Granularity of Aggregates - Normalization

[Mike Rawlins <rawlins@metronet.com>]

Again in reviewing the example spreadsheet, I am not sure about the
rules for when to use create and reference aggregates rather than when
to create new basic information entities that are children.  The
Methodology for Discovery and Analysis document in section 8.5 seems to
deal with this, but not in what seems to me to be a very complete
fashion.  "Efficient reuse of basic Core Components" and "maximum reuse
and interoperability" are mentioned as goals.   However, there seems to
be something missing to me that is somewhat similar to "denormalizing"
an RDBMS schema in order to gain efficiency.  For example:  When one
simple value is sufficient, is there a way to specify that we shall use
a basic information entity in an aggregate rather than being forced to
reuse an aggregate information entity that incorporates another
aggregate that finally has the basic entity we need?

"Maximum reuse" seems to be a dominant principle behind the spread sheet
analysis, but is this the only consideration?

Question for clarification before comment:  Is this topic addressed in
the specifications?  If so, where?

Thanks,
--
Michael C. Rawlins, Rawlins EC Consulting
http://www.metronet.com/~rawlins/