Applying cognitive linguistics techniques to content management illuminates the constant and inevitable sources of implementation problems. These tools give us a new way of looking at content management and in turn offer us an explanation of what makes implementing CM systems so difficult.

In the course of this analysis, I’ve looked at different categories of content management: the content, workflow, and roles. Because these concepts are the building blocks of any CMS, the underlying cognitive models exert crucial influence over the systems themselves.

In general, the models are inadequate for modeling reality. But this is to be expected: they are idealized cogntive models (ICM), a set of background assumptions on how the world might work. ICMs help us account for the fact that categories aren’t perfect buckets. (Lakoff’s Women, Fire, and Dangerous Things provides a description of this in Chapter 4.)

These models, the ones we use for content, roles, and workflow, have something in common: an underlying model for business. In a world where content is considered product or commodity, roles are defined as discrete sets of responsibilities, and workflows are perfect processes with specific inputs and outputs, the underlying model is business is a factory.

What post to greenonions.com would be complete without a visualization? I’ve represented this notion in a picture, showing how our conception of content production is a blend between a factory (orderly) and the typical business process for producing content (messy). Click the teddy bear for more. (By the way, I was so proud of the teddy bear, I added a couple products to greenonions gear.)

Remember that we use blends automatically to help us make sense of complicated ideas. Reality suggests that business processes are NOT like factories at all, but we project our messy reality into an orderly structure to help define it and translate it into terms computers will understand. When necessary, we will frame business processes in other ways to help us make sense of it from a different perspective. This is something humans can do nearly unconsciously with no effort, but computers can not: they have only one frame for making sense of business processes. I believe humans reframe business processes all the time, that the factory model serves as an ICM but we recognize when it cannot hold and make decisions based on current situations and scenarios, creating yet another blend that offers a solution to whatever obstacle the process faces.

Although we have many cognitive models for business, the factory model seems to pervade all thinking with respect to content management, especially for the purpose of generating system requirements — computers like the binary structures inherent in idealized factory floors. Because the factory model is so entrenched with respect to content management, we can not leverage other models for building a CMS. Even so, I’m not convinced there’s a better cognitive model out there that would permit the level of flexibility required. In other words, there is no best cognitive model of business for deriving requirements.

Instead, CM systems need to recognize that an important part of management (decision-making) is blending: making decisions requires cognitive processes currently unavailable to machines. Current implementations of CM systems have evidence of pseudo-blending, artifacts that attempt to mimic this cognitive process. For workflows, it’s exception planning; for content, it’s the content template. Sometimes we attempt to solve this problem by simplifying: creating two-step workflows or highly generic content templates. While this reduces the blending burden on machines, it also renders the CMS no more useful than non-automated publishing systems.

What I recommend instead is moving the blending to where it belongs, to the humans. Content management systems can save time, but only if they are employed to do tasks that do not require blending. This is easier said than done: humans take blending for granted. We do it automatically and it may be difficult to see where blending is happening and where we’ve attempted to shift it to the machine.

My thinking on this subject started with the desire to build a better CMS. It was during that process that I was coincidentally reading Lakoff and realized that his analysis of language could be applied to content management. We now return to where we started: building a better CMS. To do so would require time and money I don’t have. Instead, I will continue to develop the framework theoretically with the hope that one day it can be tested.

Also, there is clearly a lot of work to be done in the analysis of content management ICMs and categories. I will continue to post items that deconstruct content management’s underlying concepts.

Finally, conceptual blending analysis can be applied to more than just content management. Perhaps there’s room to explore how conceptual blending impacts other areas of design.