog's a pretty nice read. and i'm not just reciprocating.

blog on...

Beta said...

Haha - Think, I will make my next article a bit less chewy :-)

Shrenik - Thanks, mate.