I recently watch Josh Kaufman's TEDxCSU talk about his new book, The First 20 Hours. Although his idea sounds reasonable, I coudn't find any mention of scientific research to back it up. I know some research has been done on the 10,000 hour rule.

Kaufman's claim: "You can go from knowing absolutely nothing to performing noticeably well in a very short period of time: approximately 20 hours, often less."

Is there any evidence to back up Kaufman's claim?

  • 2
    I have the feeling that the claim is really "You can learn more than you think in 20 hours", which is at least as much about what people ecpect as about how much they can learn. – P_S Sep 3 '14 at 11:19
  • 2
    This may be true in very narrow areas, but I don't believe it WRT "big" topics.. No matter how good you are, you can't learn how to write assembly in 20 hours. The breadth of knowledge required (Cpu/Memory architecture, execution flow, etc, etc) not to mention all the ancillary data (what is hex). – Basic Sep 3 '14 at 12:41
  • 2
    @Basic: You can learn to write assembly "noticeably well" in 20 hours. – Flimzy Sep 3 '14 at 12:46
  • 3
    @Flimzy I suppose it depends on you definition of "noticeably well" but I have to say I strongly disagree. Try it with a family member / another guinea pig. From no knowledge to something that will build and run in 20 hours? Again, perhaps it's our definition but I don't consider doing something by rote "noticeably well". Good luck to them when it comes to debugging! – Basic Sep 3 '14 at 13:09
  • 8
    The phrase "noticeably well" is so vague that I don't see how this question can have an objective answer. I also question that the speaker means this to apply to literally any skill. – Nate Eldredge Sep 3 '14 at 14:19

The talk says that "10,000 hours" is, to within an order of magnitude, the amount of practice required to get expert-level performance: e.g. to be a professional athlete.

It says that, conversely, with a bit of practice you can get really good, really quickly.

The following is, I think, an example of the truth of that:

Those who pass their driving test have had, on average, about 45 hours of professional training combined with 22 hours of private practice. Learners who prepare this way, with a combination of plenty of professional training and plenty of practice, do better in the test.


  • Driving a car in traffic is (IMO) a reasonably complicated skill.
  • Doing it well enough to pass the driver's license test is doing it reasonably well.
  • This statement from the British Government is presumably based on plenty of experience (experiment).

The cited time (45 plus 22) is a bit higher than the "20 hours" you asked about; but it's very much the same order of magnitude (much closer to "20 hours" than to "10,000 hours").

Beware though that the above is a (one) specific example, not a proof of the general case: it's "a" skill, not "any" skill. The TED talk does actually say "any":

... about any skill you can think of. Want to learn a language? Want to learn how to draw? Want to learn how to juggle flaming chainsaws? If you put 20 hours of focused, deliberate practice into that ting, you will be astounded.

I suspect it's also possible to come up with some counter-examples: for example, that nobody gets "really good" at writing chinese in anything like 20 hours.

| improve this answer | |
  • 3
    Someone may not get really good at writing some dictated Chinese in 20 hours, but I'm pretty sure they could learn a passage of script, memorize it, learn how to write it, and learn what each of the script characters are so that when an observer points to a character and says 'what is this part' they could answer. It sort of depends on the criteria you use to measure 'really good'. – JonW Sep 3 '14 at 10:54
  • @JonW As well as depending on the criteria for 'really good' (which I think is clear in the context of the first few minutes of the talk), may be even more dependent on how large the skill is: for example I should think that you could probably memorize a sonnet in 20 hours, but not memorize the complete works of Shakespeare. – ChrisW Sep 3 '14 at 11:00
  • Can you find stronger, more general evidence? – Sklivvz Sep 3 '14 at 11:06
  • 1
    @Sklivvz I won't be looking for it. The purpose of the TED talk was to say that you can learn some things, useful/desirable things, in a relatively short time: especially that the oft-quoted "10,000 hours" was based on studies of expert-level performance, and doesn't imply that it takes 10,000 hours to learn or to become good at anything. IMO driving is a useful, real-world example of that: and very "general" evidence, in the sense that most adults have experience with having learned to drive. – ChrisW Sep 3 '14 at 11:12
  • 1
    @yokimbo There's presumably some continuum: some things can and other things can't be learned in 20 hours. What kind of evidence are you looking for, or, evidence of what? – ChrisW Sep 3 '14 at 21:53

Not the answer you're looking for? Browse other questions tagged .