The Complexity Part:
I find that neuroscience (trying to understand the brain/mind/body interdependence in sentient life) and quantum physics (specifically particle entanglement and uncertainty) are perhaps the most confusing, complex, and barely comprehensible subjects I study. The more I read on those topics the more I realize that reality is in the eye of the beholder.
The other angle from reading in the neuroscience and quantum
physics space is that one quickly comes to see that everything in the known universe is interdependent. And while our universe is knowable at some level, it is obviously indifferent. That is to say, the universe is not at all concerned whether the earth and our pitiful species survives or not.
Hence, chaos theory implies that the future is not predictable based on past events, as it used to be thought to be. Or, in words that have been attributed to both physicist Niels Bohr and baseball manager Yogi Berra, “Prediction is very difficult, especially about the future.”
The Forecasting Part
And then there is forecasting. It too is complex. Trying to forecast the weather, the global economy, the adaptation of technology, or the U.S. Consumer is fraught with peril. Today, everything is interdependent. I intend this to be the second post in a four-part series on how growing complexity is changing our lives. For no particular reason, I’m addressing forecasting in this post.
As I think back through my career, I find that I (we) spent a great deal of time and energy trying to predict highly uncertain events. Occasionally we seem to be pretty good at the task. However, hindsight lets me also see that we were actually pretty lucky in those years when we “made our forecast” within +/- 10% of the single number. Weather forecasters are smarter than we were. They, at least, give you a percentage chance for precipitation. There will be a 20% chance of rain or an 80% chance of snow accumulation between 8 and 10 inches in the next 24 hours.
We weren’t smart enough to do that. Our forecasts should have given a range of probabilities as well. We might have given our leaders a better sense of what might happen if we said, “There is an 80% chance that our region will generate between $2.1 MM and $2.2 MM this fiscal year. Instead, like just about all of our competitors, we built models that were linear, based on history, best-guess future events and they yielded a single number. We predict we will generate $2.1 MM this fiscal year. There was no equivocation, no confidence index. Just a misleading single number.
Why Forecast Anyway?
I doubt we will ever give up trying to forecast the future. We want to know where we should put our efforts and resources. I still try to forecast macro trends like where technology might take us as a society. What should I do to prepare? What “jobs” will there be for me to earn a living? Will there be any human activity NOT affected by the advance of technology?
I hope we can get smarter about things though. I hope we can learn about and use probabilities. Even in political life, we need to understand that when a poll indicates that a candidate has, say, a 30% chance of winning an election — that means they absolutely can, and might, win the election. Likewise, when we forecast how a certain product might perform in the market place, we would do well to state and understand the probabilities.
Instead of stating that the product will achieve 4% of the available market in the first year, we would do well to state probabilities such as there is a 5% chance of gaining 1 to 2% market share, a 30% chance of gaining 2 to 3% market share, a 60% chance of gaining between 3 and 4% and 5% chance of gaining more than 4% market share.
Will Technology Help Us With Forecasting?
Of course, as a technologist, I believe technology will help us out. That’s my bias — I look to technology to solve the problems it creates. One of the reasons we fail at forecasting is that we have to make assumptions and simplifications because our world is so complex and interdependent. At this point, our existing computing capacity cannot model the “real world” economy. So we oversimplify, and we generally fail.
On the horizon, of course, is quantum computing. That technology seems well suited to building ever more realistic and complex models. We may actually have a shot at getting pretty good at predicting what a particular “change” to a law, tariff, tax, or application of Artificial Intelligence will do in our global economy. Predicted within a range of probabilities, I hope!
Meanwhile, I hope we learn to stop forecasting single numbers in the business world. I hope we learn to love Monte Carlo models.