Fog Creek Software
Discussion Board




(mis)use of the term "Learning Curve"

Why doesn’t anyone use the term “learning curve” correctly?  Almost everyone you hear uses it in the complete opposite sense of what I’m sure they believe they are saying.  You know what I’m talking about, someone says that “boy, 3D Studio Max sure does have a steep learning curve, but once you get the hang of it, it’s awesome”.

The description of the learning curve should have more to do with relative level of mastery versus time spent performing the task.  The learning curve of Tic-Tac-Toe is so steep that mastery is achieved quite quickly.  Checkers mastery is achieved over some time and chess mastery may never be achieved.

Let me offer the alternate examples which demonstrate the technically correct usage:

“Tic-Tac-Toe has an extremely steep learning curve.”
“Checkers has a moderate learning curve.”
“Chess has a very gradual learning curve.”
“The game of Golf has a very gradual learning curve.”
“The game of Cup-with-ball-on-string has a steep learning curve.”

It is possible that people have modified the definition to indicate the amount required to learn in the amount of time given, but the misuse still kind of amuses me.

Talk amongst yourselves…

cheeto
Monday, April 21, 2003

Wow. I'd never thought of it that way. I suspect the term "learning curve" now has an additional definition, meaning "level of effort required to achieve basic useful knowledge." Thus "steep" conveys the idea that there's a significant amount of effort required to "climb" it.

I don't think anyone uses "shallow learning curve" any more.

Philo

Philo
Monday, April 21, 2003

Part of the beauty of the term "steep learning curve" is that it conjures in the mind the idea of scaling a high wall as a metaphor for learning a difficult task.

As phrases go, it's not as bothersome as, say, "head over heels", or "I could care less" ...

If it bothers your analytical sensabilities, picture the curve with the axes flipped.

Alyosha`
Monday, April 21, 2003

Isn't this also the case with the 80/20 rule? That's another one that conjures a lot of great images, but the original meaning was completely different from how most people use it.

I'm aware of the Learning Curve mis-usage. It's not nearly as annoying to me as Peruse.

pe·ruse    ( P )  Pronunciation Key  (p-rz)
tr.v. pe·rused, pe·rus·ing, pe·rus·es
To read or examine, typically with great care.


--------------------------------------------------------------------------------
[Middle English perusen, to use up  : Latin per-, per- + Middle English usen, to use; see use.]
--------------------------------------------------------------------------------
pe·rusa·ble adj.
pe·rusal n.
pe·ruser n.
Usage Note: Peruse has long meant “to read thoroughly” and is often used loosely when one could use the word read instead. Sometimes people use it to mean “to glance over, skim,” as in I only had a moment to peruse the manual quickly, but this usage is widely considered an error. Sixty-six percent of the Usage Panel finds it unacceptable.

- from Dictionary.com

www.marktaw.com
Monday, April 21, 2003

“The game of Cup-with-ball-on-string has a steep learning curve.”


man i must be a complete 'tard

apw
Monday, April 21, 2003

So thaaat explains it!

When Jesse Liberty wrote the book "Learn C++ in 10 Minutes" he must have heard that C++ had a steep learning curve and got confused.

Nick
Monday, April 21, 2003

A Learning Curve plots time spent on the horizontal axis and ability on the vertical axis. So if it is almost flat, you put in a lot of time to get a small increase in ability.  If it is steep, a small investment in time results in a large increase in ability.  So "steep learning curve" really means something is easy to learn, although people often mean the opposite when they use that phrase.

T. Norman
Monday, April 21, 2003

This has been a pet peeve of mine forever!!!  Thanks, cheeto.  I glad someone finally mentioned it.

commanderSpock
Monday, April 21, 2003

This question comes up fairly often on alt.usage.english.  See:

http://tinyurl.com/a090

(The link re-directs to Google Groups.)

Alex Chernavsky
Monday, April 21, 2003

T. Norman wrote:  "A Learning Curve plots time spent on the horizontal axis and ability on the vertical axis. "

Reference?

Pavlov's dog
Monday, April 21, 2003

"I could care less"

In my experience only Americans seem to have this problem. People from every other nation seem to be able to remember the correct form "I couldn't care less".

I have no idea why this is.

And the horse you rode in on
Monday, April 21, 2003

you dog, I think it is universally accepted convention to plot time on the x axis.

tapiwa
Tuesday, April 22, 2003

that was supposed to be "Yo dog"

tapiwa
Tuesday, April 22, 2003

I think the typical usage of learning curve is correct, and they are not referring to a straight line effort versus.

Rather it is a log(x) curve, and 'steep' is referring to the inital climb, not the full extent of the graph.

For example, wordperfect had a steep learning curve.  Before you could really use it you had to lean all the keyboard control combinations. I spent hours referring to f5 for the first few weeks.

Word's learning curve was not steep, because you had all these icons and menus just begging to be used.

After the inital climb, the path to mastery is a completely different issue.

So, tic tac toe has a shallow learning curve.  Somebody shows you the rules, and your away in seconds.

Chess has a steeper learning curve, before you can play you have to learn all the different pieces and what they can do.

Ged Byrne
Tuesday, April 22, 2003

Here is the actual formula:

http://www.jsc.nasa.gov/bu2/PCEHHTML/pceh.htm

Ged Byrne
Tuesday, April 22, 2003

I always thought of it as plotting effort (y) against result (x). Something with a steep rise requires lots of effort at that point (usually the start) before you have any results. Something with a more flat learning curve gives you a lott of result without much learning efforts.

Just me (Sir to you)
Tuesday, April 22, 2003

tapiwa wrote: "you dog, I think it is universally accepted convention to plot time on the x axis. "

Maybe, but I was hoping that someone could actually point me to a definition / discussion of learning curves -- you know, an article or a book chapter or something that deals with learning curves as they are used in the wild (or are they used at all?  Maybe there's no such thing as a learning curve.)

