Main Idea:

The main idea of this book is not only that diversity or difference in skill sets, knowledge, attitudes, and other human qualities is important and beneficial in achieving objectives, but that it is more important then individual ability. In addition to this interesting idea author also brings in commonly used political meaning of diversity as diversity of skin color, religious believes, and ethnic background, arguing that all these differences provide for different approaches and therefore improve performance.

Details:

Prologue: How Diversity Trumps Ability
This statement based on computer based experiments when two groups of people were selected to solve some problems. One group was selected randomly and therefore diverse while another included best individuals performers. The counterintuitive result was that diverse groups were consistently superior in problem solving than groups of superior individuals. Additional reference demonstrating the same idea is provided from “Wisdom of crowds” and other literature.
Introduction: Unpacking Our Differences
The two examples of successful use of diversity are provided: Eli Lilly web site seeking solutions for posted problems; and WWII decoding success of Bletchley Park by collection of diverse individuals. From these examples comes the Diversity Conjuncture: DIVERSITY LEADS TO BETTER OUTCOMES.
The logic of book implemented in 5 parts:
1. Review of five different types of diversity
2. The Benefits of Diverse tools
3. Diverse values
4. Formulation of diversity benefits claims:
a. Diverse perspectives and tools enable collections of people to find more and better solutions
b. Diverse predictive models enable crowds to make better predictions
c. Diverse fundamental preferences frustrate the process of making choices
5. Application in real world.
This introduction also puts a very unusual twist on hot issue of affirmative actions: they make sense as long as different identities correlate with cognitive diversity. This correlation is taken as given, but it is not necessarily so.

PART ONE UNPACKING THE TOOLBOX
1. Diverse Perspectives How We See Things
The perspective framework is provided: A perspective is a map from reality to an internal language such that each distinct object, situation, problem, or event gets mapped to a unique word.
Examples: perspective of mathematician vs. chemist; various perspectives on quality of ice cream. Very interesting presentation of the same simple game Tic Tac Toe from different perspectives making it unrecognizable.

2. Heuristics: Do the Opposite
Definition: A heuristic is a rule applied to an existing solution represented in a perspective that generates a new solution or a set of possible solutions.
Example “Do the Opposite of what you initially wanted” provided from Seinfeld episode as “successful” heuristic for George Costanza.
Heuristics reviewed:
• Traveling salesman heuristic
• Topological heuristics
• Gradient heuristics
• Error allowing heuristics
• Population heuristics

3. Interpretations: Our Own Private Flatland
An Interpretation is a map from objects, situations, problems, and events to words.
Projection interpretation ignores some dimensions of a perspective.
A clumping interpretation creates categories of similar objects that are not simply projections of attributes.

4. Predictive Models: Judging Books by Their Covers
A predictive model is an interpretation together with prediction for each set or category created by the interpretation. Simple example of predictive matrix is provided.

5. Measuring Sticks and Toolboxes: Calibers for the Brain
This chapter adds measurement to previous discussion of tools: Perspectives, Heuristics, Interpretations, and Predictive models. The measurements discussed based of multiple IQ measurements and their use.
PART TWO: DIVERSITY’S BENEFITS: BUILDING FROM TOOLS

6. Diversity and Problem Solving: Darwin’s Brass Tacks
The rule of diversity trumping individual superiority for problem solving restated and explained as following: the best individual perfumers tend to be similar and therefore are looking for solution in the same place achieving local optimization. Diverse problem solvers are looking for solution all over the place including areas of absolute optimum, which often is outside scope of best performers. The rule is not absolute. It requires fulfillment of four conditions:
a. The problem is difficult. No individual problem solver always locates the global optimum
b. All problem solvers are smart enough to recognize solution
c. Any solution other then the global optimum is not a local optimum for some nonzero number of problem solvers
d. The initial population of problem solvers must be large
A very interesting idea comes from Darwin: selection reduces diversity. In other words by selecting the best and brightest to solve problems we actually cutting off all possible solution that could come from not the best and brightest.

7. Models of Information Aggregation: Mindless Signals
This chapter about wisdom of crowds and conditions when it happens: different individuals have knowledge of different parts of solution while various errors distributed randomly and therefore cancel each other.

8. Diversity and Prediction: The Crowd of Models
For predictions the diversity matters as much as ability. Not less and not more. From here comes the Diversity Prediction Theorem:
Collective Error=Average Individual Error – Prediction Diversity
The potential downside: madness of the crowds. People in the crowd conform to prevailing views and suppress non-conformist opinions. The result is dramatic decrease in diversity or even its change to negative value in formula.

