Deliberate Practice

What it is

The 1993 paper is Ericsson, Krampe and Tesch-Römer, "The Role of Deliberate Practice in the Acquisition of Expert Performance", in Psychological Review. Study one looked at forty violinists in four groups of ten at the Music Academy of West Berlin, with the fourth group of professional orchestral players excluded from most of the inferential analyses. Study two compared twelve expert and twelve amateur pianists.

The definition is restrictive on purpose. It exists precisely to separate the practice that improves you from the volume of work that does not, and the restriction is the reason the theory was interesting in the first place.

One correction belongs high on this page because the popular version of the claim rests on it. The figure of ten thousand hours does not appear anywhere in the 1993 paper. The best group's accumulated solitary practice by age eighteen was about 7,410 hours, the good group about 5,301 and the music teachers about 3,420, and the paper invokes Herbert Simon and William Chase's ten-year rule rather than any hour count. The ten-thousand-hour framing is Malcolm Gladwell's, in Outliers in 2008, built by extrapolating one group in one study forward to age twenty. Ericsson publicly and repeatedly rejected it as a distortion of his work and wrote Peak, with Robert Pool, in 2016 partly as a corrective. The threshold was never a finding and never a threshold.

In effect

The critical evidence comes almost entirely from work Brooke Macnamara led.

Macnamara, David Hambrick and Frederick Oswald published a meta-analysis in Psychological Science in 2014 covering 88 studies, 111 independent samples and 157 effect sizes, with more than eleven thousand participants. Deliberate practice explained about twelve per cent of the variance in performance overall, rising to about nineteen per cent of the reliable variance once corrected for unreliability. By domain the figures were about twenty-six per cent in games, twenty-one in music, eighteen in sport, four in education, and under one per cent in the professions, where it was not significant. The effect is smallest exactly where the popular claim is loudest.

The methodological finding inside that paper matters at least as much as the headline. Broken down by how practice was measured, retrospective interview gave about twenty per cent, retrospective questionnaire about twelve, and an actual practice log about five. The more valid the measure, the smaller the effect.

Macnamara, David Moreau and Hambrick then ran the sports version in Perspectives on Psychological Science in 2016, finding about eighteen per cent of the variance overall and about one per cent among elite performers. Practice separates sub-elite athletes from average ones and does not reliably separate elite from elite.

The decisive study is Macnamara and Megha Maitra, in Royal Society Open Science in 2019, a preregistered, results-blind, double-blind direct replication of the 1993 violin design with thirty-nine players in three groups. The group rated merely good had logged an average of 9,844 hours. The group rated best had logged 8,224. The best did not differ from the good and were numerically behind them, and on current weekly practice the best practised significantly less. The authors carry their own hedge, which this page carries with them: practice explained about twenty-six per cent of the variance across the three groups, and they say plainly that twenty-six per cent is not an inconsequential amount. They also report that the confidence interval around the original 7,410-hour mean runs from roughly 2,894 to 11,926 hours, which contains the good group's mean, so the original study was never powered to separate the top two groups.

Ericsson's side is legitimate and should be given properly rather than as a formality. His objection is definitional. The meta-analyses count structured practice, not deliberate practice as defined in 1993, and many of the included studies never use the term at all. A reanalysis restricted to studies meeting the original definition puts practice at about twenty-nine per cent, rising to about sixty-one per cent once measurement unreliability is corrected for. He made that case in Intelligence in 2014, and with Kyle Harwell in Frontiers in Psychology in 2019. His reply to the sports meta-analysis appeared in the same 2016 issue of Perspectives on Psychological Science that carried it, arguing that the meta-analysts defined deliberate practice far too broadly, sweeping in group activities, competitions and play, and that summing hours of any practice is not the theory. Miller and colleagues argued on similar lines in Current Psychology in 2021. Measuring the wrong thing does give the wrong number.

The rejoinder belongs with it. Macnamara, Hambrick and Moreau replied in that same 2016 issue, arguing that Ericsson's definition has shifted over time in ways that make the theory unfalsifiable. That is the live objection and it is not answered anywhere in the record this vault has read.

What it does not say

It does not say practice does not matter. Both sides agree practice is necessary. No responsible figure on either side claims practice explains expertise, and the 2019 replication authors themselves call a quarter of the variance substantial.

It does not support ten thousand hours. That number is Gladwell's extrapolation, not Ericsson's finding, and Ericsson spent years saying so.

It does not hold at the elite end. Among sub-elite and average performers practice separates people. Among elite performers the effect shrinks towards nothing, which is precisely where the advice industry aims the claim.

It does not rest on solid measurement. Retrospective self-reported hour counts are the literature's weakest joint, and the better the measurement, the smaller the effect gets.

It does not resolve the definitional argument. Ericsson is right that the meta-analyses count something broader than his construct. Macnamara is right that a definition which moves when tested is hard to falsify. Both objections stand.

And it is not what Cal Newport's readers were given. Deep Work disputes the critics in one footnote whose sole authority is Ericsson's own rebuttal, characterises the critics as people suspicious of all things Gladwellian, and then never revisits the question, going on to build a daily working budget on the 1993 figures. No critic appears anywhere in twenty-three pages of endnotes. The underlying advice in that book, that the quality of concentration determines what practice is worth, is not what is in dispute and survives all of this. The footnote does not.

---