Decision-making is and has always been a passion of mine. During my Ph.D. research at MIT, I studied how to use computation techniques to make decisions for aerospace design problems based on relatively fuzzy projections. And while my research started out as an investigation into how one is able to make ideal decisions given large computation resources and adequate time for analysis, I quickly realized that there were equally interesting discoveries to be made around very quick, high-stakes decision-making, too. Although my learnings started out in the lab, I eventually discovered that they are also extremely relevant within both the business landscape and within everyday life, too.
One of the best books I have ever read on the subject of quick decision-making is “Sources of Power”, by Gary Klein. In his book, Klein describes specific situations where firefighters, police officers and military officers were able to make amazing and prescient decisions in mere moments based on very limited information. In each case, based on extensive analysis of the decisions ultimately made in those tense moments, Klein reveals out that human minds were able to synthesize the scarce tidbits of information available at the time to make very sound decisions that held up long after the heat of the moment had passed.
One of the key findings of my own research involved using computational methods to handle a situation where a person needs to make many interconnected decisions. The key, I argued, was to order the decisions properly. What decisions had to be made immediately? Which decisions could be delayed as long as possible to preserve optionality? The truths that I discovered in the lab back in the early 2000’s made their way into the technology that powers my company DataXu’s product today: a real-time decision engine for digital advertising that makes millions of AI-powered decisions every second. DataXu’s product functions as a system that not only helps our customers make millions of data-driven decisions in real-time on a daily basis, but also provides an interactive analytics interface to help senior marketers make more strategic decisions monthly, quarterly and even at yearly intervals.
Even as far removed from the lab as I am these days, decisioning still follows me everywhere I go. This week, in fact, I found myself coaching an employee as he was forced to make a decision very quickly that would normally require at least a week of analysis. He was visibly uncomfortable as we sat together and looked at his screen. “Now’s your chance to play CTO,” I said. “This is rapid decision-making in the extreme. You have limited information available, and 10 minutes to make a crucial decision. So let me ask you—what do you recommend we do?”
My team member took a few moments to think. He looked at the data available. He looked at me. He looked at the data one last time. And then he made a great recommendation.
For a moment, the situation transported me back in time—back to a conversation with my MIT Ph.D. advisor, Professor Ed Crawley. I remember sitting down, years ago, for our weekly research meeting. I had big news to share with him, and for once it had nothing to do with computational techniques. “Ed,” I said, “I’m excited to tell you that I’m getting married next summer.”
Ed, being Ed, smiled and then immediately found a way to work some wisdom into the conversation. “You know, Bill,” he said, “The thing about planning a wedding is that it will take up exactly as much time as exists before the day itself. If you are getting married in a year, planning will take a year. If you are getting married tomorrow, your planning will take one day. It’s funny that way. But either way, and most importantly—the outcome is still the same. You’re getting married!”
Professor Crawley was right, as usual. My wedding day was delayed due to family circumstances, and in the end it took a full two years to plan our wedding. As one might expect, our wedding was extremely well-planned and quite wonderful, but in the end, the same outcome occurred. I got married.
Over time, I have realized how important Ed’s piece of wisdom is when it comes to life’s unending stream of decision-making. One can stretch out his decision-making or planning for years using careful research and computation if given the time, but the human mind can also be amazingly great at making decisions quickly when time is short. I could computationally design an optimal route to work if I had the time, but when someone is drifting into my lane on the highway, I’ve got to make a split-second decision to react in time and save my life. The incredible thing is that the human brain is capable of doing an equally excellent job in both decision-making situations.
And in the end, although the wedding and all of its planning took a great deal of time (two years in fact) and many small decisions were made along the way, the outcome was the same. I married my beautiful bride—my wife. And I did not need any computational methods to confirm that that decision was the very best one of my life.