Re: [tlug] Pedant's corner

[Christian Horn <chorn@??>]

On Sun, Jan 25, 2026 at 03:24:13PM +0900, Stephen J. Turnbull wrote:
> Christian Horn writes:
> 
>  > - One of the interesting topics afterwards was the 'steep learning
>  >   curve' of the GNU radio software, and how that could be improved.
> 
> This drives me nuts.  I have no idea what people who say "steep
> learning curve" (derogatory) have pictured in their heads, and I guess
> they're not gonna stop[1], but I had to
> 
> <rant>
> The learning curve is a graph, a picture of a function.  The original
> learning curve (Ken Arrow, one of my advisors) was unit cost versus
> time.  It was quantified in World War II airframe manufacture and
> quickly adapted to all kinds of wartime production.  It was a major
> contribution because it made resource planning substantially more
> accurate.  It has also been adapted to productivity vs time.
> 
> So, a *steep* learning curve means costs drop *quickly* or
> productivity rises *quickly*.
> 
> *Steep is good.*
> </rant>
> 
> Footnotes: 
> [1]  You probably don't want to, since so many people think of steep
> as bad.  And "shallow learning curve" is pretty awkward to think
> about.  I just avoid talking about learning curves unless I've
> calibrated one, graphed it, and can talk about numerical rates of
> costs reduction.  But that discipline is not easy, either. ;-)

You might have a point there.
I actually was also thinking of emacs as another example when I
referred to 'steep learning curve' there.  I wonder if this is
something which just many folks now understand as one thing
("takes quite some time and effort to get into") which differs
from the original meaning.

Actually great material for a post (and potentially good 
comments) on the excellent 
https://languagelog.ldc.upenn.edu/ .

Christian

Attachment: signature.asc
Description: PGP signature