Tuesday, May 28, 2013

Trial and error is not random; rather, it requires intelligence and planning

I find this Thomas Edison quote fascinating:

Negative results are just what I want. They’re just as valuable to me as positive results. I can never find the thing that does the job best until I find the ones that don’t.

What feels strange about these words is the equating the value of negative results (failures) to positive results (successes). Why would they be equally valuable? Couldn't you spend your whole life producing negative results? How depressing!

Not if you're like Edison. Because what's not stated in the quote is that, to pull this off, Edison had to work in a certain frame that bounded his search for an answer. Within a bounded frame, each negative result is valuable, because it indicates an area of the frame that is not what you're looking for, making it more likely that the next test, within that boundary, is successful. And so on.

Nassim Nicholas Taleb crystallized this in his book "Antifragile," when he described how to approach trial and error "rationally."

Trial and error has one overriding value people fail to understand: it is not really random, rather, thanks to optionality, it requires some rationality. One needs to be intelligent in recognizing the favorable outcome and knowing what to discard.

And one needs to be rational in not making trial and error completely random. If you are looking for your misplaced wallet in your living room, in a trial and error mode, you exercise rationality by not looking in the same place twice. In many pursuits, every trial, every failure provides additional information, each more valuable than the previous one - if you know what does not work, or where the wallet is not located. With every trial one gets closer to something, assuming an environment in which one knows exactly what one is looking for. We can, from the trial that fails to deliver, figure out progressively where to go.

I can illustrate it best with the modus operandi of Greg Stemm, who specializes in pulling long-lost shipwrecks from the bottom of the sea.... He does an extensive analysis of the general area where the ship could be. That data is synthesized into a map drawn with squares of probability. A search area is then designed, taking into account that they must have certainty that the shipwreck is not in a specific area before moving on to a lower probability area. It looks random but is not. It is the equivalent of looking for a treasure in your house: every search has an incrementally a higher probability of yielding a result, but only if you can be certain that the area you have searched does not hold the treasure.

Stemm first maps the area where the shipwreck could have been, in the same way you focus on the living room to look for your lost wallet. That's the frame of the problem, and it takes intelligence and thought to figure out the appropriate frame (Edison would agree). Within the frame, trial and error becomes a powerful tool, in which each failure incrementally reduces the size of the frame, and gets you closer to your ultimate success.

Are "negative results just what you want"? If you have defined a proper frame for your search, the answer is yes.

[While I have been digesting "Antifragile," Albert Wenger has posted a fantastic and very thoughtful synopsis of the book and its ideas. Highly recommended.]

No comments:

Post a Comment