Jim:

> That article has one statement that sums it up nicely for me:
>
> "Of course, anything that forces us think carefully about index design
> is a good thing"
>
> I still think that that is the essence of what Water was getting at.

The following may be just a misconception on my part.

I believe the biggest proponents of surrogate keys on UT work
alone of as part of a small team. They want to deliver
applications as quickly as possible and they use surrogate keys
because it works for them (I am not arguing against that). They
do not consider the relevance of a natural key because (1) it
requires some research that may be perceived by their client as
unproductive, and (2) it requires changes to their framework
which may not be billable. The application may benefit by
choosing natural keys for some/all of the tables but it will
work with all surrogate keys and they can certainly deliver
quicker doing so. IMO, these are valid reasons for always using
surrogate keys but it does not mean their use is a Best Practice!

Daniel