Monday 2 April 2012

10,000 hours, 10,000 questions

I read Malcom Gladwell's Outliers and now feel bad for such a waste of perfectly good paper and ink. Having avoided purchasing a copy, I am still hoping to have my library history expunged. It's a mockery of critical thinking, however thinking for yourself was probably the last thing on Gladwell's agenda. In fact I cannot be sure what was even the first. I found no properly supported line of reasoning, nor any shade of doubt cast on some rather contentious issues.


The book's initial, supporting thesis statement is that hard work, not inborn skill or talent, is the key to success. This statement is meant to support a second line of reasoning being that societal support (i.e. financial and emotional) from family and friends is the single most important key to success. In other words to be successful at, say, computer programming you had to have been born in the United States during the mid 1950s. I too have observed the pre-cambrian era was a comparative dry spell for quality C+ coding. But it is beside the point whether either thesis statement is true. It is Gladwell's job is to provide convincing arguments why life is so. Though I certainly agree that nurture is a greater predictor than genetics, this book failed to teach me anything about either.

In chapter 1 Gladwell opens with a tale of hockey in Canada and our mis-appropriation of talent. I had high hopes for this start (it was clear where he was going with this) but to borrow from Charlie Wilson, he fucked up the endgame. He points out that most of the top players in Canada in the NHL, CHL, junior tournaments, et cetera are born in the first three months of the year, January, February, and March (a total of 40%, instead of the expected 25%; for note, Wayne Gretzky was a January baby, and Gordie Howe was born in March). The reason for this discrepancy is that Canada is fiercely competitive about hockey, and good players are selected early on. Players are in fact chosen so early that an early-year birthday is a competitive advantage, having up to 364 days more maturity than late-comers. Gladwell then explains that the advantages accumulate with time, cumulating right up to the professional levels. Apparently the same problem persists in the Czech as well. So far so good, though none of this data or observation analysis is due Gladwell himself, only the end passage:
If all the Czech and Canadian athletes born at the end of the year had a fair chance, then the Czech and the Canadian national teams would have twice as many players to choose from.
This sounds reasonable, except that it's not true. In fact it contradicts his own reasoning. Gladwell argued that early advantages in hockey players are cumulative, hence innate talent was not screened by our system but instead just those lucky enough to be born early in the year. This bias is unfair, yes, but if Canada were to create two streams (one for early-year players and one for late) we would not have more players; instead we would have different ones. There is a finite pool of money in Canada to develop our best players (yes, even for hockey). We would need two separate leagues to develop two streams of players separately (perhaps akin to an American and National baseball league?). That is, the only way to cumulate advantages all the way to the top would be to double our resources. In such a scenario the best players would still not surpass our current crop. In fact adding an extra round of competition into making our national team may only serve to further drain our best players. Recall that Canada's best hockey players come from small towns (like Gretzky) to degree that dwarfs the birthday bias. My hypothesis is that small town hockey players delay a highly competitive environment (while still maximizing exposure to the game) for longer periods than big city players. The only way Gladwell's quoted statement could be true is if genetically talented October-December players were being left behind at the early stages, but this is precisely what he is arguing against. Strike one.

In chapter 2 we reach the centre of this rotten egg. Herein lies the now (in)famous 10,000 hour rule which acts as a foundation for the remainder of the book. The premise is that ten thousand hours of hard work may (or may not) lead to success for a given subject, or as he alternatively rephrases, within about ten years. Within 33 pages (fewer when accounting for the five-page table of historically wealthy figures) Gladwell explains to us that all successful people have achieved their success through hard work. Had he left the statement here he may have had an acceptable, though trivial, premise. He attempts however to quantify this hard work by comparing the hours of practice spent by successful people including Bill Gates, Bill Joy, Mozart and The Beatles (he probably meant to specify McCartney and Lennon) and showing they are equal:
Researches have settled on what they believe is the magic number for true expertise: ten thousand hours
Which researchers? (for not all agree with this idea) His only direct citation comes from Daniel Levitin, a professor at McGill University. From his book This is Your Brain on Music I found he is actually more reticent about this theory than Gladwell would lead you to believe.
Like many scientific theories, the ten-thousand-hours-theory has holes in it, and it needs to account for counterarguments and rebuttals
Though Levintin ultimately defends the 10kh theory against a total of one counterargument (focusing on Mozart), his biased measure of 'success' is more or less safely constrained to practised skills of the musical or hand-eye co-ordination type. He also admits genetics, exposure, practice, and random luck may all play a confounding role and does not emphasize the rule easily translates business professional or abstract language-dependent skills (I am assuming, as he does not mention them). Also note that the his quote was in a chapter titled "What makes a musician"; I believe Levitin's brief attempt to generalize was likely a momentary err in judgement.

Gladwell, by contrast, seems only too happy to expand this rule into all domains and professions, as if it were the discovery of a new Feigenbaum constant. He uses the rule as a basis to explain innate ability has less to do with success than the combination of early environment exposure and years of practice. Were that the end of his observation it would be a fine, even optimistic, argument. But the damage comes with his attempt at quantification. I have beef with both Gladwell and Levitin's assertion of this 'rule', but moreso with the former, lacking any defense -or devil's advocacy- for alternative explanations or possible stress tests.  That said, I will offer some. The following doubts passed through my head while reading about the 10kh rule. I will even provide, as a bonus, my own explanation why this 'rule' apparently holds true; the answer is far simpler than these writers would have you believe. Because this blog is supposedly about running, I'll include relevant references when possible. Also movies too. I like movies.

A few doubts and assorted questions I felt were not adequately addressed in the original premise:
  1. Division of time: 10,000 hours is indeed a long time. Gladwell estimates it translates to 20 hours of work a week for ten years and leaves it at that. But what about 10 hours a week for 20 years, or why not 5 hours a week for 40? Might one cram success into twelve-hour bouts for 834 consecutive days? How might the hours be divided within a week, i.e. 12 hours one day, then two days of rest? Is taking a break (i.e. for a month or two) always a bad thing? In a later chapter he feels compelled to argue this is so for education, but to generalize there would be ridiculous (hockey seasons, anyone?). The rather important options are never discussed and makes the rule sound as useful as guessing how many pennies are in my pocket by knowing my bank balance. 
  2. Start time: When do these 10,000 hours officially begin? At birth? From age five? What if someone were to begin a profession at the age of forty. It is well-known that runners starting later in life, sometimes not until 70 years of age, will make improvements. Learning to fly a plane cannot take place until your teenage years, nor becoming a doctor until your 20s, and surgeon in your 30s. Elected political leaders do not hold office usually until their 40s. Do age-category achievements count, such as winning a senior PGA tournament? Certain abilities such a language are designed to be mastered when very young. Fassbinder's directing career was over by age 37, but Clint Eastwood's did not begin until after 40. 
  3. 10,000 hours of what? An egregious oversight, Gladwell never admits that quantifying practice time is much, much harder than it sounds. Examples abound. How many hours of math practice did Ramanujan get in his early years? Or how much formal training for that matter? For runners, how many hours are divided into easy running, sprinting, strength drills, stretching, mental preparation, resting, sleeping, and racing? Alfred Hitchcock described actual film directing as 'monotonous' since he already had most of the film planned on sketch pads. Werner Herzog by contrast spent a minimum of time on planning and scripts (he wrote Aguirre: Wrath of God in a weekend) and filmed quickly, mostly on instinct. So which hours here are counted? Paul Erdos was famous for always thinking about math, in his later years comprising about 18 hours a day. Conversely Glenn Gould was known for playing very little actual piano, choosing instead to study music by sight. Myself, I prepare for oral presentations much the same way; I often avoid rehearsing all together, instead I focus on studying the material itself. Roger Bannister was noted for running low mileage but probably prepared in excess mentally, and Sebastian Coe was notorious for underreporting his own mileage. Can any outside arbiter objectively measure the divisions of time and effort by all these individuals? How to quantify scientific endeavors that include geology or collecting insects, (where expertise may be a simple correlation with time spent in the field) versus the DNA findings of Watson and Crick, or Galileo's discovery of Jupiter's moons at age 45?
  4. How steadfast is the rule, really? For instance, can a zero-hour rule be claimed? Since even the most talented person can be improve with practice, there will always be an X-hour rule for each person, hence any such rule amounts to a certain amount of triviality. How universal is X? Could a 50,000 hour rule exist? Likely not, as this is tantamount to saying one would need 20 hours of practice a week for fifty years. It is difficult to imagine any talent taking this long to hone, yet there are probably examples, such as Buddhism or horse breeding. And is less than 10,000 hours ever enough to achieve expertise? Of course, including carpentry at 5,000 hours and mechanic apprenticeships that lasts about that long. These numbers also change with market demand. And as every new pilot knows, obtaining 1,000 hours of flight time is the magic number. Shirley Temple could have been said to reach her 'peak' at seven with her honorary Oscar, having been alive (and awake) for a mere 40,000 hours. What about NASA's 1959 Mercury program, which trained and launched astronauts inside four years. Clearly there are some discrepancies.
