Blog

Discussion is certainly wonderful. Steve’s responses to the “matrix model” model I posted before today’s call made me think of two features it lacked: (1) first, among the attributes associated with each category (or emphasis area) in the matrix should have been a additional dimension, entitled “interdependencies,” in which the connections with other emphasis areas could be identified and described, and (2) it should have been clear that within each dimension and category, the content would (or at least could) arise to some extent from crowdsourcing. Most crowd members probably would be affiliated with the subject matter area under discussion, but not necessarily so, and in fact some of the more interesting inputs might come from ‘outsiders’

Using the interdependency dimension, it might be possible to represent the information listed in the maturity model’s existing ‘support matrix’ (displayed on the website) in an interesting new context.

Anyway, I’ll leave it there, since the model’s value will be principally as a reference point rather than a blueprint, but I did want to mention this example of community discussion spurring extended consideration.

The community’s last conversation (on metadata repository, 11/17) crystalized some thoughts already brewing in my head on the high-level structure of the proposed maturity model. Specifically, I was struck that while the content of the drill-down discussions was becoming more granular and substantive (and thus more valuable), the connections among topics were not being clarified to an equal degree. Indeed, a noticeable amount of repetition and overlap seemed to be developing during drill-down discussions, making it difficult at times to see where particular threads of discussion “fit” into the overall conversation.

In my line of work (as a risk taxonomist) this kind of overlap usually signifies a need to look at the high-level category structure. To some, “uber classifications” of this type are a bit of window dressing, useful for reports yet contributing little of substantive value.

I have always disagreed with this view. Specifically it seems to me that top-level categories are like high-level winds in the Caribbean during late summer when tropical depressions are meandering about. These winds are important in determining the course of the future storm systems brewing below, influencing their direction and (in many cases) creating interference patterns that can prevent the storm from taking shape at all. (This is not that I consider our community discussions to be an incipient hurricane; no analogy is perfect!)

We might want to take a look at the metadata maturity model from a similar perspective, to see how well our high-level categories are driving the project. I am separately posting a PowerPoint that outlines some specific ideas along these lines.

One reason for posting a blog on this point (versus merely posting) is to explain why I think it is important to interrupt the ongoing community discussion process for a side-bar conversation.

In short, it seemed important to acknowledge that I’ve seen this kind of intervention happen before, and not always to good effect. In taxonomic consulting engagements, I dread the moment in a project when—after achieving some degree of progress and consensus—some well-meaning workgroup participant suggests that a better general method is available to address the problem, and that we would do far better to restart the discussion using some alternate approach.

I cringe at such moments because re-opening a discussion of high-level architecture in project taxonomy—while often worthwhile (and sometimes project-saving)—do place additional demands on the team, which can jeopardize all of the accomplishments and momentum a group has struggled to achieve so far. Hence, a significant “hurdle cost” must be associated with any such suggestions.

Nonetheless, I think the balance here is in favor of an additional conversation. I guess I’ll learn shortly whether or not all of you others agree. I’ll look forward to additional thoughts and reactions.