Picture two photographs of Mount St. Helens. One was taken before the 1980 eruption. One was taken after. The mountain in the first stands nearly a thousand feet taller than the mountain in the second.
Which photograph is wrong?
Neither. Both were perfectly accurate the instant the shutter clicked. The first captured the mountain that existed then; the second captured the mountain that exists now. The error appears only if you hold up the first photo decades later and insist this is Mount St. Helens — because the mountain is not the photograph. The mountain is a living geological process, and a photograph is a single frozen frame of it.
Hold onto that, because it explains something the technology industry has spent forty years failing to understand.
A canvas is accurate the moment it’s finished — and starts decaying immediately
Think about the difference between a painting and a camcorder.
A painter captures a subject on canvas. The instant the last brushstroke lands, the painting is accurate — a faithful likeness of that subject at that moment. And in that same instant it begins to decay — not as an object, but as a description — because the subject keeps living and the canvas doesn’t. A year later the portrait is still beautiful and increasingly wrong, and nothing about the painting has changed. The subject moved.
A camcorder is different. It never freezes the subject at all. It keeps recording, frame after frame, staying accurate for exactly as long as you keep it running. The moment you stop, you are back to a still photograph — a canvas — and the decay begins again.
I first wrote about this distinction in 2008: build camcorders, not canvases. It has taken the emergence of AI for our industry to arrive at the same place.
Most technology paints a moving organization
Here is the failure — and it is the same failure every single time.
A technology implementation captures how an organization works — its processes, its rules, its decisions — and freezes that capture into a tool. It paints a canvas of the organization, and then asks the organization to hold still for the portrait.
Some tools are built to adapt, and that is a genuine improvement — but most adapt the surface: the screens you configure, the workflows you toggle. The operating logic underneath — who owns the decision, how the process actually runs, whether people are aligned — stays frozen in the original portrait. The paint can move. The sitter still has to hold the pose.
It works beautifully in the demo, because the demo is the moment the brushstroke lands — the one instant the canvas and the organization actually match. Then it meets the operating floor, where the organization is still moving: people adapt, exceptions multiply, priorities shift, and the process that was captured one way is already running eighty-five ways. The canvas cannot move with it. Within months the expensive new system is a beautiful, increasingly wrong portrait of an organization that is no longer in the picture.
This is why roughly 75 to 85 percent of these initiatives have failed to deliver their promised value — a figure that has stayed stubbornly constant across every technology era. Service-oriented architecture. ERP. Big data. Robotic process automation. Now agentic AI. A different subject (the technology), every time. The same canvas, every time. The same decay, every time.
It was never a technology problem. It was a canvas problem.
The pyramids and the stacks are better canvases — which is exactly why they still fail
The most sophisticated frameworks of 2026 — the governance pyramids, the agentic stacks, the trust-and-risk models — are, to their credit, far better canvases than anything that came before. More layers. More detail. More rigor. Genuinely finer portraits.
But a more detailed painting of a moving subject decays in precisely the same way a crude one does. Detail is not motion. A governance model drawn as a static stack of layers is still a canvas — a portrait of governance, frozen at the moment it was drawn. And governance, of all things, cannot be a portrait. Governance that works is a camcorder: it keeps recording, adjusting to the organization as it actually moves — or it governs a company that no longer exists.
That is not a knock on the frameworks’ craftsmanship. It is the observation that craftsmanship of the wrong kind doesn’t solve the problem. It makes a more convincing version of it.
This is what the two lenses actually are
Everything above has names, and now they will make sense rather than sound like jargon.
Implementation Physics™ is the discipline of building camcorders instead of painting canvases — engineering for the organization as it flows, not as it looked in the demo. It is the study of what actually determines whether a technology delivers under real, moving conditions.
