Business Architecture

BusinessArchitecture is made up of people, buildings, rules, working practices, infrastructure, pool cars, the Christmas Club, IT, values, carpets, windows, beliefs, goals, extra-marital affairs, stationery, prejudices and an infinite number of other things. For that reason it is far too complex to model, and extremely difficult to engineer. BusinessArchitecture is emergent rather than intended, and blind luck is usually a major factor in getting it right.

--- JasonGorman


No system is too complex to model, certainly not one that is merely difficult to engineer (in whatever sense that word is meant). In most senses of the word, engineering rests on implicit or explicit models; the more complex the system, the more useful models need to be. For complex systems, we need powerful models. Usually, multiple simple models will be more successful than a single model, as each can then focus on what are, within the scope of the model, the more salient aspects, whilst disregarding less relevant details. Thus, an organisation structure chart models the people and departments within an organisation whilst disregarding, for example, financial aspects. Conversely, a budgetary model may eliminate the detail of people (typically becoming "headcounts"), whilst retaining departmental and some financial aspects.

There are two general problems with models:

  1. They omit details that should be included
  2. They include details that should be omitted
(See AllModelsAreWrongSomeModelsAreUseful)

The problem with models in the business sense is that seemingly insignificant factors (like the fact that the IT director had an affair with the Project Manager's wife, or that the weather is getting colder and people are feeling a little low) can significantly affect business outcomes. A large proportion of business modeling labours under the delusion that businesses are deterministic things that can be catalogued, "designed" and predicted. A great many business analysts I've worked with live in a Newtonian fantasy world where something as complex as the sales process - with all its incredibly rich interactions - can be faithfully represented using the same concepts we use to represent computer systems. I agree that nothing is beyond modeling, but we have to accept that sometimes it's the insignificant, extraneous details we leave out that actually make the difference. The problem is that there are so many factors (in fact, effectively an infinite number) that it's impossible to know which ones to leave out. Expect the unexpected!

--- JasonGorman


EditText of this page (last edited November 10, 2014) or FindPage with title or text search