Chapter 9. Bad Quality Management
If you give a manager a numerical target, he’ll make it, even if he has to destroy the company in the process.
W. Edwards Deming
The quality of a semantic data model (and any product, for that matter) is not only affected by mistakes made during its specification and development, but also by bad practices followed when measuring and managing that quality. The dimensions we choose to measure, the metrics we use for these measurements, and the ways we interpret the values of these metrics can make a big difference between a successful and a not-so-successful model. This chapter describes some common problematic quality-related practices and suggests ways to prevent them.
Not Treating Quality as a Set of Trade-Offs
We all want semantic models to be 100% accurate, complete, timely, and relevant, yet, more often than not, this is not possible or realistic. A key reason for that (apart from the fact that semantic modeling is a human activity, by and for humans) is that there are several trade-offs between the quality dimensions we saw in Chapter 4 (accuracy, completeness, consistency, etc.) that make it difficult to maximize a model’s quality in all of them at the same time.
The problem with these trade-offs is not that they exist, but rather that we sometimes ignore or forget their existence and we don’t take the time to create a concrete strategy to manage them. Let’s see the most common quality trade-offs we are up against.