learning curve love


from The American Heritage® Dictionary of the English Language, 4th Edition

  • n. A graph that depicts rate of learning, especially a graph of progress in the mastery of a skill against the time required for such mastery.

from Wiktionary, Creative Commons Attribution/Share-Alike License

  • n. An experience that teaches a lot to someone.

from WordNet 3.0 Copyright 2006 by Princeton University. All rights reserved.

  • n. a graph showing the rate of learning (especially a graph showing the amount recalled as a function of the number of attempts to recall)


Sorry, no etymologies found.


    Sorry, no example sentences found.


Log in or sign up to get involved in the conversation. It's quick and easy.

  • I'm appalled by this page. It's Wordie's dirty little seret.

    April 14, 2008

  • My comments about I could care less moved to that page.

    April 14, 2008

  • I'm going to echo yarb's comment on dork out:

    Lots of conversations on Wordie are irrelevant to the word on which they reside, but not this one.

    April 13, 2008

  • Mentally refiling it, moving it over to the cabinet containing parameter and exponentially, the one for mathematical terms which have a highly specific meaning on their home turf, but end up being used very imprecisely (as a mathematician, I'd say sloppily) in general conversation.

    Words of this kind are always likely to bother mathematicians, who are conditioned to interpret such terms only in the narrowest definitional sense.

    From a strict mathematical perspective, the phrase steep learning curve is a bit problematic, as it appears to suggest that steepness is a global feature of the curve, when in fact steepness is an intrinsically local property with magnitude that of the local derivative at any point on the axis.

    OK, I'll stop now. Even I can't take it any more.

    April 13, 2008

  • I don't disagree with frindley or sionnach about how learning curves should be graphed in an academic context, or that the popular meaning of "steep learning curve" is the opposite of what would be expected in that context. I'm arguing that "steep learning curve" is not an example of a phrase based on a widespread misconception.

    Having now waded through many examples on Google Books, I can say that "steep learning curve" was in use by the mid 1960s, and in the 1960s and 1970s had its current meaning. I found no evidence of a shift to an incorrect meaning, and no evidence that users of the phrase were referring to graphs of the academic sort. The phrase seems to have been intended metaphorically from the start, and can be thought of as showing cumulative effort versus time, even though it would in practice be difficult to produce such a graph from quantitative data.

    April 13, 2008

  • I agree with frindley 100%, and disagree with what I believe mollusque is saying. There are three variables floating around here: time, effort expended, and actual mastery of the material. We might reasonably expect some positive correlation between any pair of these variables, but none can serve as an exact proxy for another (IMO). Except perhaps one might agree to take the time spent learning as a direct measure of expended effort (in which case the two would follow an exact linear relationship, but the slope of that relationship is not informative, as it is just a conversion factor in the same way that 2.54 converts inches to cm). I think mastery of the material has to be considered the primary result of interest, and thus has to be plotted on the y-axis.

    And I wish that www.dotnetcharting.com would take their little google ad and go jump in a lake.

    April 12, 2008

  • Plotting accumulated effort (or accumulated frustration or pterodactyl's stress levels) would certainly yield a graph where the difficult task has the steepest curve. But does it then cease to be a learning curve? – given that "learning" is meant to be about the accumulation of expertise rather than the accumulation of ulcers.

    I'm going to go find my Latin dictionary now.

    April 12, 2008

  • OK, now my brain is bleeding. Solam Britannicam relinquite!

    April 12, 2008

  • Frindley, I meant cumulative effort on the y-axis. "Steep learning curve" means that someone is working harder. Someone who spends an hour trying to learn Access is likely to feel more frustrated and feel that they worked harder than someone who spent an hour trying to learn FileMaker. Hence my mountain climbing analogy.

    April 12, 2008

  • Bernice just called. It's a jungle out there, and she would really appreciate some backup.

    April 12, 2008

  • I may be misinterpreting: effort over time still gives me a steeper curve for the easier option.

    Knowing me I'll be spending sleepless nights trying to think of a variable that, when plotted over time, will yield a steeper curve for the more difficult option. There must be something!

    April 12, 2008

  • Spot on: strip appreciated very much. The next frame even better!

