Jake Ivey wrote:

     Working in a site of a type you're familiar with, or in a site you've
     become familiar with during the excavation, you know where some things
     are going to be.  It would be silly to use a random approach in a site
     whose structure you know.  So you're left with following the model you
     have built from previous experience, which is your working hypothesis
     about this site -- your theory.

Which is exactly the pitfall in question.

"Ho-hum. I've dug a hundred of these. Hell, I even published one. Stick two
squares over there, another one yonder, look for the corners, and we're out
of here."

Okay, exaggerated intentionally, but you get the picture.

Every site is unique, and every unit is a test square that will guide us to
the next unit. The adventure continues even after the last hole is
backfilled and the last artifact is described. We should assume nothing,
pre-suppose nothing, and most of all, ignore nothing.

We will never achieve ego-free research design (not with this crowd, at
least), but it should always be an objective. Nor should we ever attack any
type of site without a thorough background knowledge of its technology and
culture.

But like I say, the ice beneath our feet becomes thinner and thinner as we
theorize farther and farther from the factual content of the raw data.
Maturity as a researcher is achieved when you know the boundary between a
sound, evidence-derived, conclusion and a beer-hall hypothesis. It's a very
vague boundary.