Model Mania

Adventures of an IT Leader Book Cover I just finished an interesting new book by Rob Austin, Richard Nolan and Shannon O’Donnell, called Adventures of an IT Leader. The authors offer a fictionalized account of a business executive’s ups and downs after being assigned the job of CIO for his company. The book does a good job of describing the dilemmas faced by Information Technology leaders, and offers some helpful (although not always simple) advice on how to confront these issues.

At the end of the book, just as our hero is starting to feel like he has a handle on this whole CIO thing, he meets up with a shadowy, wizardly character who has been offering cryptic, sometimes zen-like advice over the course of the tale, and this guy proceeds to burst our hero’s bubble.

…you said you think you know some things. What you mean is, you’ve constructed simplified representations of how those things work. But don’t confuse yourself by thinking your simplified mental constructions are realistic, or worse yet, true … Nothing useful is real. If it’s complicated enough to be realistic, it’s too complicated to be useful. That’s why we build models. Representations. When we say we know things, we just mean we have mental models of those things that we like. Often we like them because they’ve been useful. But let’s not confuse having a useful model with actual knowing.

Managers have a problem … when they fall in love with a particular model of how something works. When they become convinced that a mental model they have of how something works is the right one. When they decide that they know something. … when we become too wedded to a model, we lose our ability to deal with new situations.

And while some elements of the above description might sound a bit perverse – the model-makers giveth, and the model-makers taketh away – eerily similar words appear in a recent white paper from the Software Engineering Institute called “CMMI(r) or Agile: Why Not Embrace Both!”

In some important ways, CMMI is an “ivory tower” of theoretical concepts … CMMI is not and never was meant to be a replacement or a definition of anything in the real world. That is what a model is.

Over the course of my career I’ve had occasion to become acquainted with a number of different models that attempt to represent some aspects of business reality. To name a few:

Frankenstein Movie Poster And while all of these models have proven useful at times, I have noticed that there is a disturbing tendency for all of these creations of ours to eventually grow beyond our original intentions, to assume lives of their own, and to eventually turn on their masters, like Frankenstein’s monster.

More specifically, all such models seem to bring with them the following associated problems.

  1. They become more complex over time. The original CMM was 437 pages long. The CMMI for Development, Second Edition, now weighs in at 676 pages. ITIL has grown in a similar fashion. And while the increased size of these models makes them more comprehensive and complete and accurate, one may well wonder, in line with the characterizations of models above, whether these attributes actually make them more useful. As the character from The Adventures of an IT Leader says, “If it’s complicated enough to be realistic, it’s too complicated to be useful.”

  2. They tend to grow beyond their original scopes and exhibit megalomaniacal tendencies to represent all of reality, as in the utopian quest for a unified field theory. As examples, the Quality movement gradually evolved to a comprehensive Quality Management System that encompasses almost every aspect of business management. CMII has evolved to the point where it claims the ability to solve almost all business problems through the proper application of configuration management. And the CMM, originally conceived as a solution to the software crisis, eventually grew into the CMMI, a comprehensive maturity model for all forms of product and service development. And ITIL, initially a model for IT infrastructure management, now has ambitions to describe all of Information Technology.

  3. They become authoritarian. In order to motivate changes in behavior, these models are often offered up as officially sanctioned documents, or referred to as collections of industry best practices. And while such efforts are well-intentioned, and are often helpful in that they can short-circuit endless and ill-informed internal debate, such attitudes ultimately have the unhealthy effect of making these models seem like unquestionable, authoritative bibles.

  4. They exert a hypnotic effect on their followers. Because of their size and complexity, many of these models seem to demand monk-like devotion from those who would seek to understand them. Because of their ambitions, many of their followers seem to actually come to believe that they offer the complete and unadulterated Truth, and not just a single, necessarily imperfect, view of reality. And because they are used as authoritarian instruments, their followers feel compelled to regard them as official versions of reality.

  5. They tend to encourage exclusivity. Let’s face it: if you are going to invest enough time and effort to become reasonably well-versed in a model that runs to over 500 pages in length, then you are not likely to immediately put that aside and turn to the next model in order to study that. No. You are going to stick with what you know and try to get some return on your already sizable investment in the model you started with.

  6. They form the foundations for businesses of various sorts. Authoring organizations begin to see their models, and continuing improvements to these models, and training on these models, as revenue streams. Authors and consultants spring up and see allegiance to these models as their retirement plans. Vendors build products around these models and sell them to followers at annual conferences. Businesses sink investments into these models and so become committed to them. All of these various forms of financial incentives tend to exacerbate the other problems described here.

None of which is to suggest that we should avoid models altogether. A well-made model is a wondrous thing, and a good one can help us to understand the world around us and provide helpful guidance for our actions.

But let’s keep them off the pedestals and on their leashes, please, to remind us all which is the servant and which the master.

– Herb Bowie (the Model whisperer)

November 6, 2009

Next: Service Overload