Monday, January 26, 2009

Don't Blink

Have you ever looked at a solution; be it an architecture, a network, a configuration, or code, and thought to yourself that you didn't like it but you couldn't necessarily express why? You're not alone, and if you've been doing your job for a while, over ten years, you may have reached a point where your subconscious mind sees things before your conscious mind and the subconscious is trying to tell you something.

A study conducted by the University of Iowa asked participants to randomly turn over cards from four different decks, two of the decks having red-backed card, the other two having blue. Based on which cards the participants selected, they would either win 'the game' or loose. What they weren't told is that the decks were stacked such that selecting red cards would cost them. Big.

The red decks also had some big winners, but over time, the red decks would kill you, while the blue cards offered a nice steady diet of modest wins and minimum losses. Again, the participants weren't told this at the outset, they had to figure it out. Most people did eventually figure out that the red decks were problematic, but only after turning over about eighty cards. Typically, the players began to get a hunch that the red decks were bad after about 50 cards, but that didn't translate into a reasoned explanation until 80 cards.

ESP, Intuition, or Subtle Analysis?
Here's what is interesting; as the test was proceeding the players were hooked up to all manner of physiological monitors for blood pressure, palm sweat, heart rate and so on. They would actually begin to manifest autonomic responses for the red (bad) decks after 10 cards have been turned over. Even though the players only began to get a conscious hunch after 50 cards, and a conscious explanation after 80, their bodies showed signs of stress after only ten. Subconsciously, they were already picking up the clues after 10 cards.

Blink is the title of a book by Malcolm Gladwell that deals with the subject of the subconscious mind and the clues it gives us that can be wonderfully insightful and occasionally disastrous. For this post, I want to focus on how the mind can reach conclusions, rapidly, with what you might think is very limited information.

In any organization decisions have to be made, and conventional wisdom would suggest that more information is better. Yet, our minds seem to be able to reach solid conclusions with very limited information. For instance, how many factors and time do you think it would take to be able to predict with a 70% accuracy whether a couple will remain married for 15 years. A study by the University of Washington has shown that a 3 minute composite of a fifteen minute videotaped conversations is all you need.

Wait a minute - did I just say that you can predict the outcome of a fifteen year relationship with a 3 minute video slice? Yes, because that three minute video composite has filtered out all of the noise so that all that is left is pertinent data. Expert art directors can spot a forgery in a second, even after (literally) years of evidence has been collected to authenticate a painting, statue, or composition. The experts only need a few bits of important data to know real from fake. Sometimes they don't know (consciously) themselves what those data points are.

In the book Blink, Gladwell recounts study after study that indicates that knowing a few key pieces of data is more valuable than knowing volumes of data. Gladwell cites examples of marriage researchers, war strategists, firefighters, policemen, and doctors that suffered through painful, even deadly scenarios, only to discover that their curse was not too little information, but too much.

Measures that matter
In our business we are often tasked with providing recommendations and justification to managers who are no longer intimate with the details of the newest technology. We can provide them with mountains of data, plateaus of information, and a diatribe of details. Our task is to first find out what the predictors are, and then express our recommendations in those terms.

Us I.T. folks love to point to subtle, subtle, subtle points of interest to prove that the Devil's in the details. For instance, did you know that the color of your car can affect fuel economy? OK, sure, but let's get real - how many other factors would you consider about your car's gas mileage before you would throw color into the mix? If you were asked to select a fuel efficient car as your next project, would you really recommend one particular color over another? I hope not, I hope you'd consider engine size, car capacity, requirements such as towing, or maybe - and this is a stretch; you'd consult the EPA numbers.

The next time you're asked to make a recommendation about (for instance) a new vendor solution, don't confuse clarity with detail. Focus on the elements that can predict the outcome, just like the researchers did by analyzing a three minute video to predict the outcome of a 15 year marriage. Eliminate the noise (the data that doesn't predict the outcome), and focus on the measures that matter.

No comments:

Post a Comment

Follow by Email