Thursday, November 15, 2018

managerial game theory

In a recent series of posts, The Business Bro detailed the various signals a manager can receive from management or employees regarding an organization’s imminent decline. He then broke down the different responses in meticulous detail to come up with the best option for each case.

Here is a summary of those signals and the recommended options:

If management correctly predicts imminent decline… focus on doing the job well if you are thinking long-term and play politics if you are thinking short-term

If management incorrectly predicts imminent decline… focus on doing the job well if the market is strong and play politics if your team is strong

If employees correctly predict imminent decline… focus on doing the job well if the market is strong and play politics if the market is weak

If employees incorrectly predict imminent decline… focus on doing the job well if you are thinking short-term and play politics if you are thinking long-term

The thought process employed by The BB brought me back to a subject I hadn’t thought very much about since college – game theory. It was a welcome return, however, because game theory was my favorite economics class.

The thing I initially liked the most about game theory was how its great complexity could be simplified with meticulous care and effort. In the BB’s recent example, a very complex scenario involving multiple factors is resolved by asking a series of simple questions. Structurally, these questions are basically yes-no – for example, are market conditions favorable for the organization? – or – is the team performing well? There are also other questions that break down into simple categories – is your primary motivation short or long term? Once all these questions are answered, it becomes easy enough to make a flowchart or decision tree that helps lead to the right conclusion.

By the time I finished the course, I found that the subtler aspects of game theory were becoming the most appealing to me. One especially important lesson was how knowing that the presence of one factor can render other considerations irrelevant. In everyday terms, what his means is that if you select option A for one factor, it might mean your decision between option C or D for some other factor has no effect on the final outcome.

This subtlety is very difficult to grasp because it challenges the natural human impulse to think of everything linearly. To put it another way, it doesn’t help that the natural tendency is to reframe a non-linear concept into a linear pattern just to make it a little easier to understand. A person who points out, for example, that the component articles of clothing in an outfit can’t just be rearranged to go with every other part of a wardrobe will always encounter someone who points out that this is the same thing as saying if a certain shirt is selected, it eliminates all the pants that don’t go with it. This is a good point if you don’t think about clothing in terms of outfits but it isn’t getting at the right way of thinking at all if the only way you conceive of getting dressed is by thinking about outfits.

Another metaphorical way to think about this is to consider ways to order a pizza. Let’s say you love pepperoni and always prefer this as your topping of choice. However, if you go to a vegetarian restaurant, this lifelong commitment won’t matter – though you still abstractly choose pepperoni (because you always do) the choice to go to a vegetarian pizza place took pepperoni off the list of realistic final pizza outcomes. So, you end up with the same dish as someone who always abstractly chooses fried cauliflower (your backup choice) (1).

The problem with the analogy is that the structure of the decision is still linear – you go to the restaurant first, then you make a decision about a topping. The linearity implied by the ‘if-then’ structure is an illusion that masks preferences because even if some decisions are rendered irrelevant by other decisions it doesn't mean that people don't maintain their preferences. To put it another way, game theory encourages looking at decisions in a simultaneous rather than linear way whenever there are multiple factors that go into a decision. This way of thinking helped me see that a lot of 'if-then' decision making was a narrative that explained very little about how people actually make decisions.

I’m describing all of this game theory here because I think the manager role is very conducive to someone who thinks like this. A manager who makes decisions in a linear way says first I do A over B, then I'll wait a little while before choosing C or D. This is OK in game theory but can lead a manager away from sound principles because a manager in this mode of thinking is always open to changing course given a new detail. This is fine if done sparingly but a manager who regularly breaks his or her word will quickly lose the trust of the team and find it difficult to complete long-term projects.

A manager who makes decisions as independently as possible will say I choose A over B and also C over D. Then, the manager would think about the implications of A and C interacting together to understand the most likely outcome. This style of thinking helps managers separate the relevant from the trivial. As I noted above, sometimes it doesn't matter what one factor is if another is present. A manager who chooses A and C might know that if E suddenly becomes F, it doesn’t change the plan in any relevant way because the presence of both A and C have set the course of action in stone irrespective of how the choice between E and F is made. A manager who chose C based on E, however, might reevaluate the plan when E becomes F because this manager will not recognize that the interaction of A and C render the change from E to F irrelevant.

If this sounds complex, well, it should be – complexity is a trait common to both game theory and the manager role. We're talking about stuff that you only learn at universities and jobs people do after decades as functional adults - if you thought it was supposed to be easy, I sadly inform you that you are mistaken. Like game theory, the manager role has been subject to gross oversimplification. People who think linearly think of the manager role as setting goals, keeping teams busy, and responding to new details with the same urgency as a fire chief responds to a 911 call. In the same way, people who think linearly conceive of game theory as a series of decision trees where A leads to B, or possibly C, in which case D.

These oversimplifications are accurate in the sense that they are true for very simple examples that anyone can understand on day one. However, the reality of game theory is a lot like managing – actions matter far less than interactions. A manager who understands that choosing A over B has implication on future decisions is on the right track - the type of thinking required is not A, then C, but rather knowing how A impacts the feasibility of choosing C while considering how the two choices together might influence the eventual choice of E or F.

Footnotes / cold pizza…

0. One last analogy…

A nice metaphor for managing is that it is like juggling many balls where only a few are made of glass. The complexity comes in knowing which balls are made of glass and will therefore break if they fall. This is a helpful analogy but one that encourages the linear thinking I feel often hampers managers. A better analogy might be that sometimes the balls change from glass to rubber (and vice-versa) but unfortunately this version suffers from its non-resemblance to how juggling (or physics) actually works.

1. Technically speaking (technically, in the game theory sense)...

This metaphor is consistent with how a pooled equilibrium works. In a pooled equilibrium, people with difference preferences send the same signal and an observer is therefore unable to work out the true preferences based solely on reading the signal. In this example, two people might order fired cauliflower but for at least one person the true preference was pepperoni.