Everything Is Obvious: How Common Sense Fails Us
Everything Is Obvious is sectioned into two parts, the first, Common Sense, deals with the recognition that commonsense is anything but, and explores various types of errors in commonsense reasoning. The second, Uncommon Sense, deals with attempting to make profit out of predicting mass social behavior.
The commonsense belief that commonsense is obvious falls apart when one takes a closer look. Often a commonsense truth and also its opposite may be true, only not necessarily at the same time. For example, in a poll of drivers 90% think that on average they are better than average. Statistically speaking, less than half can be better than average (at any one instant). The belief that one’s own commonsense is correct while others is not leads to the recognition that something is wrong with the concept of commonsense.
The social order we take for granted in everyday life is maintained in part by social rules that we aren’t aware of yet instinctively follow, and any attempt to express commonsense social order as a set of rules may be difficult. We often take social cues from others, for example facing front while riding in elevators, the physical separation between two people conversing face-to-face, and the duration of direct eye-contact made during conversation. Commonsense in total consists of practical experience accumulated over a lifetime of facts, observations, of received and perceived wisdom. Social commonsense varies by context, what we do or don’t say in front of a boss is different from a friend. Distinctions also vary from culture to culture and from socio-economic group to socio-economic group, for example societal norms for fairness and reciprocity. All this complexity adds up to adversely affect rulemakers whether they are executives in corporations, urban planners in cities or policy-makers in government.
Commonsense is common only “to those who share sufficiently similar social and cultural experiences.” If commonsense were to be generated as a set of rules, put into a logical consistent structure it would be complex to implement because each rule that seems right in one particular context would carry no guarantee of being right in any other context. Meanwhile commonsense of what we would do in a given situation does not necessarily work for predicting others, which might explain why policymakers seem naive and simplistic, that their plans often fail for the rest of us.
An idealized notion of commonsense simply does not work with complex phenomena. At one point it was commonsense to believe the world was flat, and before that, that lightning-bolts were thrown by gods from mountaintops. Commonsense is not a set of truths but are biases expressed as plausible sounding stories. And commonsense may also inhibit thinking when there are negative consequences for expressing alternative models (consider Galileo and the Church). Commonsense then is a model of the world that we have in our heads. And from these biases, false models, and hidden assumptions we have in our heads, we deceive ourselves into believing we can make predictions about things we really know nothing about.
An example of false cultural impressions given in the book is from widely varying percentages of organ donor cards across several European nations. The varying percentages of donation can easily lead to the (false) belief that one culture versus another has significant differences over principles of organ donation. In-depth examination however, identifies culture as not the issue but whether the donor card has a default, whether to opt-in or opt-out. The valid commonsense reasoning is that people (across cultures) attend to the default on government forms.
This in-depth reasoning has been called rational-choice theory. Rational-choice starts with the basis that if we want to understand why people do what they do, we have to understand how questions are asked, and incentives, and tolerance for risk. The problem with some of Mr. Watts assertions is that if one limits consideration of behavior to rational behavior with rationality left undefined but assumed by this reviewer to having an economics-centric point-of-view (with an ethos that ranges from Asbergerian to Ferengian to Scroogian), then one may end up with models that are not much better than before.
The scientific study of choice and behavior comes under the intersection of sociology and psychology. Given two basic assumptions of human behavior, one, that people have preferences and two, people make predictions using internal models of cause and effect, then one can develop better models of behavior, and from these models find more efficient ways to influence behavior.
Artificial Intelligence researchers call the modeling of the complexity the “frame problem.” Where the potential list of facts becomes large, the number of rules to be searched also becomes very large, and programming a task which is trivial to a human becomes a very large computer program that will often get things wrong. The model is simplified by bounding the problem context by using a limiting frame that reduces the number of choices. By reducing choices one also reduces the number of rules that result.
There are alternatives to frames. On one front, there is a new approach to programming human-like behavior and that involves statistical models of data that represents actual human behavior. On another front, humans are using advanced social networking tools, which provides a wealth of information on human behavior. Not only to collect statistics but also provides an environment from where experiments may be conducted. If users use Twitter to share information, then rational-choice theory can also use Twitter. Grassroots, meet Astroturf.
Mr. Watts uses crowd-sourcing techniques to recruit and conduct psychological, sociological and economic experiments on (unwitting?) subjects over the Internet on behalf of his corporate sponsor, Yahoo. One example provided is an experiment intended to determine the value of pay-for-performance. The results show that over a range of possible incentives, one could easily choose counterproductive incentives. And for some types of work, financial incentives are largely irrelevant to performance, (which seemingly counters the commonly expressed rationalization for higher and higher CEO pay).
Mr. Watts analyzes from an economic, social, and psychological perspective the meaning of value in aesthetics. In an economic-centric frame the “Mona Lisa” is only famous because it is famous and how it got that way is largely accident, while Shakespeare is no greater a genius than J. K. Rowling, or even the director of the movie, The Hangover. (Please, I am only reviewing; and no, I do not believe that Mr. Watts is either insincere or ironic). Given this point-of-view, given access to the right influencer (i.e. a celebrity-as-genius) one can manufacture idea “contagion” to sell anything one wishes to peddle. Kim Kardashian was reportedly paid $10,000 per tweet to mention a sponsor’s product.
The only apparent flaw in the model is in identifying the influencer. Based on computer simulations, networks are connected in complex ways, and true influencers tend to be accidents of time and circumstance. (How can one find the next Kim Kardashian?—Perhaps influencers are manufactured as well.) The alternative to finding and targeting influencers is by the mass targeting of as many individuals as possible, which when not perceived as counterproductive (i.e. spam), may simply be not cost-effective. (Note that the purchase of an influencer may be considered “selling-out,” and counterproductive).
There’s a difference between models of social behavior and models of physical phenomena. Simplifications necessary for a model to work may lead to convenient fictions, that is, the model too, may have faulty commonsense. Such models can lead to conflicting results, just like commonsense; however, the fact of using a computer to generate results generates influence all its own, and people tend to disregard the possibility that a model may not have coherence or lead to useable conclusions.
At some point it may be smarter to depend on simple stories to get a point across. A story is not any different from a computer model or scientific explanation other than being simpler and easier to understand, and human nature being what it is, whoever tells the better story has the greater potential to be an influencer. Mr. Watts references several popular books about economics and prediction, including Freakonomics by Levitt and Dubner, with Malcolm Gladwell’s Tipping Point, Nassim Taleb’s The Black Swan, and James Surowiecki’s The Wisdom of Crowds. Of these, who is the best storyteller?
To readers familiar with The Black Swan, determining what is relevant may only be recognizable after the fact. And given the difficulty in making accurate predictions, putting emphasis on the right values, asking the right questions becomes of greater importance than making predictions. So concludes the first section, Common Sense.
The second section, Uncommon Sense, begins with a discussion on prediction markets,. The section continues with the limits of prediction in planning, sports, and the wisdom of crowds. The limit for prediction markets, markets not tied to stocks or bonds but future events, is to no surprise, due to the corrupting influence of money. Money drives prediction markets up or down in order to make money on the direction of the market, not on the event. The limit on the wisdom of crowds may be put into perspective with betting on sports events, where some things are just too close to call.
As for strategic planning, one-off events are ill-suited for statistical models. Too much time may be spent on optimizing a current strategy rather than planning for change. Strategic failure often is not from bad strategy but from poor execution, failure to react effectively to changing conditions. Solutions to planning risk can come from scenario planning, from re-planning along the way, and from re-evaluating options as unknowns become resolved through time. One good design for strategy is to separate core necessities from contingent opportunities (and crises).
If some change is unplannable, then how can business react to change quicker? Market research is one method, with the proviso that market research is considered reaction. Conventional wisdom holds that the customer never creates something new - except when the customer does. Crowdsourcing is the concept of letting the market invent for you, by offering the customer something of value in exchange for ideas. Mr. Watts mentions Toyota’s “Just-in Time” manufacturing, which is not crowdsourcing but not their recent advertising campaign, “ideas for good,” http://www.toyota.com/ideas-for-good, which is. Whether one manufacturer’s acceptance of crowdsourcing is an indication of true change, or just another fad, time will tell.
This book appears to be of better use as an idea generator than as a cookbook of solutions. Perhaps as science, sociology, and psychology are primarily steered toward the goal of making profit out of social behavior, a better title might be Everything Is Commoditized. But that’s unfair. Prediction is complex, and Mr. Watts has not presented it as anything different. While some readers may not agree with the claims and biases of the author, the ideas presented are expressed clearly and in such a manner to allow all readers to draw their own conclusions.