Learning curves: up and down, steep and shallow

« previous post | next post »

Daniel Drezner, "Five thoughts about the firing of Rex Tillerson", WaPo 3/13/2018 [emphasis added]:

There is no signature idea or doctrine or accomplishment that Tillerson can point to as part of his legacy. He was woefully unprepared for the job on Day One and barely moved down the learning curve. His incompetence undercut his ability to advance any worthwhile policy instinct.

My reaction on reading this passage was that Drezner should have written that Tillerson "barely moved up the learning curve". As we'll see, this opposition in directional metaphors is apparently a cultural difference between psychology and economics, or maybe among a more complicated set of academic subgroups.  And while I was looking into this issue of directionality, I was reminded of another curious quirk of learning-curve metaphors that I've been meaning to write about for a while, namely the inverted meaning of "steep".

Let's start with what a "learning curve" is, and why someone might be said to move "up" or "down" one.

The Wikipedia entry for learning curve defines it as "a graphical representation of how an increase in learning (measured on the vertical axis) comes from greater experience (the horizontal axis); or how the more someone (or thing) does something, the better they get at it," and shows this schematic plot (among others) as an example:

The OED defines learning curve as "a graph showing progress in learning", and gives citations from psychologists starting in 1922:

1922   R. S. Woodworth Psychol. xiii. 307   Learning curve for the rat in the maze.
1924   R. M. Ogden tr. K. Koffka Growth of Mind iv. 168   All these facts..would naturally operate to shorten the learning-curve.
1967   J. M. Argyle Psychol. Interpersonal Behaviour x. 183   In fact some manual operatives also learn by doing, and learning curves can be plotted which show their rate of progress.
1968   H. Johannsen & A. Robertson Managem. Gloss. 74   Learning curves indicate how the rate of learning changes with increased practice and are used to predict labour productivity.

In fact psychologists were using the phrase for several decades before 1922, and always with the orientation that I've always assumed, namely that higher values of the "learning" measure are better.  Thus Edgar Swift,  "Studies in the Psychology and Physiology of Learning", The American Journal of Psychology, 1903, uses the term several times:

We shall return to these personal differences later in discussing the individual features of the learning curves.

It is altogether probable that all of the curves would in the end sweep more rapidly to the right and show a stage of slow progress as the physiological limit of skill in such matters was reached, but none of my subjects approached that limit. Bryan and Harter (6) found in their study of the acquisition of the telegraphic language a learning curve which had the rapid rise at the beginning followed by a period of retardation, and was thus convex to the vertical axis. The difference in form is very probably partly due to the difference in the type of learning involved, though it may also rest upon differences in the method of carrying out the test.

And all of the many graphical learning curves in that article show what the OED calls "progress in learning" as increases in some vertical-axis measure, so that progress involves moving up the learning curve. Here's an example from Swift 1903, where the measure of performance is the number of shorthand "lines" written in a timed test:

Swift's citation to Bryan and Harter refers to William Bryan and Noble Harter, "Studies in the physiology and psychology of the telegraphic language", Psychological Review 1897.  Bryan and Harter don't use the term "learning curve", but they focus is the process of "learning telegraphy", and they frequently reference things like "the curve of improvement in sending and receiving" and "the practice curves", with many illustrations of "practice curves" like this one:

By 1919, Louis Thurstone was writing about the general mathematical psychology of the issue, in "The learning curve equation" ,Psychological Monographs, 1919. And again, in all of his examples, the "learning curve" goes up the vertical axis as learning progresses.

But even among psychologists, there are some tasks where performance might naturally be measured in terms of error counts or error rates — and in that case, of course, "progress in learning" means that the "learning curve" goes down. Thus Hugh Blodgett and Kenneth McCutchan, "Place versus response learning in the simple T-maze", Journal of experimental psychology, 1947:

And when economists come into the picture, they naturally tend to define "progress in learning" in terms of measures of productivity, where progress is movement down, — e.g. this graph of "direct labor hours per unit" from Louis Yelle, "The learning curve: Historical review and comprehensive survey", Decision sciences 1979:

So someone who starts out unskilled, and then fails almost completely to learn from experience, can logically be said to have failed to move either "up the learning curve" or "down the learning curve", depending on whether our metaphorical learning curve is measuring benefits or costs.