    April 12, 2008

  • I think the graph that corresponds to the normal meaning of "steep learning curve" is time on the x-axis and effort on the y-axis. I agree that that's not what is meant by "learning curve" in academic circles.

    April 12, 2008

  • What I learned here, so far, is:

    - a new expression for ter's list (I could care less) that he typed and didn't notice (sic!)

    - that frin and c_b could like this strip

    - that my learning curve is, however you want to call it, the bad one.

    April 12, 2008

  • Before the moment passes, I would just like to assure yarb that his flashes of brilliance do not pass unnoticed, and are greatly appreciated.

    Not that all the rest of y'all aren't brilliant as well.

    April 12, 2008

  • But in summary: it does seem that all of us are agreed that:

    (a) in daily use, "steep learning curve" suggests difficulty; and

    (b) that this is because we hear the word steep and associate it with physical gradients and therefore challenges and difficulty.

    It is a natural shift of meaning that has occurred because the proportion of cognitive psychologists in the language-using population is relatively low.

    What I have learned today is that the man responsible for all this (Hermann Ebbinghaus) also developed a forgetting curve. Shall we head over there for more fun?

    April 12, 2008

  • If time is reintroduced to the databases example, it still demonstrates that one should avoid Microsoft at all costs, because at that crucial point 1 week into the learning process, expertise is considerably lower for Access than for Filemaker. That wouldn't be possible unless Filemaker had the steeper learning curve.

    April 12, 2008

  • mollusque's example is interesting. The initial stipulation that "You have to pick one, and learn enough to be productive within a week" effectively takes time out of the equation (and therefore the graph), because both things (in the example Filemaker and Access) are to be assessed within a common time frame and the assumption is equal learning time available to be devoted to each option.

    The two axes for mollusque's curve are, therefore:

    X - knowledge needed (it could be argued this equates with effort expended) ranging from none to a lot; and

    Y - ability to be basically productive with the tool in the specified time frame (so not necessarily final expertise) ranging from not ready to good-enough-to-go.

    When I plot Filemaker and Access on this graph the Filemaker curve is steeper than the Access curve. Or one could flip the axes and then Access would have the steeper curve.

    But, and here's the rub, this isn't (I don't think…) what is meant by a learning curve. It's a smart way to think about what tool to pick when you have a goal to achieve in a specified time frame (especially relevant for databases and for website design tools), but it's ultimately about assessing the tool, not the learning process.

    That's why I still think that the graph I linked to in an earlier comment (the one that plots expertise achieved over time/repetitions spent) is a better example of a learning curve.

    April 12, 2008

  • Apologies for misgendering you, frindley!

    April 12, 2008

  • frindley doesn't like to brag about being female, for she is convinced that this is behind her rapid surmounting of all the steep learning curves she has ever come across. As for her occasional struggles with the flat ones? – a measly trajectory can always be blamed on someone or something else!

    April 12, 2008

  • I think frindley might be female... though I don't know if s/he would like me mentioning it...

    April 12, 2008

  • Australians generally use 'I couldn't care less' rather than its Logic-Lite cousin.

    For a more manageable language, perhaps try the Running of the Sheep.

    April 12, 2008

  • Running the language is like running the bulls at Pamplona.

    April 12, 2008

  • You're so perspticacious, pterry! Mollusque: I think you're right, that most people think in terms of a physical gradient, rather than a graph.

    So, would anyone like to summarize our learnings from this discussion? frindley?

    evil cackle .....

    April 12, 2008

  • I'm going to agree with sionnach and disagree (respectfully) with frindley.

    Regarding sionnach's point about black and white: There's an excellent example of this in the phrase "I could care less", which is an idiom that means its exact opposite ("I couldn't care less"). Everyone seems to understand this phrase just fine, even the people who hate it (like me).

    Regarding frindley's original point, down at the bottom of the page: Imagine that two people (let's call them "Aloysius" and "Bernice") have just started new jobs. Aloysius is going to be a crate stacker. He's given three hours of training, and at the end of it, he's expected to be proficient at stacking crates.

    Bernice is going to be an emergency dispatcher. She's given three hours of training, and at the end of it, she's expected to be proficient at dispatching EMTs, police, and firefighters.

