47

What should I understand from "steep learning curve"? When a computer program (for example a translation program) has a steep learning curve, does it mean that it is not good at learning or it's hard for it to learn?

Lauren
  • 275
Mehper C. Palavuzlar
  • 35,654

3 Answers3

66

In informal usage, a "steep learning curve" means something that is difficult (and takes much effort) to learn. It seems that people are thinking of something like climbing a steep curve (mountain) — it's difficult and takes effort.

As it is technically used, however, a learning curve is not anything to be climbed, and is simply a graph plotting learning versus time. Thus, a steep learning curve would look like this (excuse the poor drawing):

Steep learning curve

One natural interpretation of such a curve, which was the predominant early usage (according to Wikipedia) and still exists in some technical circles, is that the thing being learnt is easy — a great amount of learning happens in a small amount of time. This is the opposite of the popular usage. Now there is also apparently an interpretation of the same curve in the negative sense — probably something about a large amount of learning existing, or that one never stops learning and keeps learning, but I'm not sure I understand how that's negative.

Summary

The popular meaning of "steep learning curve" is "difficult to learn"; the technical meaning is "quick to learn".

[Edit, ten years later]: I just noticed a post from February 8, 2013, by the linguist Ben Zimmer, which identifies the 1970s as when the popular usage developed. The post (also available here) gives two examples each from that decade of the word being used in public in the technical sense and in the currently popular sense (bolding added by me):

Looking through examples of the expression from the '70s, one can find both positive and negative senses. For instance, an article in the Spring 1973 issue of Sloan Management Review about the computer industry includes this line: "Due to economies of scale and a very steep learning curve, the cost of such circuits has dropped by a factor of ten in a little over one year." An article in the February 11, 1979 edition of the Boston Globe about Texas Instruments says that "part of TI's success in having a steeper learning curve — and lower product costs when produced in mass — has been its 'design to cost' system." In both examples, a steep learning curve is a good thing, from the perspective of a business ramping up productivity and trying to keep costs low.

But the phrase was also being used by individuals describing a learning process more subjectively, and in those cases the sense became more negative, with steepness equated with difficulty. Thus, for instance, in December 1978, the newly appointed chairman of NBC, Jane Cahill Pfeiffer, told the New York Times, "I'm on a very steep learning curve, and the bulk of Fred [Silverman]'s experience is not where mine is." The following month, in January 1979, Lord Kearton, chairman of the British National Oil Corporation, had this to say to The Guardian: "Everybody in the North Sea is on a very steep learning curve. What worries us is the prospect of new people coming in with practically no resources of any scale, who will have to start more or less at the bottom of this curve."

It was uses like these (notably both from titans of industry) that helped popularize the notion that a steep learning curve was an arduous and not an easy process.

ShreevatsaR
  • 27,572
  • 4
    +1 for information about actual meaning of this phrase and the way it is usually used. More about this: http://discuss.fogcreek.com/joelonsoftware/default.asp?cmd=show&ixPost=39708, http://stackoverflow.com/q/277618/95. – Marek Grzenkowicz Dec 04 '10 at 16:38
  • 4
    The technical meaning isn't "quick to learn"; by a steep learning curve, what's meant is a curve with a "cliff" in it. That is, something where it takes a while to learn anything at all, and then one "suddenly gets it". It's not such a big jump from there to "difficult", although the original literal meaning was a particular kind of difficult. – Henry Dec 04 '10 at 20:31
  • @Henry: A learning curve is always monotonically non-decreasing (the amount of learning cannot decrease with time) so it can't have a cliff either sloping left or sloping right. So I'm not sure what you mean. – ShreevatsaR Dec 05 '10 at 03:23
  • 7
    From http://www.psywww.com/intropsych/ch07_cognition/learning_curve.html: "People often speak of a steep learning curve when they mean the opposite. A steep learning curve is one in which skill improves quickly, meaning something is easy to learn. However, what most people mean by "steep learning curve" is difficult learning experience. No doubt they are thinking of steep hills and steep mountains which make climbing difficult. In actuality, the steepest part of the learning curve is the portion where learning is fastest and easiest." – Alex Dec 05 '10 at 14:24
  • 2
    I mean a curve with a long, fairly flat region, followed by a big, sudden jump. In other words, consider two skills where expertise in either can be achieved in 100 hours of training. In one task, there is a steady gain in understanding and ability as time is put in. In the other task, one gains almost no ability until after 90 hours, at which point mastery steadily climbs. In one sense, both are equally difficult (it takes the same amount of time to master each), but in another sense, the second task is more difficult. – Henry Dec 05 '10 at 18:47
  • 1
    @Henry: I wouldn't call that a cliff, possibly a plateau. Do you have any citation for such a curve being called a "steep learning curve" (or even discussed), contrary to what the citation Alex gave says? [In any case, the steep part would be the quick part, so it still doesn't make sense to use "steep" for difficult.] But I'd still like to see a citation. – ShreevatsaR Dec 05 '10 at 18:55
  • Huh. It seems that most people just don't make explicit what curve they have in mind when they talk about a "steep learning curve". I wasn't able to find any descriptions by a speaker of what they themself are imagining when they use the phrase. (Well, other than my description above of what I have in mind when I use it.) (Note that Alex's link merely speculates, without any evidence, about what third parties might be thinking, which isn't a very reliable source for what those people are actually thinking.) – Henry Dec 06 '10 at 18:29
  • I should add that Alex's link does describe the type of curve I was referring to, though obviously the writer there is using steep to emphasize different features than I would. – Henry Dec 06 '10 at 18:32
  • @Henry: I referred to Alex's citation for evidence of what a steep learning curve means in its technical context (the second sentence he quoted), not for what others think. :-) My claim is that most (all?) actual learning curves that are called steep learning curves are ones in which the learning is easy. Certainly the steep part is the easy part (if you wanted to emphasize the difficulty, you'd call attention to the non-steep part), and I don't see a good reason to talk of one curve relative only to other curves which take the same total time to reach a given level of learning. – ShreevatsaR Dec 06 '10 at 18:36
  • @ShreevatsaR: I agree that Alex's citation does support the claim about the existence of that technical meaning. Regarding your final point, it only makes sense to compare learning curves after normalizing for the underlying difficulty of the task; otherwise you're comparing things with different axes. As for calling such curves steep, the fact that "steep" already has connotations of being difficult seems like a fairly good reason for using that term. – Henry Dec 06 '10 at 19:45
  • The technical meaning matches the common-use one if you just flip the axies. – user867 Jun 27 '13 at 07:53
  • @user867: Very good point! Of course, flipping the axes inverts the meaning of "steep" (things that are steep in one graph are the opposite of steep in the other), so the common-use one is the opposite of the technical. But yes, it's possible that the curve with the axes flipped may be what some people are thinking of (takes a lot of time per amount of learning, so it's hard). – ShreevatsaR Jun 27 '13 at 08:52
  • 3
    Why is the author using time and learning as x and y axis? A steep learning curve is used in the context of a pre-existing domain of knowledge and refers to the amount one needs to learn in order to be operating effectively within that domain of knowledge. In other words, if you have to know 3 facts about a domain, it doesn't have a steep learning curve; if you need to know 1000 facts it does. Time is irrelevant in this context. The correct axis should be "facts to know" and "effectiveness to reason / act in domain". Based upon this, steep learning curve is accurate. – timpone Dec 20 '13 at 14:04
  • 2
    @timpone: The term "learning curve" has a well-established meaning for a century, and you cannot change what it means. (By the way, even with your axes, "steep" means that you very rapidly gain effectiveness after just a few facts.) – ShreevatsaR Dec 21 '13 at 04:11
  • @ShreevatsaR no it doessn't, you didn't understand what I wrote. 3<>1000 . You can see things differently since this is a very subjective question but you don't understand what I'm arguing. – timpone Dec 21 '13 at 04:29
  • @timpone: Firstly it doesn't matter, because the meaning of "learning curve" is well-defined. Anyway, just to humour you, if we adopt your axes, I'm pointing that "steep" means a large increase in y-axis for a small increase in x-axis. Do you doubt this? So this means a large increase in effectiveness after a small increase in number of facts. – ShreevatsaR Dec 21 '13 at 10:01
  • 1
    Could one not just swap the axises, where x is learning and y is time. Then a steep curve would conform to the common understanding? – Peter Sep 29 '14 at 05:17
  • 1
    @Peter: The technical meaning of "learning curve" is well-established in the literature, so we can't just swap the axes without adding to the confusion. (Besides, it seems weird to have something so regularly increasing as time on the y-axis, and have the actually variable quantity (learning) on the x-axis. Usually if one of the axes is time, it's drawn on the x-axis.) – ShreevatsaR Sep 29 '14 at 06:02
  • Those definitions are not opposite. Learning quickly does not mean that the end goal is easy, just like learning slowly does not mean that the end goal is hard (my learning curve for playing trombone is very shallow indeed, but that's because I've never touched the instrument). How fast you acquire knowledge is unrelated to how much knowledge is required to achieve your end goal. – Owen Feb 26 '15 at 19:50
24

Steep learning curve means there's a lot of facts to pick up right at the beginning.

a steep learning curve

bobobobo
  • 1,767
  • 2
    These curves seems to agree better with the common use of "steep learning curve". Moreover having proficiency/difficulty relates better with a "learning" curve since proficiency is the ultimate result of learning whereas difficulty is the hurdle that needs to be overcome to obtain that proficiency (hence the hump in the second graph). However I feel that proficiency should be the dependent variable since it is the "outcome" of learning whereas effort (the inverse of difficulty) should be the independent variable. – adib Feb 18 '16 at 05:47
  • @adib if you were to put proficiency on the y-axis you would see an apparently very gentle learning curve (proficiency must increase slowly vs effort) – bobobobo Nov 23 '19 at 14:28
  • To echo @adib: While this answer nicely illustrates what people seem to be thinking when they use “steep learning curve” in the popular sense, note that “learning curve” has a well-established meaning in the technical literature that puts on the y-axis the learning/proficiency, not difficulty. Difficulty is not the outcome of proficiency so it's weird to put it on the y-axis. The Wikipedia article (current version) also has some discussion of how the popular meaning is the opposite of the technical meaning. – ShreevatsaR Nov 23 '19 at 17:10
  • 1
    @ShreevatsaR Yes, you'd like to put # Facts Learned on the x-axis and the resultant Proficiency on the y-axis. However, that curve is only steep when Proficiency is ___VERY HIGH___ for small numbers of facts learned (implying an easy skill). So I flipped the axes to accommodate the "steep learning curve" expression. – bobobobo Nov 23 '19 at 19:09
  • If you put Time on the x-axis, and # Facts To Learn on the y-axis, I suppose an activity with a "steep learning curve" has an expectation to absorb many facts over a short period of time. – bobobobo Nov 23 '19 at 19:18
16

This phrase has a scientific basis (Wikipedia has information on its origin and scientific usage), but is most commonly used to indicate that something is difficult to learn. It refers to a person’s rate of progress in learning a new skill as it might be plotted on a graph. In this case it sounds like the computer program itself is difficult for beginners to use effectively, not that it is not good at learning. I have never heard the phrase used that way, though I suppose it could apply to a program that uses artificial intelligence.

Todd Prouty
  • 1,121