A Beautiful Grind V9

A Beautiful Grind Overcoming the Prisoner’s Dilemma and globally suboptimal Nash equilibria in the pursuit of enterprise alignment

Dr. Michael Ali [email protected]

Enterprise Alignment: Real World Stories • Divisions A, B, and C agree to standardize on a plant systems tool from Vendor X. Divisions A and C agree to do the first implementation. Division B decides to wait. Partway into implementation, Division A loses confidence in Vendor X. The enterprise council agrees not to standardize in the plant systems space. Division A reverts back to its in-house tool. Division C, further along in its implementation, cannot revert. Division C must now move forward with Vendor X without the leverage associated with an enterprise initiative. The sucker’s payoff. • Group Vice-President in charge of 3 divisions A, B, and C, wants a common ordering system for customers of all 3 division’s products. The cost per division is high due to the level of integration and change to business processes in addition to the software development costs. Division C does a cost/benefit analysis and declines to join the initiative. Group Vice-President says declining is not an option. Punishment for defecting greater than temptation to defect. • Divisions A and B must start joint product development work with Division C, recently purchased by the parent company. Division C should therefore adopt the processes and tools used by A and B. However, the processes and tools used by Division C are superior to those in A and B, and are responsible for the superior performance of Division C. The benefits of Division C doing joint development with A&B do not completely cover the impact of making Division C less efficient. Temptation to defect is greater than the reward of cooperating.

What do these stories have in common with the Prisoner’s Dilemma?

The classical Prisoner’s Dilemma Two suspects, you and your accomplice, are arrested by the police. The police have insufficient evidence for a conviction, and having separated the both of you, visit each of you and offer the same deal: if you confess and your accomplice remains silent, he gets the full 10-year sentence and you go free. If he confesses and you remain silent, you get the full 10-year sentence and he goes free. If you both stay silent, all they can do is give you both 6 months for a minor charge. If you both confess, you each get 6 years.

He denies

You deny

You confess

Both serve 6 months

He serves 10 years, you go free

R

He confesses

He goes free, you serve 10 years

T

Both serve 6 years S

P

The temptation (T) to confess (go free) is greater than the reward (R) for denying (6 months), which is better than the punishment (P) if you both confess (6 years), which is better than the sucker’s (S) payoff (10 years) : T > R > P > S

What will you do (strategy) and what will be the result (payoff) ?

The Prisoner’s Dilemma and the Nash Equilibrium DEFINITION: Nash Equilibrium If there is a set of strategies with the property that no player can benefit by changing his strategy while the other players keep their strategies unchanged, then that set of strategies and the corresponding payoffs constitute the Nash Equilibrium. From the movie, “A Beautiful Mind”

You deny

You confess

He denies

Both serve 6 months

He serves 10 years, you go free

He confesses

He goes free, you serve 10 years

Both serve 6 years

The Nash Equilibrium

The Prisoner’s Dilemma has one Nash equilibrium: when both players confess. The "both deny" strategy is unstable, as one prisoner could do better by confessing while their partner continues to deny. Alternatively, if one prisoner denies while the other confesses, it is better for the denier to change his plea (gets 6 years instead of 10). The Nash equilibrium strategy (both confess) is suboptimal for both because the payoff is 6 years vs. a best case 6 months. Both prisoners are making the rational choice, but, as Ian Stewart says, "sometimes rational decisions aren't sensible!" ttp://en.wikipedia.org/wiki/Nash_equilibrium

The Prisoners Dilemma and enterprise alignment

The Prisoners Dilemma is choosing between collaborating (denying) and defecting (confessing)

Enterprise alignment is the problem of choosing between local and enterprise solutions (products, processes, services) Divisions face the Prisoners Dilemma with enterprise alignment--should they: collaborate by automatically choosing the enterprise solution defect by choosing the optimal solution for the division (which may or may not be the enterprise solution)

The Enterprise Alignment Problem, pictorially

Local Solutions

Division B

Enterprise solutions (services, tools, processes)

Given a selection of local and enterprise solutions, the rational choice for the division is “which solution is best for me?” This approach does NOT imply that the local solution will always be the best choice for the division. Rather, every division will make the best choice for itself (i.e. optimize locally). As a result, a division will tend to contain a mix of local and enterprise solutions.

The Enterprise Alignment Problem, pictorially Local Solutions (A)

Division A

Local Solutions (B)

Division B

Local Solutions (C)

Division C

Enterprise solutions (services, tools, processes)

If there is more than one division in an enterprise, and every division makes the same rational choice to optimize itself, their selections will vary, so the effect is……

The Enterprise Alignment Problem, pictorially Local Solutions (A)

Division A

Local Solutions (B)

Division B

Local Solutions (C)

Division C

Enterprise solutions (services, tools, processes)

……to decrease the number of enterprise solutions, which leads to……..

