In a classic episode of Rod Serling’s Twilight Zone, “Nightmare at 20,000 feet,” Bob Wilson (played by a very young William Shatner) is on his first flight since his nervous breakdown six months earlier. At 20,000 feet, he repeatedly sees a creature on the wing, but whenever he pointed the creature out to someone else (his wife, the flight crew, fellow passengers), the creature would jump out of sight. When the creature starts to tamper with the plane’s wing and wiring, he grows increasingly concerned for the plane’s safety. What then should he do? If he ignores the creature, the plane might crash. If he continues to go on about a creature no one else has seen, he might likely end up back in the sanitarium. What would you do?
Well, you may recall if you’ve seen the episode, he steals a sleeping policeman’s handgun (it truly was a different time back then) and proceeds to open the exit door and successfully shoot the creature. Since no one else had seen the creature, the end result was that Bob was taken off the plane in a straightjacket, back to the sanitarium, since for sure he was having yet another breakdown.
Spoiler alert: in case you haven’t seen the classic episode and plan to as a result of this narrative, the twist at the end is about to be revealed.
At the end of the episode, as Bob is being carted off to the sanitarium in a straightjacket, the camera zooms in on the plane only to reveal the damage done to the wing and wiring by the creature. Bob wasn’t having a breakdown after all and, in fact, he may have just saved everyone on the plane.
That was Rod Serling fiction; the following account actually happened.
Recently, a very bright, albeit rambunctious, 8-year-old girl was traveling with her parents on a flight from London to New York. They were sitting two rows ahead of me and the girl was running up and down the aisle, reaching over seats to play with other people’s computers and generally creating havoc throughout the cabin. Her mother tried everything to control her – but to no avail. The father sat detached and oblivious to everything that was going on around him; indeed, his head was buried in an academic journal. However, after about 25 minutes of this, he had enough. He grabbed the girl by the arm and sternly said, “If you don’t behave, I am going to put you out on the wing!”
So, let’s examine the decision at hand for this apparently very astute young girl. Clearly the Mother held no sway and so when the father became involved, the girl had two choices at hand: 1) ignore the father and keep on misbehaving or 2) listen, sit down, and be quiet. We can represent her choices as follows:
Of course, the girl had to judge the repercussions of her actions — what would the father do? If she sat down and behaved, her father would do nothing except go back to his reading, but she wouldn’t get to do what she wanted, namely wreak havoc on the rest of the passengers! On the other hand, if she continued, she would have to judge the probability that her father would follow through on his threat of putting her on the wing. She was a smart young girl — clearly, he wasn’t going to follow through and put her out on the wing. Any 8-year-old child would realize that this is simply not a credible threat. Consequently, she continued to misbehave (much to the dismay of the other passengers) and the father went back to his reading. Smart girl.
Backwards Induction and Tactical Moves
What the story illustrates is the power of backwards induction — logic forward and reason backwards. When we’re faced with a strategic choice, we can indeed use logic to play out all of the alternatives and work through logical outcomes and responses. This enables us to choose the most desirable outcomes. We can then choose the correct actions that lead to the most favorable outcomes and monitor the process so that we stay on the correct path. The first part requires the use of decision trees; the second part can be played out via a tool called “Bayesian updating.”
In the case of the little girl, she knew the outcome she wanted (continuing to misbehave). She quickly realized that she wouldn’t end up on the wing no matter what she did. But, what could the father have done differently to have the girl choose a different path? Indeed, putting her out on the wing wasn’t feasible, which is precisely why this threat wasn’t credible; if he had played out the girl’s decision tree, he would have quickly realized that she wouldn’t listen based on this threat and the only way to get her to sit down would have been to incentivize her — either positively or negatively. For example, he could have threatened to withhold her allowance for a week (and then follow through), purchased an interesting movie on the plane’s inflight entertainment system, or anything else that was (a) credible and (b) provided the proper incentive to move to the “right” path on the decision tree. However, clearly, he wasn’t playing the game from her perspective.
Chess or Checkers Anyone?
We can use this same set of principles to logic out competitive moves in business. They’re actually similar to the ones we use when playing chess or checkers. Extend this thinking to a contemplated price cut, an impending capacity decision (e.g., build a plant, renovate an existing one, or take one offline), a potential acquisition, or a whole host of strategic decisions you might be considering for your business. Now, think back to the little girl on the plane. Imagine drawing a decision tree for your business and follow these guidelines:
Process for applying this to your business (key “take-aways”):
- List the set of potential options available to your business for the strategic issue at hand. Be complete.
- Draw one branch of a tree for each potential strategic move; each branch will represent one available option.
- One at a time, list the set of options available for each of your competitors for each branch of the tree you just drew. Imagine you actually did what was listed on each branch in turn — what is the complete set of feasible responses by rivals? (Note that the tree is getting increasingly complex.)
- For each of the larger sets of branches, assess the outcomes for your firm — how good or bad would this be if the sequencing of moves (first yours and then theirs) were to actually happen?
- Pick the best branches for your firm. What are the common elements of a “good” outcome? Is there a common first step? Are some of the outcomes particularly bad for your firm? If so, can you limit the likelihood of these bad outcomes by not doing the first step that leads to that bad outcome. You can essentially take the bad outcomes “off the table” by choosing first, but this requires some work and forethought.
For any of you who have played a game of chess or checkers, you already apply this process to the game. You can choose a move to maximize the likelihood of an end result being most favorable when you first think ahead multiple moves (i.e., for a current set of strategic alternatives) and reason backwards. In fact, if you understand chess and/or checkers, applying decision trees to sequential games is something that should be quite intuitive.
short url:
