Simplicity vs. Simple Answers –
Naming is necessary to order, but naming cannot order all things – Tao te Ching
Anyone with enough experience in any given profession can see patterns and similarities. Project X looks a lot like project Y , so therefore the solution is XYZ. But experience also provides caution – you need to do due diligence to ensure your instincts are correct. The obvious pattern may have some traps. Characteristics that indicate the need for one solution may be masking deeper symptoms.
In the Data field it often takes serious diligence to not oversimplify things and go to the obvious. Data people are problem solvers that use a deep well of tools and techniques. Most data professionals relish this aspect of our profession.
But there is a disturbing trend that is veering away from that, looking for the simple answer. Shilling an approach as the ‘one and only way.” Recently I have seen a lot of “We just need to do this….” articles. But there is always a risk with these. That risk is that someone cherry picks a few words and runs off with “the” answer. The point of this blog is to reinforce something I consistently try and relay to others in data management.
THERE IS NO ONE WAY. OR ONE RIGHT ANSWER
This isn’t a new challenge.
Granted, data science, AI and machine learning are generating a lot of board-level mind share. But the success rate is still reflective of organizations adopting the window dressing of a few tools and failing to manage the underlying asset.
The thinking about what works and does not work is falling into the common trap of magic answers. If you are not clear what that means, harken back to the heady data warehouse growth days – where it was Kimball vs Inmon in terms of approach. A common request back then was for me to tell an organization which approach was “best.” Truth is, neither one was a perfect answer for everyone. The real answer was , heaven forbid, you had to think.
It is the same for data management and governance. The extended explanation of that means we want to find simplicity, but not be over simplified
Let me present some examples. These are paraphrased from recent articles and conversations. I do not mention individuals or organizations saying these things because these comments were heard downstream of the original author or vendor. So far, I do not think they intend to be oversimplified. Nor are they narrow-minded or intending to be self-serving. Rather, they are looking for simplicity in approach. But simplicity is not necessarily the simple answer. Words are important. And too many people, stressed out by the fierce scrutiny of critical leadership, grab for the quick and simple answer.
Here are some examples of statements that encourage risky thinking:
- “There are three ways to do data governance. A, B and C. C is best.”
- “All we have to do is tell better stories.”
- “’Governance’ has the same root as govern, and that means enforcing and punishment, and that is bad.“
All of these statements can and were taken out of context. All of these have led to oversimplified approaches that then became more complex than promised.
To be clear — These are all misinterpretations of someone trying to simplify our challenges. But they have been taken out of context and oversimplified. Let’s look at them in more detail.
“There are three ways to do data governance. A, B and C. C is best.”
These types of statements come from observations that certain styles of implementing DG seem less irritating to stakeholders and cultures than others. This is true. If the situation permits, organizations want to shy away from big bang, or obtrusive initiatives. It tends to be easier to start with what you have and make small changes. Corporate memory is long-lived. Many companies still echo the disappointments from re-engineering, information engineering, Lean, and early large-scale data warehouse efforts. This is a valid concern.
However, it is unreasonable to say there is only one best way. Sometime there is no data capability other than some scraggly reporting. Other times there is a heavy gaze of a regulatory effort that requires immediate and uncomfortable participation.
Your effort will lay along a spectrum –- you may have a path that is easier than another. But always be prepared to deal with more severe obstacles. Before any implementation style is selected, make sure you know where you sit.
“All we have to do is tell better stories.”
I have heard many versions of this of the past few decades. Essentially it speaks to the difficulty in getting leadership to understand the importance of data. It has taken a lot of forms over the years, but it speaks powerfully to the need to communicate in organizationally acceptable terms and avoid abstractions and shallow business cases. Strong messaging is essential to any successful effort where change is involved, and data governance or data management is change.
The story must be about the business. But once sold, you need to stay that course. Too many efforts hit an obstacle, then panic and scurry to get something done. Then we get the “foundational” technology effort. These do not result in sustainable results. Don’t forget to make the business case the story. Also, remember you still need to execute. Once leadership is engaged, the lens is on you, and you need to work hard to deliver. That means develop an approach that works for your organization, not walk through a bunch of canned steps, and hope it all works out. Also, make sure you can measure when your story is failing, or becoming distorted.
“’Governance’ has the same root as govern, and that means enforcement and punishment.“
Govern is not an evil word. It is necessary. Without governance, we would not have a trusted bank system, healthy food, or relatively safe transportation systems. Governance is a necessary capability as systems become complicated. Data systems and data supply chains are complicated. They require controls. All organizations establish guard rails and constraints. Data should be no different. To say some sort of central oversight of enterprise rules is bad is not being honest. If that oversight says we need to be federated or distributed, it needs to be enterprise policy. At some point, changing enterprise mindsets to an approved and required set of standards or principles is mandatory. Even if you can’t use the word (for some reason) nothing has changed. Change is uncomfortable. Not governance. Don’t get them confused.
A lot of data efforts have been oversold as simpler than they turned out. A lot of promises have been made that have been dashed by extenuating, but detectable, circumstances. Many, many executives have bought in, engaged, then become disappointed.
So even if you see a pattern, or think that something you have read or heard will suddenly make it all easy, check again. Most data efforts turn out to be more difficult than intended, so beware overpromising.