Pavlov's dog
Tuesday, April 22, 2003

http://wombat.doc.ic.ac.uk/foldoc/foldoc.cgi?learning+curve
http://www.jsc.nasa.gov/bu2/learn.html

Just me (Sir to you)
Tuesday, April 22, 2003

I don't think we are ever likely to reverse the use of 'steep learning curve'. WIthout getting into original references, you could plot Effort on the Y axis (in which case steep means difficult) or on the X (in which case steep means easy). Most people are gong to associate steep with hard, so we had better stick to it.

Any other phrases people can think of that have come to mean the opposite of what they originally meant? My favourite is 'quantum leap' to mean a big jump instead of a very small one.

My favourite actual misuse is 'cannot underestimate' etc.. Even reporters say things like 'you cannot underestimate the impact of wireless technology today'.

David Clayworth
Tuesday, April 22, 2003

As someone whose degree is in psychology (I focused on cognition, perception, and neurobiology, not that "here's my leather couch, now tell me about your childhood" stuff), I can confirm T. Norman's assertion that learning curves are plotted with time (or something analogous, such as number of experimental trials) on the X axis and increasing capability on the Y axis. I'm not sure I could quickly locate a definitive source to substantiate this claim, only that every piece of primary research I can recall, as well as every textbook, lecture, etc. on the topic, has followed this convention. (It's perhaps so ingrained in the literature of learning and memory that it's considered unworthy of explicit comment.) There's a logic to it as well, insofar as time is the independent experimental variable (X) and learning is the dependent one (Y, with higher performance being plotted further up the vertical axis, as you would expect).

The apparent confusion with the use of "steep learning curve" arises, I suspect, because the word "steep" as used in everyday speech does not always map quite precisely to the concept of a steep (i.e., high-slope) curve where the X axis is time. Consider a steep trail to the top of a mountain. Steep implies a lot of exertion relative to a flatter trail. But it also means you get to the summit after traversing less horizontal distance. When describing a steep learning curve, it's really the latter fact that's more relevant -- you achieve mastery after a shorter period (of time, not distance).

In typical usage, the word "steep" more strongly conveys the sense of difficulty. Yet there are exceptions: If you said you had experienced a "steep rise in income" over the past few years, that would clearly be a positive thing, right? That's how psychologists would tend to read "steep learning curve" -- something like "steep rise in proficiency". In common usage, though, it's interpreted as "steep" == "difficult", and "learning curve [implied: plotted against time]" == "learning process", resulting in an opposite connotation.

Really, this is just one of many cases where a specialized term has found its way into common usage in a more-or-less corrupted form. Not unlike "relational" in "relational DBMS", perhaps.

John C.
Tuesday, April 22, 2003

Great points, David. One of my personal favorites is "I could care less", when of course the real intent is "I could not care less" (i.e., I already care so little it's impossible to care less; incidentally, does this imply that caring is quantized? ;-).

John C.
Tuesday, April 22, 2003

"You can't underestimate" is being used in the sense of you "mustn't underestimate". Think of "you can't go to the board meeting dressed as a pink rabbit".

Modal verbs are notoriously shifty. That's what keeps language teachers like me in a job!

"I could care less" is American. It's the Brits who say "I couldn't care less". I suppose the Americans are being ironic.

Stephen Jones
Tuesday, April 22, 2003

David, to paraphrase Neil Armstrong a quantum leap may be a small step for you but it's a hell of a jump for an electron!

Stephen Jones
Tuesday, April 22, 2003

In a the X axis is Quantity of Knowledge, the Y axis is the time to learn. A steep learning curve is where it takes a long time to learn the initial portion of the thing that you are trying to learn. Therefore, most people use it correctly.

Tim Sullivan
Tuesday, April 22, 2003

Eh, happens all the time, same way 'showstopper' has been completely corrupted. Both defs here: http://dictionary.reference.com/search?q=showstopper
You'll go equally mad trying to get people to stop saying "I'm going to try and..." or any number of other misuses.

anomolous coward
Tuesday, April 22, 2003

Tim Sullivan: sorry, time elapsed is traditionally considering the ultimate in independent variables, and thus, conventionally placed on the x-axis. 
Of course, Einstein modified that a bit.
Great topic, guy, this has always bothered me too. 

Ethan Herdrick
Wednesday, April 23, 2003

"I could care less" is defensible.  You can't apply strict rules of logic to idiom.  See:

http://alt-usage-english.org/excerpts/fxcouldc.html

Alex Chernavsky
Wednesday, April 23, 2003

Okay, I give up.  What's wrong with "I'm going to try and..."?

One-Armed Bandit
Wednesday, April 23, 2003

People often say things like "I'm going to try and fix this" when they should say "I'm going to try to fix this." The only time you should use 'and' is when the trying and the other thing are two separate events, like "I'm going to try [to do whatever it was we were just talking about] and then I'll tell you how I did." (Actually, it's OK to say it, same as 'gonna', but looks bad when written.) Bugs me almost as much as people saying "He should of" instead of "He should have", as in "He should of left it alone." Then again, my boss is from PA and says things like "The car needs washed" instead of "The car needs to be washed", which is one construction I'll *never* get my head around. Excuse me--around which I'll never get my head. :-) Just remember what the man said--"This is the sort of English up with which I will not put." ( http://www.wsu.edu/~brians/errors/churchill.html )

anomolous coward
Wednesday, April 23, 2003

One which I hate which I've seen around here is "my bad" as in "sorry, my bad". That's not English!

John Topley
Friday, April 25, 2003

*  Recent Topics

*  Fog Creek Home