The good news: Talent is meaningless. You can be world-class at anything.
The bad news: The formula requires absolute commitment: 3-4 hours daily of pleasureless, intense practice, and fifty to eighty hours a week of total domain-related time… for a decade. You must always get a full-night’s sleep. And you can’t catch up to anyone who’s got a head start.
Annotated selections from the paper follow.
Experts’ memory for representative stimuli from their domain is vastly superior to that of lesser experts.
Novel programmer interview technique: Show the programmer a program. Give them a moment or two to study it. Take it away. Take a short break. Ask them to reproduce it.
Adults perform at a level far from their maximal level even for tasks they frequently carry out… “It is that we have too many other improvements to make, or do not know how to direct our practice, or do not really care enough about improving, or some mixture of these three conditions.”
“Do your best” is hard advice to follow.
When Tchaikovsky asked two of the greatest violinists of his day to play his violin concerto, they refused, deeming the score unplayable. Today, elite violinists consider this concerto part of the standard repertory.
“World-class” is a moving target. World records are broken by performers with more resources available for more practice.
To make an eminent achievement, one must first achieve the level of an expert and then in addition surpass the achievements of already recognized eminent people and make innovative contributions to the domain.
The 10-year rule
It takes normal individuals approximately a decade to acquire [their adult vocabulary.]
The time between chess players’ first learning the rules of chess and attaining international international chess master status was 11.7 years for those who learned chess rules late (after age 11).
J.R. Hayes confirmed that 10 years’ experience is necessary in another domain, musical composition.
Simon and Chase’s “10-year fule” is supported by data from a wide range of domains: Music, mathematics, tennis, swimming, and long-distance running. Long periods of necessary preparation can also be inferred for writers and scientists.
In many other domains, the highest level of expert performance is displayed by individuals with more than 10 years of experience: evaluation of livestock, diagnosis of X-rays, and medical diagnosis.
Motivation and perseverance are necessary for attainment of eminent performance.
The relation between acquired performance and the amount of practice and experience was found to be weak to moderate in the earlier review. We propose that the reason for this comparatively weak relation is that the current definition of practice is vague… we must analyze the types of activities commonly called practice.
We want to distinguish activities invented with the primary purpose of attaining and improving skills from other types of everyday activities, in which learning may be an indirect result. On the basis of several thousand years of education, along with more recent laboratory research on learning and skill acquisition, a number of conditions for optimal learning and improvement of performance have been uncovered. The most cited condition concerns the subjects’ motivation to attend to the task and exert effort to improve their performance. In addition, the design of the task should take into account the preexisting knowledge of the learners so that the task can be correctly understood after a brief period of instruction. The subjects should receive immediate informative feedback and knowledge of results of their performance. The subjects should repeatedly perform the same or similar tasks.
There is no activity in the common education of programmers that matches the description above.
In the absence of adequate feedback, efficient learning is impossible and improvement only minimal even for highly motivated subjects. Hence mere repetition of an activity will not automatically lead to improvement in, especially, accuracy of performance.
Studies show that providing a motivated individual with repeated exposure to a task does not ensure that the highest levels of performance will be attained. Assessment of subjects’ methods shows that inadequate strategies often account for the lack of improvement.
How many times do programmers receive feedback on their programs? A few times per semester for a student, then once per code review in industry?
As the complexity of a desired skill increases beyond the simple structure of most laboratory tasks, the logically possible methods to correctly and incorrectly perform the task by subjects increase as well. To assure effective learning, subjects ideally should be given explicit instructions about the best method and be supervised by a teacher to allow individualized diagnosis of errors, informative feedback, and remedial part training. The instructor has to organize the sequence of appropriate training tasks and monitor improvement to decide when transitions to more complex and challenging tasks are appropriate. Although it is possible to generate curricula and use group instruction, it is generally recognized that individualized supervision by a teacher is superior.
Tutoring yields better performance by two standard deviations—the average tutored student performed at the 98th percentile of students taught with the conventional method.
Close supervision is rare. Most feedback for students comes from overworked TAs who take a quick read through a program.
How much would the software industry earn and save if its programmers were two standard deviations better? Is there a positive ROI on tutoring?
Given the cost of individualized instruction, the teacher designs practice activities that the individual can engage in between meetings with the teacher. We call these practice activities deliberate practice and distinguish them from other activities, such as playful interaction, paid work, and observation of others, that individuals can pursue in the domain.
What is deliberate practice for programmers?
You can’t practice on the job
Individuals given a new job are often given some period of apprenticeship or supervised activity during which they are supposed to acquire an acceptable level of reliable performance. Thereafter individuals are expected to give their best performance in work activities and hence individuals rely on previously well-entrenched methods rather than exploring alternative methods with unknown reliability. The costs of mistakes or failures to meet deadlines are generally great, which discourages learning and acquisition of new and possibly better methods during the time of work.
The workplace is unsuitable for learning advanced programming skills. That leaves undergraduate programs. Yet, a low GPA is the (severe) consequence of failure on most undergraduate programming projects. Small wonder undergraduates are reluctant to try new techniques and tools such as unit testing, formal specification, debuggers, and IDEs.
Recent analyses… reveal an enjoyable state of “flow”… an enjoyable state of effortless mastery and execution of an activity. This state of diffused attention is almost antithetical to focused attention required by deliberate practice to maximize feedback and information about corrective action.
We claim that deliberate practice requires effort and is not inherently enjoyable.
A necessary precondition for practice… is that the individual be fully attentive to his playing so that he or she will notice areas of potential improvement and avoid errors… Practice without such concentration is even detrimental to improvement of performance.
Practice Time is Everything
There is a complete correspondence between the skill level of the groups and their accumulation of practice time.
At the age of 18 the expert pianists had accumulated 7,606 hr of practice, which is reliably different from the 1,606 hr of practice accumulated by the amateurs.
During the diary week experts were fully engaged in music and spent close to 60 hr on music-related activities.
We’re a very immature field compared to others like music. How many CS graduates had even started programming at 18? And yet we’re the ones programming the rockets, robots and reactors.
Talent is Practice in Disguise
Given that acquired skill resulting from prior accumulated practice cannot be observed, it could easily be incorrectly attributed to native talent.
Years of intensive preparation under the supervision of a master invariably precede the attainment of international recognition.
The real innate difference are in practice precursors
In fact, within our framework we would expect that several “personality” factors, such as individual differences in activity levels and emotionality may differentially predispose individuals toward deliberate practice as well as allow these individuals to sustain very high levels of it for extended periods.
Intelligence necessary but not sufficient
“high but not the highest intelligence, combined with the greatest degree of persistence, will achieve greater eminence than the highest degree of intelligence with somewhat less persistence.”
How to succeed wildly at anything
We view elite performance as the product of a decade or more of maximal efforts to improve performance in a domain through an optimal distribution of deliberate practice.