Thursday, December 1, 2016

Kahn Academy on game theory

Chapter 12 of my book, "Maximizing Project Value", posits game theory as tool useful to project managers who are faced with trying to outwit or predict other parties vying for the same opportunity.

When John von Neumann first conceived game theory, he was out to solve zero-sum games in warfare: I win; you lose. But one of his students challenged him to "solve" a game which is not zero-sum. To wit: there can be a sub-optimum outcome that is more likely than an outright win or loss.

For this most part, this search for compromise or search for some outcome that is not a complete loss is throughout the business world, the public sector (except, perhaps, elective politics), and certainly is the situation in most project offices.

The classic explanation for game theory is the "prisoner's dilemma" in which two prisoners, both arrested for suspected participation in alleged crimes, are pitted against each other for confessions.

The decision space is set up with each "player" unable to communicate with the other. Thus, each player has his/her own self interest in mind, but also has some estimate of how the other player will react. The decision space then becomes something like this:
  1. If only you confess, you'll get a very light sentence for cooperating
  2. If you don't confess but the other guy does, and you're found guilty, you'll get a harsher sentence
  3. If both of you confess, then the sentence will be more harsh than if only you cooperated, but less harsh than if you didn't cooperate
  4. If neither of you confess, risking in effect the trust that the other guy will not sell you out, you and the other prisoner might both go with a fourth option: confess to a different but lesser crime with a certain light sentence.

From there, we learn about the Nash Equilibrium which posits that in such adversarial situations, the two parties often reach a stable but sub-optimum outcome.

In our situation with the prisoners, option 4 is optimum -- a guaranteed light sentence for both -- but it's not stable. As soon as you get wind of the other guy going for option 4, you can jump to option 1 and get the advantage of even a lighter sentence.

Option 3 is actually stable -- meaning there's no advantage to go to any other option -- but it's less optimum than the unstable option 4.

Now, you can port this to project management:
  • The prisoners are actually two project teams
  • The police are the customer
  • The crimes are different strategies that can be offered to the customer
  • The sentences are rewards (or penalties) from the customer

And so the lesson is that the customer will often wind up with a sub-optimum strategy because either a penalty or reward will attract one or the other project teams away from the optimum place to be. Bummer!

There are numerous YouTube videos on this, and books, and papers, etc. But an entertaining and version is at the Khan Academy, with Sal Khan doing his usual thing with a blackboard and and voice over.

And, you can read Chapter 12 of my book: "Maximizing Project Value" (the green/white cover below)

Read in the library at Square Peg Consulting about these books I've written
Buy them at any online book retailer!
Read my contribution to the Flashblog