Invariant Physics™ is the one thing in all of this that does not flow. Technologies change completely — the water is unrecognizable from one era to the next. But the requirement that the operating logic be in place before the technology can deliver on it has never changed. That is the river bed beneath the moving water: the channel that keeps directing the flow even as it reshapes, while the water itself is replaced era after era. Until proven otherwise, that requirement has not once failed to hold — and “until proven otherwise” is itself the camcorder’s stance, a claim that keeps recording rather than freezing into dogma.
One guardrail, so none of this is misheard. It does not mean the technology doesn’t matter, and it does not mean nothing is ever true. A camcorder still needs a good lens — a better tool records a better flow; it simply can’t substitute for the recording. And a photograph can absolutely be wrong — a doctored image is a lie regardless of when it was taken. Fidelity to the moment is real and checkable. What decays is not fidelity. It is currency — how well a true snapshot still describes now.
The proof isn’t the metaphor. It’s the outcome.
A frame this clean is still only a way of seeing — until it produces a result. So here is the result.
In 1998, this approach — the one I would later formalize as Strand Commonality™, the practice of capturing an organization’s many moving attribute-streams on an ongoing basis rather than freezing them once — was applied in the real world at the Department of National Defence, in an engagement funded in part through Canada’s SR&ED research program.
The operating logic went in first. The technology went in last. And the outcome was documented: on-time delivery rose from 51 percent to 97.3 percent in three months; cost of goods fell 23 percent and stayed down for years; the team required to run the contract went from twenty-three full-time equivalents (FTEs) to three.
That was 1998. Every piece of technology involved has since been replaced many times over — the water is entirely different. What held was the channel’s role — the operating logic that governs where everything flows. Riverbeds reshape; they don’t stop governing. That is the basis for sustainable success, regardless of the era of technology.
That is the whole of it. Accuracy is not a canvas you finish and hang on the wall. It is a camcorder you never stop running. Truth is believing; accuracy is knowing — and knowing is not something you achieve once. It is something you keep doing, for exactly as long as the world keeps moving.
Which is always.
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
-30-
Related
The Canvas and the Camcorder: Why Technology Fails in Every Era
Posted on July 7, 2026
0
A companion to Until Proven Otherwise: The Lens for Sustainable AI Adoption. That post named the two lenses. This one shows why they hold — and the proof.
Picture two photographs of Mount St. Helens. One was taken before the 1980 eruption. One was taken after. The mountain in the first stands nearly a thousand feet taller than the mountain in the second.
Which photograph is wrong?
Neither. Both were perfectly accurate the instant the shutter clicked. The first captured the mountain that existed then; the second captured the mountain that exists now. The error appears only if you hold up the first photo decades later and insist this is Mount St. Helens — because the mountain is not the photograph. The mountain is a living geological process, and a photograph is a single frozen frame of it.
Hold onto that, because it explains something the technology industry has spent forty years failing to understand.
A canvas is accurate the moment it’s finished — and starts decaying immediately
Think about the difference between a painting and a camcorder.
A painter captures a subject on canvas. The instant the last brushstroke lands, the painting is accurate — a faithful likeness of that subject at that moment. And in that same instant it begins to decay — not as an object, but as a description — because the subject keeps living and the canvas doesn’t. A year later the portrait is still beautiful and increasingly wrong, and nothing about the painting has changed. The subject moved.
A camcorder is different. It never freezes the subject at all. It keeps recording, frame after frame, staying accurate for exactly as long as you keep it running. The moment you stop, you are back to a still photograph — a canvas — and the decay begins again.
I first wrote about this distinction in 2008: build camcorders, not canvases. It has taken the emergence of AI for our industry to arrive at the same place.
Most technology paints a moving organization
Here is the failure — and it is the same failure every single time.
A technology implementation captures how an organization works — its processes, its rules, its decisions — and freezes that capture into a tool. It paints a canvas of the organization, and then asks the organization to hold still for the portrait.
Some tools are built to adapt, and that is a genuine improvement — but most adapt the surface: the screens you configure, the workflows you toggle. The operating logic underneath — who owns the decision, how the process actually runs, whether people are aligned — stays frozen in the original portrait. The paint can move. The sitter still has to hold the pose.
