Notes on Think Twice by Michael Mauboussin

I recently enjoyed reading Think Twice: Harnessing the Power of Counterintuition by Michael Mauboussin. It made for a nice counterbalance to Michael Gladwell's Blink, which I had read previously and which makes some fairly different arguments on the power of intuition. The biggest lesson is that both have their place, and wisdom comes into play in distinguishing which is relevant when.

The book's introduction mentions some famous examples of really smart people getting duped by relying on their intuition. Stephen Greenspan wrote an entire book on gullibility; Alan Greenspan lost money in Madoff's scheme; the LTCM geniuses "failed;" the Columbia disaster occurred despite early warnings; the banks in Iceland collapsed. Smart people can make really big mistakes!

The author's three step approach to deal with this is as follows:

1. Prepare: learn about mistakes
2. Recognize: find mistakes in context
3. Apply: use various tools
   

The author also outlines three major areas of mistakes:

1. Overconfidence
2. Pure irrationality
3. Non-multidisciplinary thinking
   

Ch. 1: The Outside View

• Example of Big Brown race horse in the Triple Crown contest
• Example of overconfidence, frothy UPS sponsorship
• Horse ended up finishing last
• Need to check how successful were other horses in his position (with his win/loss record)
• Most people favor the inside view over the outside view (outside view doesn't see problem as unique and has better perspective).
• Least capable people often have the largest gaps of overconfidence.
• Illusion of optimism
• Example: beating the odds of M&A history
• Testimonials or stories more influential than data
• Planning fallacy: underestimate time and cost
• Look at others in same situation for more objective, outside-view analysis
• Select a reference class
• Assess the distribution of outcomes
• Make a prediction
• Assess the reliability of the prediction and fine tune
  

Ch. 2: Open to Options (How your Telephone Number Affects Your Decisions)

• Kahneman's research on anchoring, bias (even when anchor/priming source is irrelevant, like phone number)
• Tunnel vision
• Presentation of problem affects how we solve it.
• Anchoring affects negotiation
• Groupman (author of How Doctors Think): always be open to unpredictable
• Representativeness bias/availability bias
• Historic pattern recognition skill less relevant in modern world
• Confirmation bias (Cialdini's theory on the persuasiveness of consistency)
• Example: famous video of counting white shirts and missing something glaring
• Stress increases tunnel vision.
• Bad incentives
• To prevent, seek dissent.
• Keep track of decisions to avoid hindsight bias; write down in decision-making journal.
• Avoid decisions in emotional situations.
• Understand incentives.
  

Ch. 3: The Expert Squeeze (Why Netflix Knows More Than Clerks Do About Your Favorite Films)

• The Wisdom of Crowds
• Collective error is individual error minus collective diversity.
• 3 requirements for success of crowdsourcing estimates/predictions: diversity, aggregation, incentives
• Inappropriately relying on intuition
• Match problem with appropriate solution
• Seek diversity
• Use technology when possible
  

Ch. 4: Situational Effects

• People conform to group judgment.
• Group affects perception parts of brain, not decision making. Group affects how you see, not think!
• Fundamental attribution error
• Experiment: ambient music statistically affects French or German wine purchase preferences.
• Opt out organ donations: most people just go with defaults.
• Milgram studies on roles
• Zimbardo prison study on roles
• Inertia in old business practices
• To stop this, use checklists (e.g., for IV line insertions in hospitals).
• Be aware of situations.
• Consider situation before individual.
• Watch out for institutional imperative.
• Avoid inertia.
  

Ch. 5: More is Different (How Bees Find Hives Without Real Estate Agents)

• Swarm intelligence, no leader
• Aggregation of noisy individuals
• Complex adaptive system
• Interactions create structure.
• Individuals inept; whole smart
• Market irrationality does not follow from individual irrationality.
• Changes in one component affect whole (problems at Yellowstone Park)
• Watch for tightly coupled systems.
• Use simulations to build complex worlds.
  

Ch. 6: Evidence of Circumstance (Outsourcing the Dreamliner)

• Birth order affects personality, but context changes this.
• Outsourcing bad when coordination is difficult with many sub-components.
• Outsourcing design and engineering bad
• Correlation not causality: Superbowl stock market indicator, Bangladesh butter production stock market indicator
• For causality, x must come before y, x must have a functional relationship with y, and there must be no z that causes both x and y.
• Mistake: inflexibility in face of new evidence (Norse failure to adapt in Greenland)
  

Ch. 7: Grand Ah-Whooms (Making the Millennium Bridge Wobble)

• Arbitrage and negative feedback resist change and stabilize systems.
• Positive feedback creates trends.
• Small incremental change can instigate a big phase transition (liquid to ice, critical mass of new fad): called ah-whoom moments
• Synchronized crowd behavior swayed Millenium Bridge
• Nassim Taleb/Black Swans: extreme outliers in power law distributions
• Problem of induction
• Falsification better than verification
• Mandelbrot: normal distribution wrong for asset prices
• LTCM: changing correlations can wreak havoc
• Our prediction ability is very poor; big role of luck
• In social networking or online games, luck of initial draw/initial picks/users critical to determining future path
• Understand distribution, model, and prepare for gray swans.
• Prepare for ah-whoom moments.
• Beware of forecasters.
• Mitigate downside, capture upside
• Kelly formula: never bet too much when big losses are possible.
  

Ch. 8: Sorting Luck from Skill (Why Investors Buy High and Sell Low)

• Dalton: research on reversion to mean
• Children heights: revert to mean, stable distribution
• Feedback should focus on skill and process, not outcomes, because these revert to mean.
• Halo effect: only focus on winners' strategies, not losers'
• Focus on sample size to see if conclusion worthwhile.
  

Conclusion

• When stakes high, think twice.
• Raise your awareness.
• Put yourself in others' shoes, find a reference class.
• Keep a decision-making journal.
• Use checklists.
• Conduct premortems.
• Know what you can't know.