We Like Bad Data
Ian Bailey
From the process model, either the BA, or a data modeller will attempt to ascertain the information flows between processes (somewhat akin to astrology or divining for water). These flows will inform the development of a data model. Data modelling is a bit (only a bit) like philosophy. It’s about producing a rational model of a particular aspect of the world. In philosophy, any decision is backed by three thousand years of the finest minds in humanity in continual debate and peer review. Similarly, in data modelling, someone with a degree in computer science makes a 30 second decision; “I’ll use a text string…oh, and I’ll make it 20 characters long”.
The data model will then be given either to the IT department, who think they know better and build something that uses really cool technology, or to a software development company who will just re-badge the last application they developed. About half way through the development cycle, a management consultant will come in and re-engineer all the processes the BA thought he/she had captured, sack half the staff and move most of the processes offshore. When we get to User Acceptance Testing, the users quickly twig that what they are being shown is going to keep them in work for the next ten years so they gleefully accept, and everyone’s happy.
del.icio.us