Modeling does not require so much

Object-role modeling approaches are sufficient

Opponent: The ORM (Object-Role Modeling)[Halpin] or the NIAM (Nijssen's Information Analysis Method)[Nijssen, Falkenberg, Halpin] allow to represent the application on the basis of atomic attribute types and relationship types. Nothing else is allowed. These two constructs are powerful enough to represent any application.

Proponent: You are completely right. These two constructs are powerful enough to represent any application. The relationship type construct is relationally complete, i.e. any relational database can be specified on the basis of ORM schemata.

Opponent: OK! Then let us stay with ORM or with NIAM!

Proponent: Please dont throw out the baby with the bath-water! ORM is useful. But ORM allows modeling only with a very high granularity. I remember the example of the Boeing company. They modeled the construction of the plane entirely by NIAM diagrams. Unfortunately, the size of the diagrams was so high that they could place the diagram using all walls in the research building. Nobody was able to capture these diagrams.

Opponent: This argument is unfair! What is the reduction in size for ER diagrams for such applications?

Proponent: The reduction is a linear factor estimatable by about 8 to 10. But there is another severe difference in the approaches. Extended ER models such as HERM allow abstraction of sub-diagrams. They allow further stepwise construction by decomposition in a consistent form.