But whether the learning curve is rising or falling, a steeper curve ought to mean that learning goes faster, and therefore is easier. If a task is harder to learn, or more resistant to practice, it should have a shallow learning curve (or maybe a gentle learning curve?), not a steep one.

Even in elite publications, however, a steep learning curve is taken to mean a difficult task, not necessarily a rapid rate of improvement:

[link] Trump’s aspirations in North Korea might reflect more than a bit of naive ambition and misdirected effort. Trump’s goal for the meeting — a denuclearized North Korea — is laudable. Still, is a meeting, now, the best way to accomplish it? With a turnover in key personnel, including the secretary of state, the learning curve promises to be steep.

[link] “Going online and selling online was for me hitting a brick wall. The technological aspect of selling art online was foreign to me,” said Ms. Massion, who had a steep learning curve, starting with learning how to upload digital images.

[link] As you know from the calendar, this week’s drills will focus on following rules, sportsmanship skills, and not using game time to sell people on your new diet system. Clearly, we've got a steep learning curve. We can't all have the natural ability to sit passively while our children kick a ball around on a beautiful spring day. We have to work at it!

[linkEvery president faces a steep learning curve when he enters the presidency. There is, as John F. Kennedy reportedly said, no school for commanders in chief. Yet even by that standard, recent interviews show a Donald Trump who is genuinely surprised by the size of his duties, the interests he must balance, and the methods required to get that done.

Presumably, this is because the metaphor of learning as climbing a mountain — gradus ad parnassum — is overwhelming the metaphor of documenting progress by plotting a measure of performance as a function of time. If the mountain is steeper, the task is obviously harder — though if measured progress is expected to be more rapid, the task is obviously easier.

The two metaphors can be made consistent, if initial performance is very low, but rapid improvement can be expected — so there's a lot to learn, but it's easy to learn it.  My impression is that this is not very often what people mean by talking about steep learning curves, but maybe sometimes they do.

Update, just to clarify — "a steep learning curve" has clearly become an idiomatic way to say that something is difficult to learn, not (as the technical origin of the phrase implies) that something can be learned quickly, with relatively little time and/or effort. I suspect that most people who use the currently-common meaning have no idea what a "learning curve" actually is, or in any case are not thinking about whatever they may know about learning curves when they use the phrase.

Thus in some ways this is like the "raise the question" meaning of "beg the question", with the difference that very few people are annoyed by the eggcorn-ish overlaid meaning of "steep learning curve". And I'm not trying to tell people not to use the new meaning, just explaining what has happened.



  1. FM said,

    March 14, 2018 @ 2:31 pm

    The "steep" metaphor makes sense if we think of progress as motion _forward_, which is slower when the mountain is steep. I think what people usually mean by it (well, what I've implicitly imagined when I use it) is something like "in order to advance a little in performance [the x-axis], you have to do a lot of learning [the y-axis]." Time is a third variable that's not even considered — after all, different people, or the same person with different strategies, may do different amounts of learning over the same period of time/amount of experience. So this is the only way to make this metaphorical version of the "learning curve" into an objective, non-learner-dependent metaphorical measurement. But yeah, I think that makes "steep learning curve" officially into a mixed metaphor.

    [(myl) But there's never been any technical version of a "learning curve" where the x-axis is "performance" — the horizontal axis is always some measure of "experience", such as days or trials or etc. The y axis is where the measure of "performance" goes — quantity of stuff accomplished, percent correct, error rate, etc.]

  2. Gwen Katz said,

    March 14, 2018 @ 2:34 pm

    Don't we sometimes use "move down" to mean "move along?" For instance, "go down the road" doesn't imply that the road itself descends. So by "move down the learning curve," could they just mean "make progress along the learning curve"?

