Friday, September 10, 2010

A colleague asks, "Can we really make predictions?"


"Knowledge of the future automatically alters our actions. These, in turn, alter the future. Are attempts to predict the future futile?"
I would not strictly say so, but you have to modify what you mean by "predict".

Before the Modern Era, people believed that the state of the universe was determined by fate (pre-Enlightenment) or strict deterministic mechanics (Enlightenment). For the most part, they did not believe in a two-way dynamic entanglement between the state of the universe and human intention and action, nor did they strongly believe in random events. So when people spoke about making predictions, they believed that if a prophet did make a prediction about some future state, the prophet would be able to foretell the exact state, if the prophet possessed true clairvoyance or spoke with authority from God. Colloquially, people still have that idea in mind when they think about the term prediction.

Since the advent of the Modern Era, we now have a better understanding of dynamic systems and "random" events and how to characterize them. Now we know that a prediction must be accompanied by a statement of probability or degree of belief about anticipated outcomes. We can still make statements about the state of the universe at some time in the future*, but we also have to understand (and potentially manage) the odds when we do.

Believing that, as a decision analyst, I'm not so concerned about making forecasts or predictions about how the universe will turn out before I make a decision. Rather, I am more concerned with how many resources I need to apply and where in order to reduce that likelihood that I will regret one course of action over another.

[*Or when we have a greater resolution of the facts. By that, I mean that we don't always want to know what will occur. Rather, we want to know what has occurred but the actual state of the event is for the time being hidden behind a curtain, so to speak. Think petroleum reserves, the latent demand for a product (ok, that's a kind of hybrid time/resolution problem), whether you have cancer or not, or whether a pot of gold or a goat sits behind Monte Hall's curtain.]

No comments: