Review: Malcolm Gladwell’s “Outliers”
Malcolm Gladwell’s Outliers is a collection of anecdotes about successful human careers, ranging in domain from computer science to hockey to the law. In these episodes he reveals hidden environmental precursors to eventual success or failures. An example is that of hockey players in the elite Canadian junior leagues, who are far more likely to have been born in the first few months of the year than would be expected from a random selection from the population. Gladwell suggests that the responsibility lays with the birth date cut-off of January 1st that determines team eligibility in the Canadian children’s leagues; older players in each cohort have had more time to develop physically than their autumn-born classmates, and their size gives them an advantage at try-outs. This sets them on the road to athletic success from an early age with better competition, more ice time, and better coaching at each level. By the time they hit 20 or 21 years of age, the gap in talent between a January-born and a December-born player of similar natural skill might be the difference between the Detroit Red Wings and the Pensacola Ice Flyers.
Similarly, Bill Gates was a brilliant person who was fortunate to attend one of the few schools in the country with mainframe computer access in 1968; the Mothers Club had put up the funding, and a connected parent at the school was able to get him access to UW machines when the funds ran out. By developing these skills at an early age, Bill Gates had a head start on the coming computer revolution, and was able to experiment and innovate faster than the IBM career-men of an older generation. But without the stroke of luck provided to him by the Mothers Club in his environment, the book implies, the world might never have been graced with DOS.
The most coherent point I got from Outliers is “nurture matters.” Success occurs when smart and talented people take advantage of the opportunities given to them. That point seems inarguable. I do have an issue with Gladwell’s singular focus on the “freaky” attributes of the environments that nurtured the success of his characters. I think it leads to an unhealthy understanding of causality and the nature of outliers. A probabilist would say that all environmental factors predict success, but they do so with varying degrees of predictive power – computer access in school, for instance, might be a better predictor of success than eating Wheaties for breakfast – but so might be birth order, intelligent friends, and inspirational teachers. By focusing the book on the Strange but True, he slights such factors that have greater predictive power. This is important, since Outliers is overtly prescriptive…the jacket cover claims the book is “a blueprint for making the most of human potential.” He sees something unexpected in P(born in January | hockey player), but this leads to incorrect value judgments about P(hockey player | born in January). For example, P(hockey player | born in January) < P(hockey player | having a hockey coach for a dad), and P(hockey player | born in January) « P(hockey player | being really big and really fast). The probability of being a Jeff Beukeboom is greater if you are 6’5” and 230 pounds than if you are born in March.
The human eye/brain instinctively saccades to outliers, and its first question is usually “what are they doing there?” – but this comes from a misunderstanding of outliers and what to do with them. There are two plausible explanations for the presence of outliers. One is that your data is noisy; you were unable to capture the model accurately due to crappy measurement, so the outliers should be ignored. The other explanation is that the outliers were generated from the same model that produced the rest of your data; God flipped The Big Nickel and it landed not on heads or tails but on its side…a freak occurrence, but surely one that has some probability, however infinitesimal. And when She flips it a hundred billion times (the estimated number of humans who have ever lived), there are going to be some Bill Gateses and Barack Obamas and Bobby Orrs out there on the fringes of the distribution. Certain aspects of ones nature and environment can be strong predictors of the variables people commonly associate with success (power, wealth, fame), but it’s not really fruitful to focus on any one feature more than any other, unless you can say unequivocally “birth date relative to January 1st explains more variance in Canadian hockey players’ success than presence of favorable alleles responsible for building fast-twitch muscle fibers.” Gladwell says of the hypothetical January-born hockey phenom, “he didn’t start out an outlier. He started out just a little bit better.” But the thought begs continuation; he started out just a little bit better than the tiny fraction of Canadian boys good enough to be at the tryout in the first place! And what caused those boys to be there in the first place? A whole mess of factors that go unmentioned and unexplored…but their potential contribution is no less important.
Ed Ricketts ruminated on the pitfalls of such “teleological thinking” seventy years ago on a survey of intertidal invertebrates in Mexico, catalogued in Steinbeck’s The Log from the Sea of Cortez. Take a box of matches, Ricketts says. Why is the longest match longer than its fellows? Because, the teleological thinker responds, the machine that created it was depressed slightly longer than usual when it was created, slicing it off at a point further than usual from the end of the match. Why did that happen? Because the air in the factory was slightly more humid that instant, affecting the spring mechanism in the machine. Why was it more humid that instant? Because a butterfly flapped its wings in Africa….and back it goes, to “causes” that individually are weaker and weaker predictors of match length…but collectively, they completely determined the outcome.
The neo-Rickettsian psycho-probabilist’s response to the question is that the longest match is the longest because it is the longest. There is some distribution of match length out there, and you are staring agape at the outliers because the human eye/brain is naturally attracted to them. No one pays second thought to an average-length match, but it comes from the same mother generating-model as long ones, fat ones, short ones, blunt ones, double-headed freak ones… and there’s really nothing left to say, except to calculate the probability of generating a given length of a match, either globally or given specific information about air humidity, temperature, the age and consistency of the oil in the machine, the make and model of the butterfly…
The joint probability was determined, The Big Nickel was flipped, and one peculiarly long match was deposited in a box. Q.E.D.
If you subscribe to this approach, then you are left with a bitter taste in your mouth as Gladwell leads people to focus on the “gotcha,” the factoid, instead of teaching people to think holistically about probability and causality. People are the way they are because of the gajillion factors that led them there, in their DNA and in their environment. Some flips of the coin are more important than others, but focusing too much on a single flip misses the forest for the trees. Bill Gates is Bill Gates, and we should write his history and marvel at his achievements. To create more like him, we should tackle the challenge of calculating marginal probabilities of all factors that predict the success metrics we collectively agree are worthy of pursuit, and focusing our energies on the best predictors we are able to affect. Unfortunately, we can only do this if people are taught how to approach these problems critically, and Malcolm Gladwell does his readers no favor in this respect. This ( http://www.ted.com/talks/arthur_benjamin_s_formula_for_changing_math_education.html ) is where I would start.
2 years ago