  5. Is the time 'to expertise' for all experts 10,000 hours, or is it an average? Averages are dangerous tools when misused. Can it be that for half the field people reach their peak in 8,000 while the other half do it in 12,000? These are important distinctions.
  6. How long does an expert-phase last? Admittedly this isn't what Gladwell set out to discuss, but it is rather important to distinguish careers in terms of length. Which is better, running a 3k in 7:35 once or ten in 7:55 (repeat performances are a better sell to sponsors)? In movies, some directors hone their abilities over decades, as in Ozu or Bergman. Picasso changed his art style so regularly that he had several distinct periods, each worthy of a career. Conversely James Dean's career born and died in three movies yet they made him a legend.  
  7. When is an expert an expert? When exactly did Haile Gebrselassie become an expert at running? Was it when he made the Ethiopian national team, when he beat Paul Tergat for Olympic gold in 1996 or when he set the (then) marathon world record of 2:03:59 in 2008? And when were Samuel Pepys' abilities as a raconteur first realized? Did Bill Gates succeed when he created window 3.1 or 95, when he made his first million or billion? 
  8. What kind of environment did experts practice in? Here Gladwell states that early success comes from early encouragement, but no-one is self driven to success merely by kind words. Many, many hours are done in private reading, but helped along with advice from tutors. The list includes professors encouraging Riemann and Feynman to study certain math books. Bode Miller is known for training in downhill skiing in rustic environments, whereas Annika Sorenstam uses spreadsheets to calculate all of her golf scores. Often Kenyans are said to be social runners and that is their key to success, yet it has also been argued that El Guerrouj reached his legendary status when he stopped running with pace groups for his speed work.
  9. This last doubt is a subset of the 10k rule. Gladwell seems to believe wholeheartedly that more is better when it comes to spending time on a problem while young. This is true in some cases, but can it really be a general rule? Is there a proportional relation between time spent on a problem and solving it?  The example he provides is the asian success in mathematics and their tradition of hard, persistent work. It is true that Asians do well at rote math, but dominance at more creative levels fades. As expected, victories in the International Math Olympiad, held for high school students, are held by China. But take a look at the Putnam contest, held for college students, and the hardest math test you will ever see. Here the Chinese winners are comparatively fewer (the contest is american, but we assume the best math students China might attend the finest universities in the world). Then at the very highest level of achievement (for persons under 40) is the Fields Medal, where the Asian dominance is all but gone; instead a European advantage is observed. Meanwhile the Abel prize open to people of all ages shows a clear American dominance. So the Chinese system is failing to produce outliers (What was Gladwell's book called again?) Rigorous teaching of math is a tricky enterprise; South Korea students spend ridiculous hours at school, including arithmetic. The country is not known for mathematical innovation, though they are certainly hard workers.
  10. Could the 10,000 hours be a self-fulfilling prophecy? When people know what to look for, it is easy to find. Has a double-blind test been conducted (i.e. researchers are not told that a target count is being sought). I know of runners who have a tendency to include miles run when they are short of a mileage goal, and omit these same miles when in excess, myself included. Worse, we are collectively prone to making such theories true by definition: see the same issue that haunts Moore's Law
As I already stated, no-one is born with a talent, not really. If you were born to play piano but lived in Greece circa 5000 BCE you would have never known the difference. A Kenyan born to ski is a Kenyan with no future in sport. A Canadian downhill skier born in the prairies has a severe disadvantage. Few skills are innate; language must be taught, even locomotion takes humans years to master (whereas a gazelle learns to walk in minutes). We are born 'awaiting' certain imprints, that is all. Biology guides us, more loosely than for other animals. What part of us desires to keep learning and achieve excellence is hard to explain. The measure of IQ is of course meaningless; read through Gould's excellent Mismeasure of Man if you have any doubts on that. Gladwell agrees IQ is not enough, though he believes it exists, and we're born with a certain IQ (but that is correlated poorly to success). After early encouragement from parents, add to this the 10k rule and voila: if you have the talent plus hard work plus the societal influence you win. But I see it differently. 