    [(myl) But I don't think this applies when there's an obvious vertical gradient — you wouldn't say "down the hill" to mean higher on the hill then you are now, or "up the hill" to mean lower. And I think it would be weird to say "take route 70 from Georgetown CO (at 8500 ft) down to the Continental Divide (at 11,150 ft)".

    In more or less horizontal travel, there's a whole thing about what directions are "up" or "down" — see "down east", "down south/up north", "uptown" vs. "downtown", etc. And COCA has 1,086 instances of "up the road", along with 6,831 instances of "down the road".]

  3. Philip Taylor said,

    March 14, 2018 @ 3:15 pm

    Gwen K's analysis corresponds exactly to my own.

  4. Brian said,

    March 14, 2018 @ 3:41 pm

    To build upon what FM said, the idea of a learning curve being "steep" is that it's a reference to future requirements, not past progress. Your learning curve will need to be steep in order to pass this class, beat this game, etc. But I agree that it does seem to wind up getting confused with climbing metaphors.

    And I suspect that Gwen Katz's explanation is how the original statement "move down the learning curve" originated, but it still sounds completely wrong to my ears.

  5. Ethan said,

    March 14, 2018 @ 3:48 pm

    I read the original Drezner quote as being a snarky way of saying "he was bad at the start and got worse from there".

  6. David Morris said,

    March 14, 2018 @ 3:48 pm

    My intuition was the 'up the road' is just as common, but Google Ngrams does not support me. Even if 'down the road' doesn't imply that the road itself descends, it would be strange (to me) for someone to say 'down the road' when the road actually went up. (By the way, the Ngram for 'up the road', down the road, up the street, down the street' is one of the most interested I've seen.)

    My other intuition was that 'up the learning curve' was more common than 'down the learning curve'. Ngrams shows that 'down' was more common than 'up' from the late 1970s to the late 1990s, but that 'up' has been more common since (with a slight tick up for 'down' from 2005 to 2008).

  7. David L said,

    March 14, 2018 @ 4:19 pm

    I think of productivity as being intuitively measured in terms of amount of stuff produced per unit time, rather than time needed to produce one unit of stuff, so the graphical example you give seems back to front.

    The point about a steep learning curve is fascinating. The simple-minded metaphor in my mind is that a steep learning curve is like a steep hill — it takes a lot of effort to go a short distance. But if the vertical axis is "amount learned" and the horizontal axis is "effort expended," then you're right, a steep learning curve means you acquire a lot of knowledge for a modest outlay of effort.

    Mind blown, as they say.

  8. Jonathan Smith said,

    March 14, 2018 @ 5:02 pm

    I agree down = along as in down/along the positive x-axis.
    LL's own space/time metaphor conundrum seems to be that from the front page, "<>" (right) both lead to p. 2… with the four choices offered on pp. 2+ a still thicker cognitive stew :D

  9. Jonathan Smith said,

    March 14, 2018 @ 5:03 pm

    from the front page, "previous entries" (left side of page) and "next page" (right) both lead to p. 2…

  10. 번하드 said,

    March 14, 2018 @ 5:18 pm

    I still remember that when I first saw the LL style navigation, it struck me as "ah, finally somebody got it right!" :)
    One could also argue for up/down buttons, seeing the posts stacked with the most recent one on top, in extension of the overview pages' arrangement.
    And I seem to remember that time's arrow can point in different directions in different cultures.

    [(myl) See "Time is space: When fronter is farther behind", 10/16/2004.]

  11. Luc said,

    March 14, 2018 @ 5:52 pm

    What Ethan said. Drezner has been rather vocally anti-Tillerson, and can be rather snarky, so the "Tillerson got worse" explanation is valid.

    That being said, we could just ask him, he's a rather active Twitterer.

  12. Rubrick said,

    March 14, 2018 @ 6:50 pm

    You have an unbalanced tag at the end of your update. Not surprising; tag usage in WordPress has a broad learning curve…

  13. Rubrick said,

    March 14, 2018 @ 6:52 pm

    Heh. WordPress swallowed my embedding of [less-than slash "font" greater-than] in my comment, which probably proves some kind of point.

