Quality is not an act, it is a habit


The Vision of Rob Arntz

Model, reality, direct objects and reports

The following really happened. And it actually saddened me a bit to see, how an IT-department could create something that was so far away from … the facts!

"A model is a schematic representation of reality." That is what you find, when you look for the word "model" on Wikipedia. That is to say, apart from an explanation about a totally different type of models, which I leave omitted here for the sake of convenience.

Recently I was asked to review a model for an organization of physicians I was not familiar with. The goal of the logical data model that was presented to me was to provide the basis for an enterprise wide business intelligence application, in which "all relevant data" of the enterprise were stored.

When I saw this model, a few trivial things caught my attention. Form errors as ambiguities regarding the notation used, unclear semantics of attributes, entities and relations, questions about cardinalities… ‘the usual stuff’.

When I had reported my first findings, I experienced an unpleasant feeling that "something else" was wrong with the model as well. And suddenly I realized what it was. The model had entities such as "Invoice", "Invoice line", "Service", "Department". All seemingly normal, but something was missing. If it is an organisation for health care, you would expect something like "Patient" or "Person", or perhaps "Client", wouldn't you? Nothing of the kind! The grammatical direct object was missing completely. Well, some fields in invoice address might eventually lead towards such direct objects, but then only in a very implicit way.

The sad truth was that the model wasn't based on their own vision of their reality, but rather on the financial package which was already used for years within the organisation to their full dissatisfaction. Thus the model had become a reflection of the administrative reality of the organisation ‘imposed from the outside world’, but it showed no insight at all on the mission, vision and strategy of the organisation. A peek on the website of the organisation learned that patients ought to be central in the health-care process. This was not reflected in the model at all.

Instead of finishing the review, I assembled the modelers. In a joint session we went through the annual report and the website of their own organisation. Subsequently, we investigated in a workshop to what extent their own model matched this (external) communications. The word invoice was literally not present in their annual report nor on their website, while both were full of terms like "health service", "treatment", and of course "patient". It became a very fruitful workshop that led to a lot of enthousiasm with the modelers, because the model now fully reflected their way of talking about reality!

The next version of the model that was offered to me for review, did a lot more justice to the definition in Wikipedia and Van Daele.




Related sites: