Monday, October 4, 2010

The Predictioneer's Game

I just finished reading 'The Predictioneer's Game' by Bruce Bueno de Mesquita.  If you are interested in the future this is a must read.

Bruce is one part economist, one part social scientist, and one part mathematician.  He has developed a formula that describes the negotiation process.  He studies each party to a negotiation and assigns numerical values (0 to 100) to their positions, goals, and power.  Then he solves the formula and predicts the results of the negotiation.

His work is based upon game theory.  So he calls each negotiation a 'game'.  He gives examples of wars, negotiations leading up to wars, bidding wars, lawsuits, elections (and sham elections), multi-party political struggles, and car buying.  His explanation of how to get the best price for a new car is the most buyer-empowering process I have ever seen.  (Unfortunately it only applies to new cars.)

I find this approach very appealing.  He doesn't give away any of the formula itself.  He only explains the variables that he assigns as inputs.  Frankly, that's enough for me.  I don't want to do the math.  That's why we have computers.

He cites several examples of correct predictions.  The future predictions that he gives appear to be reasonable, and not obvious to anyone who does not approach the subject through this type of disciplined methodology.

He has posted an online version of his game.  So you can prepare a tab-delimited text file with a series of values representing each party of the negotiation, and then upload the file and see the results of the game.  He doesn't show you the formula.

Of course this is only a methodology, and not a guarantee.  He gives several examples when his predictions failed.  And it appears that he has a pretty good grasp of the failure conditions.  The biggest two failure conditions are outside interruptions (natural disasters, sex scandals, and other exceptional events that interfere with the negotiation process) and people who do not seek their own advantage (idiots and idealists.)  But these two issues are problems for game theory in general, and not unique to his methods.

There are a few modifications that I would like to see implemented in Bruce's formula and process .  The following won't make much sense if you haven't read the book.

Three Goals
His process only accounts for a single goal for each party.  I don't think that humans usually work that way.  I think that we usually have three different goals in mind:
* Ideal - The perfect state that we hope to achieve.
* Expected - The result that we expect to be able to achieve.
* Minimum - The poorest result that we are willing to accept.

The intervals between these goals affect the way that we play the game.  It's probably a fallacy to identify a 'normal' game.  But the most obvious case for these three goals would be that a person's Ideal goal is higher than their Expected goal, and both of those are higher than their Minimum goal.  In this situation a person is likely to negotiate and act in good faith.

If a person's Expected goal is very close to either their Ideal or Minimum goal, above their Ideal goal, or below their Minimum, then it will greatly alter their approach to the negotiation.

When a person's Expected goal is very close to their Ideal goal then they will shift into either a more flexible or a more rigid negotiating strategy.  Pragmatists will become more flexible in order to get the process over with quickly.  True believers, idealists, and other extremists will become more rigid as they get caught up in the fervor of reaching their Ideal goal.  (This might be the true test of an idealist.)

When a person's Expected goal is very close to their Minimum goal they will use a very rigid negotiating strategy and expend a great deal of time and energy on the negotiations.

When a person's Expected goal is lower than their Minimum goal they are going to either undermine the process, cheat, and otherwise act in poor faith, or they will expend a tremendous amount of time and energy in an incredibly flexible negotiating strategy.  For example, a dictator will not join a negotiation where they expect to be hanged as a result.  But a murderer who has been arrested will negotiate nearly anything to get out of the death penalty.

When a person's Expected goal is higher than their Ideal goal they are going to treat this negotiation as a means to an end to improve their negotiating position in some other game.  This is the ugly political process of giving votes on one issue to gain votes on a separate issue.

Every negotiation takes place within a context.  Each participant has other concerns that they are concurrently negotiating (or expect to begin negotiating soon.)  The participants have an existing relationship, and expectations for a future relationship.  And there are other parties who are not directly involved in a particular negotiation, but who have an interest in either a resolution to negotiations or to one or more of the participants.

Each participant has many ongoing negotiations.  If this negotiation is not one of their top priorities then it will not get their full attention.  They will probably delay it where possible.  When delay becomes impossible they will adopt a very flexible negotiating strategy in order to get done quickly.

The more negotiations that a person has going on, or expects to have going on soon, the more likely they are to be willing and able to be very flexible in their negotiation strategy.  This is especially true when some of the other participants to this negotiation are also participants in other negotiations.  This is the type of situation where politicians trade votes.

When two or more parties negotiate multiple separate games, over time, they establish a relationship.  Each participant will form an opinion of the other participants, especially concerning their trustworthiness.  Participants who have a long history of negotiating successful games will be able to negotiate more flexibly with each other because of the trusting relationship.  And conversely, when trust has been betrayed in the past  then current and future negotiations will be less flexible.

When one or more participant wants to negotiate more games after the current one, then that person will adopt a more flexible negotiating strategy.  For instance, a businessperson who is attempting to establish a reputation and earn repeat business will negotiate more deeply than an established company.

Sometimes there are external parties who may choose to interfere with a particular negotiation game.  There is a near-unlimited number of possibilities here.  Friends, enemies, and parties to other negotiations (current or future) are the prime suspects here.  A mistress, for instance, may be content to remain hidden until the money and notoriety from exposure gets too great.  Or someone may see an opportunity to hurt an old enemy by interfering with a negotiation.

I'm not certain that all of these scenarios are worth codifying within the prediction formula.  But I believe that for important negotiations each of these possibilities should be examined as a possible tactic for improving one's position.  That type of strategic and tactical research is what Bruce's consulting company does.

Humans aren't always rational.  Sometimes people just fail to see an opportunity.  Sometimes people have hidden constraints that limit them from exercising an obvious negotiation tactic.  Sometimes people bluff.  And sometimes your estimations of their goals and values are simply incorrect.

This is something that the predictive formula should be able to test and adapt for.  Almost all negotiations take place over a series of rounds.  As each round concludes you can compare the predicted and actual positions.

It is not as simple as updating the formula with the actual values and recomputing.  Each variation needs to be examined and explained.  Those explanations need to be fed into the strategic and tactical plans for future rounds of negotiation.

If you decide that someone is bluffing then you need to decide if you want to extract more concessions from them, at the risk of moving their Expected goal above their Minimum goal and causing them to begin subverting the negotiations.

If you decide that someone has a hidden constraint and cannot ask for a concession that they need, then you can use that information to your advantage.  Maybe you can threaten to expose their weakness to coerce greater concessions.  Or maybe you can offer the concession as an olive branch, in order to gain the upper hand.

If you decide that someone is simply stupid then you can play them in order to gain greater concessions.  You can arrange the situation to make the stupid person look heroic while they make much larger concessions than they should.  Call this the Neville Chamberlain tactic.

1 comment: