Lizard tails and autogeneration

I grew up in a part of the planet where lizards are kept as pets in homes (more accurately, the homeowners really have no choice). Occasionally, a cat would pounce on one. Then, something magical happened: the lizard would detach its tail, and scamper away. The cat was left with the lizard's twitching tail—not the outcome it quite expected!

Bruce Silver, in a recent post, offered interesting insight into the capability of a BPM product to generate process models 'automatically.' He has surfaced some of the critical issues that confront analysts and architects. He makes excellent points about wasting enormous amounts of time in process modeling, if indeed the underlying BPM platform can easily autogenerate process models by examining the underlying systems.

While every bit of automation helps (if it is wrapped in the right methodology), this generates a bigger question: does the model define the business or does the business define the model? From my perspective, I feel that by overly relying on the model autogeneration capability, there is a danger that we miss out on some of the most compelling rationales for BPM.

First, I grant that process modeling shouldn't occupy everyone's time. After all, the business of business is business, not modeling, simulating, or even building applications. Anything that helps speed up modeling is good, since it would help refine and improve the business.

At the same time, modeling is more than documentation. The verb 'to model' implies creating a representation of reality by applying certain constraints, which can be simplifying assumptions or the directing influence of some purpose.

Without such constraints, the model can end up becoming as complicated as reality, thus defeating its purpose. But the selective application of constraints and purpose is a human activity. We are still a couple of hundred years away (optimistically) from letting Artificial Intelligence take on this task (at which point, we may all just as well pack up and go vacation on the beach). So, when a process is modeled, it is done so from the perspective of a business goal, such as improving productivity, eliminating cost, innovation, or compliance. Each of these goals brings different perspectives to that same business process. Autogeneration, in this context, is like generating data dictionaries from databases. Useful for a start, maybe, but nowhere as useful as a full metadata (or business semantic) representation.

Second, autogenerating process models by deriving them from systems functionality is a very IT-centric definition of business processes that may or may not be an optimal view. In reality, business processes don't constrain themselves to just systems. There is quite a bit of human interaction going on. Even in the most faithful and comprehensive ERP implementations, there is some level of underground communication going on to work around some of the systems limitations. I do not suggest that BPM should automate all of it (perhaps it is not even desirable to do so), but BPM should at least be able to capture that complexity.

Third—and this is the most serious indictment of autogeneration of process models—the purpose of process modeling (or documentation, in the sense of Bruce's post) shouldn't be to just capture the as-is, file it away, and go home.

The fundamental purpose of modeling is:

• to bring together people from various functions and goals onto a shared understanding of the business processes
• to foster cross-functional communication
• to provide them with the opportunity to get some ah-ha's about their business
• to give them the time and opportunity to innovate.

All of these real benefits are lost if we overly rely on pushing a button, generating a business process, automatically indexing it, filing it, and storing it away in an ECM. Without humans participating in the process, there won't be any innovation. We'd be like that cat that doesn't quite know what to do with the lizard's tail.

The lizard, I was astounded to learn as a kid, can regenerate its tail. But, it can only regenerate a tail, nothing more. Only human imagination, with some help from nifty technology, can create a T-Rex from a lizard tail.