  14. Ralph Hickok said,

    March 14, 2018 @ 8:27 pm

    Years ago, I read about a "U-shaped learning curve," purportedly showing how children learn language. IIRC, the gist of it was that children first learn irregular verb forms by hearing speech. Then, as they learn rules of grammar, their use of language actually gets worse for a time. For example, a child learns to say "I heard" but, after learning how the past tense is usually formed, begins to say "I heared."

  15. Ray said,

    March 14, 2018 @ 9:42 pm

    this whole thing is like how when people say that a meeting time has been "pushed back" or "moved up" — I never know what they mean!

  16. CPHuxley said,

    March 15, 2018 @ 12:29 am

    I agree with Brian. For me, a steep learning curve makes sense when the y-axis represents the level of competence that is required by a certain point in time along the x-axis.

  17. Scott P. said,

    March 15, 2018 @ 12:47 am

    My mental 'learning curve' has time as the x axis and 'amount needed to master' as the y axis. A shallow learning curve means you don't need to learn very rapidly. A steep learning curve means you do.

  18. Andreas Johansson said,

    March 15, 2018 @ 1:28 am


    For that reason, at work we speak of meetings being moved "right" (later in time) or "left" (earlier in time). The metaphor, of course, is a calendar where dates increase towards the right.

  19. Rhona Fenwick said,

    March 15, 2018 @ 3:04 am

    In more or less horizontal travel, there's a whole thing about what directions are "up" or "down" — see "down east", "down south/up north", "uptown" vs. "downtown", etc.

    Where I live there's frequently an actual third dimension involved in speaking of "going up to" or "going down to" a place (living near the east coast on the eastern side of the Great Dividing Range, as the majority of Australians do, west is nearly as often "up" as north is, east nearly as "down" as south is), and it's often fascinating to see how people choose their directions, especially when someone's heading both west and south (east and north is rarely possible, particularly in Queensland).

    Also, your mention of "down east" made me smile. It reminds me of Daffy Duck (or, more specifically, Duck Dodgers in the 24th-and-a-Half Century!) trying to baffle Porky Pig with three-dimensional navigation, including a reference to the direction "south by down-east".

    Here's the relevant clip.

  20. Bob Ladd said,

    March 15, 2018 @ 3:33 am

    Having taken a first-year psychology course in the heyday of behaviourism, I've always had the vague feeling that something wasn't quite right about the standard metaphorical meaning of "steep learning curve", but never thought more about it than that. Thanks for looking at this, Mark!

  21. Bob Ladd said,

    March 15, 2018 @ 3:38 am

    With regard to words for more or less horizontal travel, residents of Edinburgh and Glasgow go "through" to the other city. I think this might be more generally true in Scotland but I'm not sure of the details.
    And more generally, Brits tend to go "out" to foreign countries – my impression is that the likelihood they will use "out" increases with geographical and cultural difference.

  22. rosie said,

    March 15, 2018 @ 4:00 am

    > when I first saw the LL style navigation, it struck me as "ah, finally somebody got it right!"

    LL's navigation misleads me at least half the time. Your "previous post" link links to the next post. Granted, that one is earlier in time, but the links are for navigating through the sequence of posts, and in relation to that sequence the words "previous" and "next" are the wrong way around. The more usual convention is
    Prev 1 2 3 … Next
    where "Next" is next in sequence, not next in time.

  23. Jen in Edinburgh said,

    March 15, 2018 @ 4:48 am

    I'd never thought of 'through to Glasgow' as unusual. (I think I'd go 'through' to the rest of the Central Belt, but 'over' to the southern west coast.)

    There's a dividing line somewhere south of here where people stop going down to London and start going up. Friends in Newcastle still go down, but I've had people from further south argue that London is *always* up, no matter where you start from.

    'Going out to' makes me think of emigrants to Commonwealth countries – if someone said it now, I think it would still imply to me that they were going there to work for a fairly substantial period of time, if not indefinitely. (I'm willing to be proved wrong by examples, though.)

