Learning Curve

Refers to the plot of time or attempts (x) versus progress (y), at least in the discipline of psychology. The more popular use of the term plots progress on the x axis and time or effort on the y axis, so that "steep learning curve" can conjure up images of hard-to-climb steep mountain faces or something.

An interesting treatment of the "steep learning curve" phenomenon (for which I myself got flamed for once; I think it's still in some mailing list archive somewhere) see http://www.crh.noaa.gov/library/Grammar/Learn-curve.html.

Here is a similar grammar misunderstanding: I once startled someone by telling her that I got too much positive feedback. She didn't understand.

--EricJablow

Some people say that Windows or MacOS have a shorter learning curve than LinuxLikeOperatingSystems. I think that since most curves are infinite that this makes very little sense.

I would argue that the curve for LinuxLikeOperatingSystems would be something like y=x. On the other hand, the curve for Windows would be something like y=x^3. So, for small values of x (x<1) Windows would be easier, but for large values of x, LinuxLikeOperatingSystems would be easier.

Or, to simplify, the very first few steps of windows are easier, but users quickly get to the point where they would have had an easier time if they had used LinuxLikeOperatingSystems from the start.

You can find a very similar argument made by LarryWall in ProgrammingPerl (for perl vs awk or C). However, it is nonsense. A learning curve is specific to the task you wish to learn how to do. Learning how to install a new device is easier in Windows than in Linux. Learning how to write a device driver is easier in Linux than in Windows.

An alternate formulation: GnuLinux has a steep learning curve, but a high payoff function. By contrast, there's a low ramp for Windows, but you also tend to top out early.

When there are interactions between things you (should have) learned already and what you learn next, the curve will be much more interesting. Probably bumpy. I am reminded again of the BalloonModelOfKnowledge.

It seems to me that MicrosoftWindows is relatively simple in this respect. It's a collection of things, you just need to remember where each is and what it does.

Un*x systems are different because you need to pick up an assortment of skills and use them with each other, and to further your other skills. A thorough understanding of the assorted documentation systems makes reading and cross-referencing docs easier. An understanding of shell quoting will help when some script starts behaving oddly. Maybe this wouldn't be an advantage if ItJustWorks...

I picture the adventure game in which you have to find the match, oil and lamp before you can get the key from the dark room. That sort is much more interesting than the simple branched ones where you wander around making incorrect decisions and reloading. -- MatthewAstley

See also FuzzyAndSymbolicLearning with a graph of the LearningCurve, MindOverhaulEconomics

CategoryEducation