Caveat: Hindsight bias

Sees past events as being more predictable than they were.

How many similar issues arise at NASA for which nothing bad happens?

Small groups?

Benefits?

• can act quickly
• synergy and swing

Common limitations?

• more prone to direct influence from one another
• fewer resources

Confirmation bias

• Favors information that confirms their beliefs or hypotheses.
• Conclusion to evidence; rather than vice versa:
  • cameras can’t see damage to Columbia
  • damage couldn’t be fixed

“Twelve Angry Men”

What difference would…

Cognitive diversity

Stasser’s experiment showed people talk about what they know in common, rather than what was rare/valuable.

Dissent

Solomon Asch people erred in estimating the length of a line less the 1% the time when alone, but when influenced by group pressure they erred 36.8% of the time (Sunstein 2003, pp. 18-19)

Polarization

1. proportion of arguments/limited argument pool: a group whose members are already inclined in a certain direction will offer a disproportionately large number of arguments supporting that same direction, exaggerating initial inclinations (Sunstein 2001, sec. 4)
2. social comparison: people want to be favorably perceived by others, and so shift their views to what they perceive others’ to be (Sunstein 2001, sec. 4)

Expertise

• talkativeness != expertise
• experts often don’t know how wrong they are

Even so…

Small groups can actually perform better than their best members.

Exercise

You’ve been appointed to a small committee intended to look into a problem.

What six ground rules would you recommend?

Gangster models

• *Godfather II: top-down hierarchy;
  • plus: quick decisions; long-term planning
  • minus: hard to get information; hard to scale
• Heat: small and tight-nit group
  • plus: trust, specialization, mutual awareness
  • minus: limited resources and room for error;
• Asphalt Jungle: specialized, ad hoc job
  • plus: hand-picked diversity; well incented
  • minus: high TC, low trust

Coase and transaction costs

Firms emerge when external TC > internal TC (e.g., search, contracting, and coordinating)

How might feedback relate to Coase?

• You are more likely to get accurate feedback at the boundary.

Paradox of inclusion

• at GM it took 15 meetings (with the CEO in on the last 5) to change the design of a headlamp.
• endless discussion versus distributed decision making

Problems of hierarchy

• worker had to idea of consequences of mistake, or even what her bracket did
• Peters and Waterman noted a new product idea had to traverse 223 steps
• upward mobility correlated with not telling boss bad news
• little diversity and executive isolation

Why couldn’t the customers take it right to the software people?

targets → lowballing & deceit

“Goldbricking in the machine shop.”

When a measure becomes a target, it ceases to be a good measure. – Marilyn

decentralization → ownership & coordination