PART THREE DIVERSE VALUES: A CONFLICT OF INTERESTS (OR IS IT)?
9. Diverse Preferences: Why Tapas
Notions defined of Fundamental Preferences (about outcomes) and Instrumental Preferences (how to get there). Preferences are different from choices. They are ordering of alternatives rather then selection of alternatives. Example fundamental preference: reduce crime first and care about better housing second. Example of instrumental preferences: try to reduce crime by increased policing first and by education second.

10. Preference Aggregations: Four (Not So) Depressing Results
Here is an important mathematical results going back to Kenneth Arrow:
• Collective preferences may fail to exists
• Unconstrained voting process may result in arbitrary choices
• People may have incentives to manipulate the choice process
• Common resources (public goods) may be underprovided

I personally love Arrow’s theorem so much that I have to include it: “No complete, transitive collective preference ordering based on individual preference ordering exists that satisfies unanimity, independence of irrelevant alternatives, and nondictatorship if all possible preferences are allowed.
Also provided other mathematical analytics confirming the main result: no collective objective could be defined without manipulation and/or dictatorship or at least suppression of minority objectives.

11. Interacting Toolboxes and Preferences: Go Ask Alice
There is complex interaction: Diverse fundamental preferences produce diverse values and different sets of possible solutions. The effects of interaction: diversity begets diversity; diverse predictive models create diverse instrumental preferences.
PART FOUR THE PUDDING: DOES DIVERSITY GENERATE BENEFITS?

12. The Causes of Cognitive Diversity: Family Vacations, College, or identity?
Summary so far:
• Cognitive diversity produces benefits
• Fundamental preference diversity creates problems;
• Collections of people with diverse cognitive toolboxes and diverse preferences have higher-variance performance.
Causes of diversity: DNA, Training and Experience, Identity, finally and very important – Serendipity.

13. The Empirical Evidence: The Pudding
This chapter is about proper collection and analysis of data. Example provided of hypothetical experiment to define if bike is faster transportation then walking. The result really depends on user having skill to use bicycle.

Here are the areas were diversity hypothesis is supported by facts: Predictions and Problem Solving. The confirmed Formula: Net Benefits = Gross Benefits of Diversity – Cost of Diversity.
The overall empiric results:
Cognitive diversity always improve collective performance
Identity diversity performance impact is not clear. Sometimes cost of identity diversity is too high when groups fight each other.

PART FIVE GOING ON THE OFFENSIVE
14. A Fertile Logic: Putting Ideas to Work
This chapter contains more or less detailed recommendations for effective use of diversity:
• Move beyond stock portfolio analogy in building diverse groups by taking into account superadditivity of diverse tools
• Contain multitudes (Whitman) by not being afraid of contradictions
• Bring in and listen to dissenters
• In politics encourage diverse citizens
• Encourage interdisciplinary efforts
• When building groups distinguish the Fundamental from the Instrumental
There is also a sub-chapter about use of diversity in hiring and admissions.

Epilogue: The Ketchup Questions
This epilog is pretty much call to invest necessary effort not only into tolerance of diversity, but go farther and embrace diversity because it would help to solve problems and improve lives.

My Take on It:

The term “diversity” is pretty much spoiled for me by its use to justify racial preferences, or support for illegal immigration, or refusal to reject intolerance, or even expression of sympathy to historical grievance or religion motivated terrorism.

However I agree with thesis of this book and I think that application of recommendations provided in it would be very beneficial for everybody’s wellbeing and prosperity. The only thing that I believe did not get proper treatment is tolerance. I think that if society to survive diversity, it should be extremely tolerant to everything except of intolerance of actions. Somehow people get excited when somebody said some stupid thing about race, but find it wonderful to have official race discrimination in college admissions. I would much prefer opposite attitude: everybody ignoring with contempt stupid talks by individuals and get excited and seriously fight all forms of race, religious, and any other form of discrimination.

As to the problem of people not getting proper treatment in childhood and schools, I am all in support for additional training to close gaps that prevent such people from competing on equal basis. I am completely against of having different criteria for anybody. Somehow in sport nobody suggests that athletically not very proficient people participate in Olympics running shorter distances or getting some seconds shaved off their time. The real consequence of this is not advance of disadvantaged, but cultivation of resentment resulting in inevitable increase in cost of diversity to the point of negative consequences.