    Bernice's training is going to be a heck of a lot more stressful than Aloysius's, because she has to learn a harder skill in the same amount of time. In other words, she has a steeper learning curve -- and this is not a good thing for her.

    April 12, 2008

  • Okay, let's see if I can explain better. A steep learning curve must be steep in relation to something, presumably a different learning curve. I chose software packages to illustrate. I've heard it said, for example, that Microsoft Access has a steeper learning curve than FileMaker Pro (another database application).

    You don't need to learn as much in Filemaker to start using it effectively as you do in Access. But I guess that means I'm not using the same Y-axis as frindley specified. He made it "progress", I'm picturing it as "learning" or "knowledge". Access requires you to learn more to be competent than Filemaker does.

    I think when people say "steep learning curve" the analogy is more to a physical path than to a graph. You can walk straight up the mountainside or take the shallower path that loops around the side. If you graph progress (time to altitude), the steeper path might take less time, but most people prefer the shallower path, because they are not willing to work that hard.

    April 12, 2008

  • Yeah! Who's running this language, anyway?

    April 12, 2008

  • Sionnach: what makes you Saussure about that?

    April 11, 2008

  • All of this raises an interesting question. If one believes (and I do) that most people assign a meaning to the phrase 'steep learning curve' that is the exact opposite of what it actually means, why does it survive at all? It would seem to be a completely unreliable vector for transferring information.

    My hypothesis is that the word 'completely' is key in allowing this term to survive. That is, if everyone interprets it incorrectly in the same way, this eliminates the potential for ambiguity. So that a speaker who wants to say something is hard to learn can say it has a steep learning curve, secure in the knowledge that everyone will give it the logically incorrect but intended interpretation, thereby avoiding a misunderstanding.

    This is as if everyone agreed that, from now on, the meanings of the words 'black' and 'white' would be swapped. In theory, no ambiguity, provided everyone agrees to the same convention. But in practice, it's no way to run a language.

    April 11, 2008

  • But my toboggan says the steep learning curve is more fun.

    April 11, 2008

  • There were no learning curves that I could find, but over at Indexed Jessica Hagy has some great graphs and diagrams.

    April 11, 2008

  • Here be diagrams

    In the above example the X axis is reps rather than time, but since repetition occurs over time, it's the same thing. The Y axis is defined as expertise, what I would call progress.

    Of course, the common misconception is natural enough, because we're inclined to associate steepness with difficulty.

    April 11, 2008

  • mollusque: Our brains must be wired very differently indeed. Because I've read what you wrote - three times now - and I still have no idea what you could possibly mean. Not trying to dis you, just really not understanding what you mean.

    Specifically, what's the distinction between 'progress' and 'being productive'? Also, your introduction of 'software packages' seems to introduce an extraneous floating variable into whatever relationship the learning curve purports to describe.

    April 11, 2008

  • I think the convention is correct. Say you're comparing two software packages. You have to pick one, and learn enough to be productive within a week. Choose the product with the shallower learning curve; not as much progress is required to be productive.

    April 11, 2008

  • I used this once and got the retort 'More like a learning cliff'.

    April 11, 2008

  • I think what frindley might be saying is that rocket science isn't brain surgery, Asa.

    But I agree that there is something very odd about the whole 'steep learning curve' concept. It's peculiar that such a widely encountered expression, which is almost universally* taken to refer to something particularly difficult to master, should be based on such a confusing graph.

    frindley's observation is astute - it is the steepness of the inverse curve that is really being referred to in the standard interpretation of this phrase.

    *: based on three out of three respondents in my completely unscientific poll

    April 11, 2008

  • This just blew my mind. You're saying rocket science isn't rocket science, then?

    April 11, 2008

  • Counterintuitive it may seem, but the steep learning curve – assuming that the curve is agreed to be a graph of progress (Y axis) and time (X axis) – means that the thing has been very easy to learn indeed, as progress occurs rapidly.

    However common usage implies that a steep learning curve represents difficulty and great challenges. The only way this would be possible is if time were plotted on the (Y axis), which would go against all graphing conventions.

    April 11, 2008