Predicting Randomness
- May 26th, 2013
- By Lettergram
- Write comment
The Back Story
About six months ago (December 2012) a friend and I were discussing ideas for beginning a business. Our methodology was simple, throw out as many ideas as we could, weed out the silly ones and then form a few models for the remaining ideas in hopes one seems like it has a fair possibility of success (hopefully with minimum risk). Turns out, although we had several interesting ideas, the most interesting part of the whole event was determining how to determine what would be best.
My friend had recently read a book titled: The Black Swan: The Impact of the Highly Improbable by Nassim Taleb and had been discussing it with me. In the book the author discusses the his ideology which explains that the most successful endeavors are the one that no one ever saw coming, although in retrospect most people believe it a certainty that that event would happen. In other words, looking at the past it seems likely that historical events do not seem to be random, but rather follow a logical track. Unfortunately, when in the present we cannot (even with the aid of computers) determine an event before it happens, even though the evidence certainly should exist. The author reasons that this is because as humans our brains simply don’t work like that, we attempt to find patterns and unfortunately, we simply don’t have a large enough sample size and perhaps never can.
These two discussions (about the book and starting a business) have been troubling me ever since. Clearly, there is an issue. Starting a company obviously is coupled with risk, however the idea that the most successful companies form from seemingly improbable ideas is troubling. Then what about the other side of the spectrum? What if I attempt to start a company that produces similar goods to another (better established) company with similar technology? Then I am out of luck. Leading to the conclusion that all highly successful companies must form from an idea which has currently not been predicted or created. That being the case, it would be beneficial to predict such cases would it not?
The Point
Not only would the ability to predict seemingly random events be beneficial in business, it would be beneficial in all aspects of life. Imagine the ability to predict traffic accidents, political upheavals, terrorist activities, economic downturns, inflation rates, perhaps even crimes.
The simple fact is, we cannot predict nearly as much as we could with the data we have at hand and although there are some serious challenges that would have to be over come I have no idea why we have no model to predict the seemingly “random” of history.
My Naive Solution
According to the author of the before mentioned book (The Black Swan: The Impact of the Highly Improbable) one major reason that we (as humans) have an issue predicting random events is our brains simply are not set up that way. Sure we can think, but clearly our evolution favored reflexes, strength, and adaptability over prediction what we needed to adapt to. An example would be, when tossed into the water we initially try to splash around stay afloat, we don’t (normally) consider how to swim before we are tossed in. The reasoning is simple. We have complex and fast computers which can do massive amounts of computations and have companies such as Google which can organize data efficiently. It should then be possible to rank current and semi-historical events similar to PageRank in order to predict likely outcomes. The trouble with this is that human history has trillions of important events and even more minor events which lead to outcomes. That being said, the issue according to the author (as well as inductive reasoning) points out that because most historians claim that their is a logical chain of events that lead to outcomes, history should lead to specific outcomes based on those facts. The problem with that chain of thought is that historians (as well as businessmen, economists, stock traders, etc) all look at ALL of history rather than each situation as a new one.
Therefore, if somehow we could determine a time-frame for which most seemingly “random” events or ideas occur we could therefore make accurate predictions about the short term future. For example, if in this randomness program there is a high likelihood that a company will develop a technology which can increase cell tower reception by 100% and it would be successful it would be a good idea to invest in that sector of the market. Similarly, if the program could determine a possible housing bubble fiasco it would be beneficial to fix take money out of the market. Obviously, there are many MANY clear issues, however the idea that I was attempting to convey was that although random events of today seem to be explosive and almost unpredictable there is a solution which on the surface seems possible.
And perhaps creating something which can predict the seemingly random, seems like a good business venture…



