In the late 90s, my friend Sean handed me a battered copy of Daniel Quinn’s Ishmael that had been passed from person to person across the country, a common practice for Quinn’s philosophical novels. As a 90s kid, I found the novel incredibly influential, and it helped shape my worldview. His work did not read like the abstract philosophy I was accustomed to in academic settings; it felt like a lens that revealed how much of what we call normal is simply inherited belief, reinforced until it becomes invisible. It stuck so deeply that I marked it permanently: a tattoo of Cain and Abel, not as theology but as Quinn framed them — Leavers and Takers. Competing visions of how humans relate to the world. One oriented toward participation and continuity. The other is toward control and exception.
Ishmael introduced that idea. The Story of B, a companion novel, clarified it. The argument wasn’t that civilization is malicious or misguided, but that it operates within a cultural vision so dominant that it no longer seems like a choice. We don’t question it. We design within it.
That distinction — between vision and programs — stuck with me long before I had the language to apply it to my professional life.
At the time, I didn’t connect that worldview to data. Then, during my doctoral research, I returned to The Story of B, and this time the parallels were impossible to ignore.
Vision Comes Before Programs
One of Quinn’s core ideas is that programs do not create change on their own. They are a response to existing behavior. They are corrective, not generative. Vision, by contrast, defines what feels natural, reasonable, and inevitable.
Programs follow vision. They never lead it.
That distinction is important to understand because data governance, as practiced today, is almost entirely programmatic. We talk about frameworks, operating models, policies, councils, tooling, and maturity levels. As I’ve discussed ad nauseam in my recent articles, my research shows that most organizations already have extensive governance structures. What they lack is shared meaning.
Practitioners describe environments where governance exists on paper but not in practice. Roles are defined, yet unclear. Policies exist (maybe) but are inconsistently enforced. Data Governance capabilities are operational but lack resilience and collapse under organizational change.
From a program-centric view, this looks like an execution failure.
From a vision-centric view, it looks like misalignment.
The Unspoken Data Vision We Already Live With
Every organization already operates under a data vision—even if it’s never articulated.
It’s embedded in incentives, workflows, and organizational design. It shows up in assumptions like:
data is produced for local use first,
speed outweighs coherence,
ownership is contextual,
quality is situational,
governance is something added after the fact.
This vision is not malicious, but it is dangerous, nonetheless. It quietly shapes behavior long before any governance program is introduced.
Governance initiatives then arrive not as expressions of that vision, but as correctives — attempts to counteract outcomes the vision produces. That is why governance so often feels heavy, fragile, or oppositional. It is asking people to behave differently without changing the story that tells them what “normal” work looks like.
Note: I am intentionally calling out a data vision detached from the organizational vision. We’ll come back to this.
What the Research Actually Points To
The most important finding in my research is that effective data governance relies primarily on social architecture. It is about understanding.
Structure alone does not produce desired behavior. Policy and enforcement help, but only when people understand why data governance exists and how it connects to their decisions.
Across hundreds of practitioners and dozens of industries, most organizations report that data governance structures are in place. The problem is coherence. Organizations that view data governance as successful consistently focus their programs on culture and literacy, establish accountability for communicating about data governance, and view their programs as sustainable. Compared to value non-believers, characterized by:
limited understanding of data governance,
weak alignment between policy and day-to-day decisions,
inconsistent enforcement,
and vulnerability to organizational change.
These negative characteristics are the symptoms of running data governance as a program layered on top of an unchanged worldview.
Organizations with stronger data governance outcomes are those where people can explain — simply and consistently — what data governance is for, how it affects decisions, and why it exists.
Programs explain what to do.
Vision explains why behavior makes sense in the first place.
Without vision, data governance remains effortful.
With vision, programs become almost invisible.
Governance as an Expression of Vision
The argument here is not that programs are unnecessary. Quinn is clear on this point: programs are not forbidden. They are provisional. They exist to support vision, not substitute for it.
Data governance becomes resilient when it stops trying to manufacture behavior through structure alone and instead reflects a shared understanding of data as a critical resource, a decision surface, and a source of responsibility.
That shift — from program-first to vision-first — is what enables data governance to survive reorganizations, technological change, and new waves of innovation, such as AI.
You cannot program your way into that.
You have to see it first. With the vision comes the stories and the beliefs.
In my experience, I can now see that I’ve had programs fail because I confused them with vision. Those early programs never had a chance. And almost all data governance today is run as a program.
This isn’t a new idea. It comes from a much older idea: programs are what societies build when their underlying vision produces outcomes they don’t like. Programs react. Vision leads.
Programs Are Reactive by Design
In The Story of B, programs are described as corrective mechanisms. They exist to counteract behavior that emerges naturally from a dominant worldview. Programs don’t shape reality; they attempt to compensate for it. They require constant reinforcement, justification, and energy because they are always pushing against the current.
That framing maps uncomfortably well to how data governance operates in practice.
Organizations rarely introduce data governance as a proactive design choice. More often, data governance arises in response to failure — when existing data practices begin to fall apart.
Note: I have written extensively on this point, and I doubt most readers will need convincing, so I won’t elaborate. If you are unconvinced, I suggest stopping here and catching up on my other posts.
This matters because programs that exist primarily to correct behavior are inherently fragile. They don’t persist on belief. They persist on effort.
If we don’t accept this, we will keep mistaking effort for progress — and calling it data governance.
Why Governance Programs Don’t Become Resilient
We operate in a dynamic world. Change is constant for everyone, but for those working with data, the effects are more pronounced. Rapid technological shifts, evolving regulatory demands, and ongoing economic and geopolitical uncertainty collide directly with how data is produced, managed, and used. In that environment, resilience is not optional.
Resilience is about what survives change. Data Governance rarely has the resilience to survive organizational restructuring, executive turnover, shifting priorities, or major technology transitions, even when formal programs exist — complete with operating budgets, skilled practitioners, and defined structures
Observed from a programs-versus-vision perspective, this is expected.
Programs are provisional. They depend on sponsorship, enforcement, and continual explanation. When pressure increases, programs are the first thing to be cut — not because they’re unimportant, but because they are not foundational.
Programs cannot compensate indefinitely for a misaligned vision.
Vision, by contrast, survives disruption. It travels through people, not org charts.
On Vision and Data Governance
A data vision is a shared belief system about how data fits into the organization’s way of working. Here’s the kicker, though: it is not, nor can it be, a separate vision from the organization’s vision.
To put it bluntly, the broadly accepted idea of a Data Strategy in its current form, which includes programmatic data governance, is a dangerous myth.
Okay, Tony, nice dramatic effect — way to be a rebel.
You can come at me on this, folks — maybe I am being dramatic — but consider that it is logically inconsistent to run data governance programs that require constant justification for why rules should be followed, while senior leaders simultaneously claim the organization is — or aspires to be — data-driven.
Our organizations’ visions must assume that:
data is a shared, enduring asset,
decisions about data have enterprise consequences,
stewardship is part of normal work,
quality, risk, and value are inseparable,
governance exists to coordinate decisions, not to police behavior.
People wouldn’t need to be sold on data governance if it reflected how they already understand their responsibilities.
With a corrected vision, data governance won’t feel like a heavy lift. It will feel boring — and most things that are resilient tend to be, well… boring.
The Core Reframe
Here is the uncomfortable conclusion that ties the philosophy to the research:
Data governance does not fail because programs are poorly designed.
It fails because programs are trying to correct outcomes produced by a vision that never changed.
That’s why organizations keep rebuilding data governance. The cycles of failure focus on the mechanisms, without touching meaning.
The organizations that will survive and thrive in the years to come won’t be the ones with the most elaborate governance programs. It will be the ones whose vision of data already supports coordination, accountability, and learning. You cannot program your way into that, either.
The data governance capability is resilient when it stops fighting the river and supports the flow. The stark reality is that if your data governance capability is still fighting the river, it is not resilient, and if it's not resilient, neither is your organization, because its vision is unfit for the AI revolution.
This is the impetus to stop being polite, to stop ignoring the cracks we have learned to work around, and to finally confront the reality that after everything collapses, all that remains is a tattered banner of data governance standing alone in the ruins of our organizations — raised too late, asked to symbolize order only once coherence has already been lost.
On Vision and Data Governance was originally published in My Column Has NULLs on Medium, where people are continuing the conversation by highlighting and responding to this story.