We learn certain skills as a young person. Some useful for later, some not. Physical and mental abilities develop, some may or may shine through over the years. Sometimes exposure time shows poor correlation to memory. Consider one of Citizen Kane's most serene passages, a scene between Thompson (the never-seen interviewer) and Mr. Bernstrein:
Thompson: We thought maybe, if we can find out what he meant by that last word - as he was dying - 
Bernstein: That Rosebud? Maybe some girl? There were a lot of them back in the early days, and-
Thompson: Not some girl he knew casually and then remembered after fifty years, on his death bed- 
Bernstein: A fellow will remember a lot of things you wouldn't think he'd remember. You take me. One day, back in 1896, I was crossing over to Jersey on the ferry. And as we pulled out, there was another ferry pulling in. And on it there was a girl waiting to get off. A white dress she had on. She was carrying a white parasol. I only saw her for one second. She didn't see me at all. But I'll bet a month hasn't gone by since that I hadn't thought of that girl.
If our memory were as a simple function of exposure time, our deathbed memories would mostly be of work and trips to the bathroom. If memory of a subject is poorly correlated to skill or understanding, then I really have no idea what makes up talent (and I really do not). Talent is a tricky beast, especially when the fundamental units are hard to pin down. In a world of an infinite and ever-changing talent basis, it is also impossible to isolate talent by genes, nor would we benefit by doing so (were it even possible) due to the lack of predicability of what, pray-tell, a future talent actually is. Often the diamond in the rough remains hidden until too late to change course. Then, just as Gladwell says, the feedback loop enhances the abilities of those who shine early, though some ultimately win by not shining through too soon (ever notice the prettiest girls in high school are not the prettiest at age 40?). And unforgiving talent-filtering methods such as the many rungs on the Canadian hockey ladder could be hurting young talent. Gladwell misses the boat here; the bias in January babies is a symptom of a problem, not a cause. As promised earlier, now for an alternative explanation to the 10kh rule:

We are a culture of specialization. Not everywhere on earth, but in North America and China where this theory is cited to work. Assuming 'the best' means being on top, in a very high percentile, then no matter how much talent you begin with (not much at birth, of course), it will never compare to someone who works hard. The world's best linguist is far worse than a young child in all but 25 languages (out of 5,000 or more). Therefore you must work hard. Assuming time is devoted to eating, sleeping and doing unrelated activities, only so many hours can be spent learning one subject. If that subject requires complex knowledge (say quantum physics or surgery), then the hours are further split. A lower limit of hours spent on is required by competition with others, an upper limit created by other duties in life. Once you reach your 20s, it is time to choose a career. By this time thousands of hours will have been spent in your endeavor, as an inevitable byproduct in life. How those hours are counted is clearly dependent on the subject itself and varies widely (see point #4). Perhaps a more useful definition would be to define talent as the one who works the hardest without realizing it. In that vein, those who quantify every mile they run might eventually associate 'mileage' with 'obligation'. This is a delicate matter requiring as much psychology as hard work. The one invariable I have seen among successful people is they really don't spend much time thinking about how hard they work; they are too busy working hard (and often having fun). Can you imagine Richard Feynman quantifying his joy for physics? Recall he once said "Physics is like sex: sure there are some practical results, but that's not why we do it"

Returning to the book, the remaining chapters played out as expected, a series of weak anecdotes demonstrating that lawyers come lawyer families (Jews), hard workers tend come from hard working families (Asians), Southerners are sensitive to name calling (wait, what?), and how Korean pilots were far too polite for their own good (they switched to English and consequently saved lives, apparently). Nice stories, but they prove nothing about the mechanism to success for the individual 'outlier'. Gladwell finds it necessary to point out that culture is important to success. Well, of course we already knew that. People are reading this book because they think it will reveal the secrets behind that success; this is why it was a #1 bestseller. It does not answer such questions. If you would like a quick answer to one person's talent development, read Stephen King's very personal journey behind his success in On Writing, a memoir of the craft (note: King has no idea where his writing talent came from). Every page is worth more than Gladwell's book.   

I doubt anyone with a normal mental capacity takes Gladwell's ideas seriously (or for you IQ disciples, those with an IQ in the bottom quartile). Ok, so I didn't have 10,000 questions to ask, but to be fair the number 10,000 seems to have little to do with anything. Don't buy into this theory: not only does it explain nothing, it predicts nothing. To borrow from Pauli, this theory is not only not right, it isn't even wrong.