Sunday, February 3, 2008

The Gaussian paradox (?)

As computer science students, something that we encounter very often is the gaussian curve a.k.a. the normal distribution. The bell shaped curve fits the behaviour of most large scale phenomenon, including human behaviour, among other things. It says that out of some observations of something (anything measurable), most will near the average value, while only some will differ by a large amount. Laymen know this as the herd mentality, if we talk about people. The more interesting part is, that there is a small percentage will be far away from the mean, equal percentages towards the left and right ends of the scale. The paradox that I talk about, comes into being when you try to apply this to human intelligence. Now, what intelligence means is slightly controversial, but i'll stick to the usually accepted paradigm. What I intend to look at is people who we call 'geniuses' and 'luminaries' as compared to others who are deemed 'psychotic' and 'anti-social'. This is where the gaussian curve gets interesting. The simple fact is, the curve doesn't differentiate between + and - around the mean, it only cares about the magnitude of this deviation. It is we who assign the + and the -. This takes an interesting connotation to this question of accepted intelligence and socially acceptable behaviour. It is society that assigns the tag of good or bad. Keep this in mind that the currently acceptable norms of good or bad won't remain the same over time. A few hundred years ago, black slavery was acceptable to most white people, before the world wars, even Hitler wasn't considered evil. Even nowadays, every goverment doesn't call all dictators evil, even the torchbearers of democracy tend not to criticize certain absolutist regimes, while reserving their contempt for others. Which brings us back to our gaussian curve. As I said, it simply measures the magnitude of deviation, not the nature of the deviation. Problem is, that deviation can be of many types, and some are more visible than others. Over time, we have assigned different worth to different deviations, and even at any time, the measure is different for different people. Killing people is accepted for soldiers, whilst not for other people. Not that I support people killing each other, but the fact is that we overlook certain things based on certain justifications. A generalist decision on all such factors is impossible, and so pointless.

What I'm essentially trying to imply is that look at differences with a fair eye, without the views society puts on you. You never know how society might change tomorrow.

4 comments:

Spec123 said...

Maybe a gaussian curve plotted on several non-independent variables will make more sense.

apollo said...

Of course it would. But try explaining that to someone who hasn't done courses on probability :D

Taking them one at a time makes it slightly simpler.

AP said...

"What I'm essentially trying to imply.."

Now now, where have I heard this before?

apollo said...

where?? :D