The Enterprise Alignment Problem, pictorially Local Solutions (A)

Division A

Local Solutions (B)

Division B

Local Solutions (C)

Division C

Enterprise solutions (services, tools, processes)

……even fewer enterprise solutions, creating more fragmentation between divisions.

The Enterprise Alignment Problem, summary •

A Division will behave rationally, which is to select between the local and enterprise solutions based on what’s best for them, but Division B

•

multiple Brands, making the same rational choice, leads to fewer enterprise solutions available to any one Division and more fragmentation between Brands Brand A

Brand B

Brand C

Brand A

Enterprise solutions (services, tools, processes)

Enterprise solutions (services, tools, processes)

Brand B

Brand C

The Enterprise Alignment Problem, in game theoretic terms

PROBLEM: A division has to choose between always selecting the enterprise solution (collaborating) or making the optimal choice between the enterprise solution or a local solution (defecting). The choices have the following payoffs: T is the temptation to defect from the enterprise solution (preserve your right to choose), and R is the reward for always selecting the enterprise solution, and P is the punishment due to lack of enterprise solutions and fragmentation across the divisions, and S is the “sucker’s payoff”– i.e., your division is alone in using the “enterprise” solution IF: T > R > P > S THEN: The equilibrium strategy is for each division to locally optimize, which will have the effect of reducing the number of enterprise solutions available to all divisions and increasing fragmentation across the divisions. PROOF: The problem as stated is equivalent to the classical Prisoner’s Dilemma. Therefore the local optimization strategy (and corresponding suboptimal enterprise payoff) is the Nash equilibrium strategy for this game.

Enterprise Alignment and the Nash Equilibrium T is the temptation to defect from the enterprise solution (preserve your right to choose), and R is the reward for always selecting the enterprise solution, and P is the punishment due to lack of enterprise solutions and fragmentation across the divisions, and S is the “sucker’s payoff”– i.e., your division is alone in using the “enterprise” solution From the movie, “A Beautiful Mind”

Always pick the enterprise solution

Choose between local and enterprise solutions

Always pick the enterprise solution

Easier for divisions to work together, many enterprise solutions available, but your division may be suboptimized.

R

Your division gets to optimize selection based on division needs, the other division has to suboptimize in T some cases.

Choose between local and enterprise solutions

The other division gets to optimize selection based on division needs, you have to suboptimize in some cases. S

Everyone makes the optimal choice for themselves, leading to fewer enterprise solutions to choose from, and less ability to work across divisions P

The Nash Equilibrium (if T>R>P>S)

Solutions to the ‘Enterprise Alignment Dilemma’, Part I

T is the temptation to defect from the enterprise solution (preserve your right to choose), and R is the reward for always selecting the enterprise solution, and P is the punishment due to lack of enterprise solutions and fragmentation across the divisions, and S is the “sucker’s payoff”– i.e., your division is alone in using the “enterprise” solution

Since T > R > P > S creates the Prisoner’s Dilemma, don’t allow T > R > P > S!

Solutions to the ‘Enterprise Alignment Dilemma’, Part II 1.

Change the payoffs (make R & P > T) A. Increase the reward for choosing the enterprise solution (‘carrot’) i. lower the cost of enterprise solutions ii. raise the benefit of cross-division cooperation and defragmentation B. Reduce the temptation to locally optimize (‘stick’) • strong central governance • peer pressure

•

Agree that if R is within, say, 10% of T, then the enterprise solution will still be selected (make it easier for R = T)

•

If one division takes the enterprise solution that the others rejected, the other divisions have to pay compensation (i.e., reduce the sucker’s payoff, S, to zero)

•

Create regional solutions as an intermediate choice between local and enterprise solutions (reduce T, raise R)

•

If R > T translates into $ savings for some, use that money to offset T > R for the others

•

Communicate: let the other divisions know your choices (avoids creating a sucker, and enables open discussion of T, R, and P)

Conclusion Recognize that making the best choice for your division will inevitably lead to fewer enterprise options and more fragmentation. The rational choice (looking out for your division) is not the sensible choice (looking out for the enterprise). Always pick the enterprise solution

Choose between local and enterprise solutions

Always pick the enterprise solution

Easier for divisions to work together, many enterprise solutions available, but your division may be suboptimized.

Your division gets to optimize selection based on division needs, the other division has to suboptimize in some cases.

Choose between local and enterprise solutions

The other division gets to optimize selection based on division needs, you have to suboptimize in some cases.

Everyone makes the optimal choice for themselves, leading to fewer enterprise solutions to choose from, and less ability to work across divisions

Where we want to be (sensible)

The Nash Equilibrium (rational)