  24. Ursa Major said,

    March 15, 2018 @ 7:13 am

    But errors/experience isn't a learning curve, it's an error curve. And since errors is inversely proportional to learning, while the error curve goes down the learning curve always goes up…

  25. MattF said,

    March 15, 2018 @ 8:41 am

    I guess the 'steepness' of a learning curve is, like 'curvature', an intrinsic property of the curve. Which is to say, one thinks of the learner as a mobile point on the curve, and consider how much effort it takes for the point to move forward– kind of like the amount of energy gained or expended in roller coaster ride as the vehicle moves along the track.

  26. KevinM said,

    March 15, 2018 @ 8:53 am

    Yes, another glitch in the metaphor (or the meta-metaphor). The idea that the curve exists, and that a person travels along it, is not quite right. You create it as you go.

  27. Scott P. said,

    March 15, 2018 @ 9:33 am

    The idea that the curve exists, and that a person travels along it, is not quite right. You create it as you go.

    As an educator, I can attest that the curve exists prior to the student encountering it.

  28. Vulcan With a Mullet said,

    March 15, 2018 @ 9:41 am

    In a blues song, you can't go UP the road feeling bad :)

  29. Robert Coren said,

    March 15, 2018 @ 9:52 am

    My idea of what a "learning curve" matches @Scott P's. In terms of actual graphic curves, the "steepness" doesn't necessarily describe the actual progress, but what the curve needs to look like in order to succeed.

  30. H Stephen Straight said,

    March 15, 2018 @ 10:06 am

    After years of perplexed uncertainty like that described by Mark I finally realized that the steepness of a learning curve could indicate the amount of effort the learner devotes to learning as opposed to merely doing a new activity in order to achieve the minimum required level of proficiency. When, for example, a dean appointed with little college-wide experience says that they encountered a steep learning curve in their first months in office, that means they had to expend hours of study and thought to attain the knowledge and skills necessary to avoid ignorant or clumsy mistakes. A steep curve characterizes faster but not necessarily easier learning if the time dimension includes all the learner’s time rather than just the portion spent on learning, which increases with the urgency of becoming proficient.

  31. J.W. Brewer said,

    March 15, 2018 @ 11:18 am

    As a possible blues counterexample to VwaM's proposition, I offer "Farther Up the Road" (recorded 1957 by Bobby "Blue" Bland): https://www.youtube.com/watch?v=QTkJO_qq7hM It *may* be relevant (if someone is trying to figure out a pattern here rather than assuming random free variation) that the person who is going to be "up the road" rather than "down the road" is not the first-person narrator but the woman who done him wrong.

  32. BZ said,

    March 15, 2018 @ 1:38 pm

    Honestly I've never heard about moving along/up/down the learning curve before. It's an odd concept to me. Honestly, a learning curve is just a set phrase to me. Never thought of an actual physical chart. As for steep learning curve? Do people talk about steep curves in mathematics? I don't think I've encountered this turn of phrase despite minoring in math. I found one relevant example googling "steep curve -learning" and it's for something called a "steep yield curve". So even with a real mathematical curve, "steep" is not often used to describe the slope, so the word is available for another meaning in a metaphorical learning curve.

  33. KevinM said,

    March 15, 2018 @ 2:37 pm

    Oh, @Vulcan, how could you? I refute you thus:
    Bobby Blue Bland, "Further On Up The Road"

  34. KevinM said,

    March 15, 2018 @ 2:38 pm

    Oh Jeez, didn't see JW Brewer. Sorry for the redundancy.

  35. A1987dM said,

    March 15, 2018 @ 3:33 pm

    The modern meaning of "steep learning curve" makes sense if you transpose it, i.e. you put skill levels on the x axis and amounts of effort needed to achieve them on the y axis.

  36. Ray said,

    March 15, 2018 @ 3:42 pm

    @Andreas Johansson — what a great idea! to move a date right or left. I'll suggest this to my co-workers (though it might present a steep learning curve!)

