[plt-scheme] Re: Programming for non-programmers

On Fri, 15 Oct 2004 07:52:22 -0400, Matthias Felleisen
<matthias at ccs.neu.edu> wrote:
> But, I used *if* for a reason. As I have pointed out before, the
> current savings of moving a software engineering job abroad to a
> country where SEs earn a quarter of a US SE is 20, 25 at most 30%.
> These numbers are available in public studies as well as in IBM papers
> about this years migration of jobs. That's a large gap for now, and
> definitely worth "arbitraging". But the salaries in other countries are
> jumping by leaps and bounds. I predict that this 10-year old out
> sourcing wave [it's only in the news because of the US presidential
> elections; there is no other reason, it is an old phenomenon] will
> peter out in a few years.

This is especially interesting when you consider the economic effects
of programmer productivity.  A large number of studies over the past
30 years have claimed a 10:1 productivity ratio between the best
programmers and average programmers.  (In my experience, the 10:1
ratio is credible.)

Of course, the best programmers don't get paid anything like 10 times
the average rate--they get paid perhaps 3 or 4 times the average rate.
 So this is an economic no-brainer.

But 10 average programmers can't actually replace one really good
programmer.  Because communication costs increase non-linearly with
team size, a large team of average programmers will spend half their
time in committee meetings, arguing over UML diagrams.

So if can reliably staff your project with very good programmers, you
can reduce development costs by at least 30%.  So why doesn't
everybody do this?

  1) Good programmers are scarce.
  2) Good programmers are hard to identify if you're not a good
programmer yourself.

The PLT project is trying to fix (1) by fixing CS education.  And
they're doing a good job.  I've actually worked with people before and
after they've studied HtDP, and seen people with very basic skills
turned into respectable programmers.  I'd certainly pay 30% more for
one of their graduates than for a programmer from a typical Java
curriculum.

Didn't Yourden perdict "the decline and fall of the American
programmer" immediately before the last big boom started? :-)

Cheers,
Eric