It works beautifully in the demo, because the demo is the moment the brushstroke lands — the one instant the canvas and the organization actually match. Then it meets the operating floor, where the organization is still moving: people adapt, exceptions multiply, priorities shift, and the process that was captured one way is already running eighty-five ways. The canvas cannot move with it. Within months the expensive new system is a beautiful, increasingly wrong portrait of an organization that is no longer in the picture.
This is why roughly 75 to 85 percent of these initiatives have failed to deliver their promised value — a figure that has stayed stubbornly constant across every technology era. Service-oriented architecture. ERP. Big data. Robotic process automation. Now agentic AI. A different subject (the technology), every time. The same canvas, every time. The same decay, every time.
It was never a technology problem. It was a canvas problem.
The pyramids and the stacks are better canvases — which is exactly why they still fail
The most sophisticated frameworks of 2026 — the governance pyramids, the agentic stacks, the trust-and-risk models — are, to their credit, far better canvases than anything that came before. More layers. More detail. More rigor. Genuinely finer portraits.
But a more detailed painting of a moving subject decays in precisely the same way a crude one does. Detail is not motion. A governance model drawn as a static stack of layers is still a canvas — a portrait of governance, frozen at the moment it was drawn. And governance, of all things, cannot be a portrait. Governance that works is a camcorder: it keeps recording, adjusting to the organization as it actually moves — or it governs a company that no longer exists.
That is not a knock on the frameworks’ craftsmanship. It is the observation that craftsmanship of the wrong kind doesn’t solve the problem. It makes a more convincing version of it.
This is what the two lenses actually are
Everything above has names, and now they will make sense rather than sound like jargon.
Implementation Physics™ is the discipline of building camcorders instead of painting canvases — engineering for the organization as it flows, not as it looked in the demo. It is the study of what actually determines whether a technology delivers under real, moving conditions.
Invariant Physics™ is the one thing in all of this that does not flow. Technologies change completely — the water is unrecognizable from one era to the next. But the requirement that the operating logic be in place before the technology can deliver on it has never changed. That is the river bed beneath the moving water: the channel that keeps directing the flow even as it reshapes, while the water itself is replaced era after era. Until proven otherwise, that requirement has not once failed to hold — and “until proven otherwise” is itself the camcorder’s stance, a claim that keeps recording rather than freezing into dogma.
One guardrail, so none of this is misheard. It does not mean the technology doesn’t matter, and it does not mean nothing is ever true. A camcorder still needs a good lens — a better tool records a better flow; it simply can’t substitute for the recording. And a photograph can absolutely be wrong — a doctored image is a lie regardless of when it was taken. Fidelity to the moment is real and checkable. What decays is not fidelity. It is currency — how well a true snapshot still describes now.
The proof isn’t the metaphor. It’s the outcome.
A frame this clean is still only a way of seeing — until it produces a result. So here is the result.
In 1998, this approach — the one I would later formalize as Strand Commonality™, the practice of capturing an organization’s many moving attribute-streams on an ongoing basis rather than freezing them once — was applied in the real world at the Department of National Defence, in an engagement funded in part through Canada’s SR&ED research program.
The operating logic went in first. The technology went in last. And the outcome was documented: on-time delivery rose from 51 percent to 97.3 percent in three months; cost of goods fell 23 percent and stayed down for years; the team required to run the contract went from twenty-three full-time equivalents (FTEs) to three.
That was 1998. Every piece of technology involved has since been replaced many times over — the water is entirely different. What held was the channel’s role — the operating logic that governs where everything flows. Riverbeds reshape; they don’t stop governing. That is the basis for sustainable success, regardless of the era of technology.
That is the whole of it. Accuracy is not a canvas you finish and hang on the wall. It is a camcorder you never stop running. Truth is believing; accuracy is knowing — and knowing is not something you achieve once. It is something you keep doing, for exactly as long as the world keeps moving.
Which is always.
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
-30-
Share this:
Like this:
Related