    @MattF — yes, that is how I've always perceived the phrase: being on a steep hill, climbing hard, because the learner's trying to pack in a lot of new information in a short amount of time, and not in the usual, organic way. so the learning curve is steep, hard won. also, the phrase "steep learning curve" is often used along with "to get up to speed"; at some point the learner figures stuff out and can match everyone else's pace in performance, the curve levels out…

  37. Daniel Barkalow said,

    March 15, 2018 @ 4:23 pm

    I think people are consistent that a "steep learning curve" means "you'll know a whole lot more after a day of experience than you did initially", and you'll be much more effective at the task in minute 481 than minute 1. But does this contrast with "during minute 1, you perform nearly perfectly, and have no need to improve" or "during minute 481, you still can't get anything done"?

    There's also the idea that learning comes with effort, and a steep learning curve means that you're putting a lot of effort into learning rather than into performing the task, whereas a shallow learning curve means that your effort goes mostly into producing results rather than learning.

    There's definitely the perception that, if something has a steep learning curve, such that people improve greatly with lot much experience, they're also more likely to give up with very little experience, and there's the possibility that survivorship bias is affecting these results, where everyone who doesn't improve rapidly doesn't manage to get much experience.

  38. Ethan said,

    March 15, 2018 @ 4:49 pm

    @Daniel Barkalaw: That doesn't match my impression of common usage. I would interpret a claimed steep learning curve as corresponding to your "still can't get anything done at minute 481" and no expectation that you will gain much competence in the first hours/days/months.

  39. peter said,

    March 15, 2018 @ 4:58 pm

    Many of the uses in business of the term “learning curve” are references to the costs of doing some new activity, which typically fall as experience is gained and as learning increases. Going down the learning curve thus refers to falling costs over time.

  40. Roscoe said,

    March 15, 2018 @ 5:42 pm

    "Here I go down Circle Road…"


  41. D.O. said,

    March 15, 2018 @ 7:26 pm

    More logical would be speaking of a "compressed learning curve" (you have to do your learning in a shorter time than normal), but language is a great put downer for everyone too full of themselves.

  42. James Wimberley said,

    March 15, 2018 @ 7:46 pm

    I suspect that the odd BrE use of up and down in relation to London goes back to rail. The network was built radially from the capital. Main lines were double-tracked, and the two lines labelled up and down for operating clarity.

  43. Graham Katz said,

    March 15, 2018 @ 10:05 pm

    A relate thought: In Joshua Rodman's entertaining New Yorker article about paper jams https://www.newyorker.com/magazine/2018/02/12/why-paper-jams-persist he makes a mistake describing the Bernoulli effect (since corrected online but noted in the letters https://www.newyorker.com/magazine/2018/02/26/letters-from-the-february-26-2018-issue ), writing that air moves more quickly over the flat part of a wing than over the curved part. I wondered when I read this [after getting over the shock that such an error could make it through editing and fact checking, etc.] whether the source of the error was in a similar human motion metaphor: we move more quickly (by foot, car, bike or ski) in straight lines than in curves (just as we move more slowly up steep hills).

  44. Gwen Katz said,

    March 16, 2018 @ 1:28 am

    To build upon what FM said, the idea of a learning curve being "steep" is that it's a reference to future requirements, not past progress. Your learning curve will need to be steep in order to pass this class, beat this game, etc. But I agree that it does seem to wind up getting confused with climbing metaphors.

    This makes particular sense with regard to games, where the X axis might be levels and the Y axis might be the skill needed to complete that level. So a game where the first level is "put the box on the button" would have a shallow learning curve, whereas a game where you have to learn 47 different combos in order to get through the first level would indeed have a steep learning curve.

    The trick is that in this case the X is actually time-independent; if it took you 20 hours to finish the first level, the graph doesn't show that.

  45. Nick Barnes said,

    March 16, 2018 @ 5:54 am

    'Up', of course, means 'towards Cambridge'. 'Down' means 'away from Cambridge'. Ironic given the low elevations hereabouts.

  46. Nick Barnes said,

    March 16, 2018 @ 6:01 am

    With regard to learning curves: I agree with several other commenters in that my mental model has some sort of achievement or progress along the X axis, and some sort of effort along the Y axis. Very complex skills (such as emacs, or violin) tend to have steep curves in that new learners must put in a lot of effort, and experience some frustration and bafflement, to achieve basic results.

