This column takes a look at selected beliefs, business conditions, approaches, and psychological influences operating in our data management world. By highlighting these topics for your reflection, I hope you will entertain some perspectives that may be useful to you in this fascinating and complex industry.
Reductionism in Data Management
Reductionism is the organization of observable facts about multiple topics into a manageable system by proposing that one theory can be explained by, or is a subset of (“reduces to”) another. An example from the philosophy of mind is the theory that psychological phenomena can be reduced to chemistry, i.e., the notion that if brain states and emotive states are properties of a physical body, then thoughts, emotions, and beliefs can be said to be essentially chemical. In the novel Breakfast of Champions, Kurt Vonnegut expressed this reductionist view humorously: “Dwayne’s bad chemicals made him take a loaded thirty-eight caliber revolver from under his pillow and stick it in his mouth.”
Human beings tend to be action-oriented. Most of us typically practice economy of rationale and attempt to summarize multiple considerations and simplify explanations, which seems to be a more suitable posture for getting things done. In our fast-paced world, individuals and organizations frequently feel pressure to produce results quickly and as directly as possible. This creates the desire to narrow one’s focus to facilitate action, as in the common software engineering jibe “Let’s get right to the code” (versus completing requirements).
The universe of data management is essentially a complex system of inter-related activities, work products, technologies, and organizational behavior, spanning the full lifecycle of the organization’s data layer. Many explanatory frameworks are available, created by vendors, consultants, consortiums, and standards bodies, that attempt to organize the spectrum of everything we do into reasonable categories of related concepts, processes, techniques, and products.
Vendor frameworks tend to describe and advocate disciplines and tasks that comprise a narrow wrapper around the scope of their product suite; by contrast, the upcoming DMBOK 2.0, intended to provide a sound introduction to best practices in multiple knowledge areas, is projected to be over 700 pages in length – a hefty and information-packed introduction indeed. This is a testament to the breadth and depth of our industry, developed through innovative thinking, cutting edge products, and proven successful approaches over the past 40+ years.
There are persistent challenges for the data management professional in explaining ‘All Things Data’ – what the organization should do, why it should do it, and what the essential connections are – to executives, business leaders and staff, including:
- Data management has suffered for decades from the notion that it is a project, with a beginning, middle, and end, versus a core function that is continually needed as long as the organization creates and manages data, similar to Human Resources or Finance – thus funding tends to wax and wane, causing core products to stall or be abandoned – e.g., a metadata repository is no longer populated
- Many important initiatives that are key to optimizing the data layer and its design, implementation, management, and improvement are strategic and foundational in nature, versus instantly providing a tangible product – e.g., an enterprise data model doesn’t create a monetized data feed
- Listeners often lack the patience to endure detailed discussions the dependencies of multiple foundational tasks – e.g., the business glossary is mapped to logical models, which will be mapped to physical models, interfaces will be scanned and documented…etc…(…and complete data lineage traceability will result after five years)
- Business cases tend to emphasize technologies and tangible products to grab the attention of busy decision-makers – e.g. implement a Big Data platform
All of these elements serve to perpetuate limited perspectives on the part of the organization. As data professionals, we are, accordingly, tempted to reduce all of data management to one of its constituent disciplines, wherever the current emphasis and funding may be pointing – in essence, describing a broad holistic system as if it were equivalent to one its parts. This influence can be noticed in statements such as “Data Management is about”- […Data Quality…(or)… Data Governance…(or)…Metadata…]” (and so on).
When we speak in these terms, we are focusing on a single knowledge area and implying that other disciplines play a supporting or subordinate role (or effectively ignoring them). This limits our sphere of understanding and risks losing the over-arching vision and high-level connections that enable an organization to establish a cohesive, broad-based data management program that will endure and succeed in producing accessible, high quality information.
To be an effective advocate for a holistic, organization-wide approach to data management, informed by the many dependencies and bi-directional support among disciplines, we need to resist these influences. While individual, focused projects must be completed, we should strive to expand our knowledge base, perspective, and language to encompass the big picture.
So what’s the antidote to excessive reductionism? Emphasize the convergent value of giving attention to each node in the nexus of interconnection that characterizes data management.
Here’s an example: suppose you needed to explain to an executive group why the organization should support the development and approval of standard business terms. You could illustrate some examples of projects that were hindered by lack of agreement and shared definitions about critical atomic facts, for example, Customer Status. You could diagram the connection to business determinations about acceptable data quality; you could show how approved business terms underlie optimal data design, how they enable data sharing, how they empower business users with enhanced metadata, describe the critical role of data governance, and how these results are linked to business priorities.
This would be informed communication fueled by holistic thinking and continuity of vision. If you cultivate familiarity with many data management disciplines and core work products, you will be able to better embrace complexity and become more confident in your knowledge of inherent dependencies, becoming a more effective advocate for a strategic and successful data management program.
