Usability and Software Fitness

Dave Long
Tue, 21 Aug 2001 11:00:13 -0700

> I disagree, on both points. (1) Business concerns
> dominate. Good business and mediocre -- or even poor
> engineering -- almost always wins, where bad business
> dooms a product to failure...

I have read criticism of Ford for
running his company by eyeballing
stacks of invoices.  That that was
not a particular problem just goes
to demonstrate that good business
and mediocre, even poor, management
can be successful*.  Perhaps every
business cycle demonstrates that
bad business, even with excellent
management, can fail.

It may be difficult to evaluate a
fitness function when the observed
result is more dependent upon the
business than the software.

There was similar confusion among
cattle breeders for a long time.
Once people started keeping track
of genetic impact upon milk yield,
it was discovered that most of the
attributes on which "prize" bulls
were judged had very little to do
with how cheaply their daughters
could produce quantities of milk.


* to be fair, I think that judging
heights of stacks is actually a
good, and very easy, measurement,
as long as those stacks were first
sorted by orders of magnitude.
Only if the stacks are close does
one need to examine lower-order
stacks, or to actually total up
the numbers themselves.  Who was
it who thought partial ordering
was the basis of most knowledge?

> ... does NOT mean that customers get served best, nor
> that providing the best product is how to succeed in
> business.

If one's definition of "success"
is market dominance, then I would
guess that one should play near-
zero-sum games:

Play negative-sum games, and soon
everyone will have failed.  Play
positive-sum games, and there will
be too many successes, making the
industry pretty crowded.  Play a
near-zero-sum game, though, and
after a while (I believe they call
this "maturation") there will be
a few very large players, and many
many failures.