  47. Philip Taylor said,

    March 16, 2018 @ 6:22 am

    Nick B: "Up" to any ancient seat of learning; one is, after all, "sent down", no matter at which alma mater one formerly one studied …

  48. Ralph Hickok said,

    March 16, 2018 @ 8:02 am

    @Graham Katz:
    Yes, I noticed that error. A letter appeared in response to that article which I assumed was going to correct the error, but it didn't.

  49. Robert Coren said,

    March 16, 2018 @ 9:59 am

    What @James Wimberley said makes sense to me.As a Northeastern US native, I always think of going "up" to the country and "down" to "town", and was taken aback when I started reading British novels and encountered the reverse (and in contexts where "town" always meant London, which was likely to be south of wherever the country house was).

    And then there's being "sent down" from University, which seems to be a separate thing.

    As to "uptown" and "downtown", I grew up in Manhattan and the Bronx, where "downtown" always means "south"; but I think in the general case it means "at or towards the central business district".

  50. Rodger C said,

    March 16, 2018 @ 11:12 am

    When I was growing up just downriver from Huntington, WV, the business district of Huntington was "uptown." Later, coming down from the hills to West Huntington, you'd be greeted by a big sign, which annoyed my father, pointing up-river and reading "DOWNTOWN."

  51. Terry Hunt said,

    March 16, 2018 @ 1:49 pm

    Further to James Wimberley, there is (or was) a supposition in railway history circles that this terminology originated with some of the earliest rail lines to be converted from horse to steam power, in the coal fields of North-East England.

    The main purpose of these lines was to convey coal from the mines down to the geographically lower staithes on nearby rivers and canals for onward transportation by barge, and the empty wagons back up to the mines, which were also the 'headquarters' of the companies concerned.

    When the utility of steam-drawn trains became evident further lines, initially independent and not linked up, were built for multiple uses, and "up" and "down" came to mean "towards" and "away from" the headquarters of the rail company in question.

    In time the separate lines became linked and developed into a network (as railway pioneer George Stephenson forsaw, stipulating that all his lines should be built to the same "standard" gauge as those first mine wagon ways), and the headquarters of most railway companies shifted to London, so people eventually came to speak of going up to, or coming down from, London even if train travel was not involved.

  52. Jerry Friedman said,

    March 16, 2018 @ 10:35 pm

    Ray: @Andreas Johansson — what a great idea! to move a date right or left. I'll suggest this to my co-workers (though it might present a steep learning curve!)

    As an extreme step, people could say "have the meeting sooner", or "have it later" or "put it off" or "postpone it". But I'm sometimes accused of being too literal.

  53. Rodger C said,

    March 17, 2018 @ 11:22 am

    Maybe it's time for the anglosphere to generalize the Indo-English "prepone."

  54. outeast said,

    March 18, 2018 @ 9:22 am

    References to steep vs gentle learning curves are commonplace in gaming reviews. In such contexts it's definitely about the speed at which you *have to* master skills in order to progress. That is, it's required by the learning context not dependent on the gamer's learning ability: a slower learner cannot take a more gentle slope, not can a fast learner bypass the training wheels phase. In extreme cases new skills/abilities have to be unlocked by progressing through missions/levels, making the learning curve even more independent of the learner.

  55. 번하드 said,

    March 18, 2018 @ 10:26 am

    @Rodger C:

    Yes, you got my voice for the preposition of meetings, it's an overdue enrichment.

  56. 번하드 said,

    March 18, 2018 @ 10:42 am

    Which should have been preponement, of course. Sorry for that, lack of caffeine. BTW, other national dialects of English have riches on offer, too. I hereby nominate "eye shopping" and "skinship", both made in Korea.

  57. Chris said,

    March 19, 2018 @ 9:25 am

    I always thought the Y-axis was "ignorance". On the other hand, I envisioned it as a steep concave up increasing graph, which doesn't make sense.

    However, if the Y-axis were ignorance and the graph concave up decreasing, it would make sense to say "move down" the curve.

RSS feed for comments on this post · TrackBack URI

Leave a Comment