A Book on a Beach
In 1990, my wife and I celebrated our 25th wedding anniversary with a holiday in Mauritius. One afternoon, sitting on the beach at Île aux Cerfs, I was reading Eli Goldratt’s The Goal—his presentation of the Theory of Constraints as a novel. Fellow holidaymakers commented on what strange reading material I’d chosen for a beach.
They weren’t wrong that it was unusual. But something in me recognised that it mattered, though I couldn’t have said why. At the time, my consulting business was involving me intensely in developing facilitation skills. I was building System Dynamics models using software like Vensim, trying to understand the flows and interdependencies in clients’ organisations. Goldratt seemed to be pointing at something essential about how systems actually work.
Thirty-five years later, I think I finally understand what I was sensing. And I believe that if we follow where Goldratt was pointing—really follow it, past the practical applications to its philosophical roots—we arrive somewhere unexpected. We arrive at a recognition that the deepest constraint isn’t within our systems. It’s in how we think about systems in the first place.
What Goldratt Saw
The core insight of Theory of Constraints is deceptively simple: in any system, performance is limited by a very small number of constraints. Optimise everywhere except the constraint, and you’ve wasted your effort. Find the constraint, exploit it, subordinate everything else to it, and only then consider whether to elevate it.
This was already relational thinking, though it wasn’t framed that way. A constraint isn’t a thing with intrinsic properties—it’s a relationship. Something is only a constraint in relation to the whole system and what that system is trying to achieve. Move the goal, and the constraint may shift. Change the environment, and yesterday’s constraint becomes irrelevant while something else emerges as limiting.
Goldratt understood that local optimisation—the attempt to make every part of a system as efficient as possible—actually degrades system performance. The parts don’t exist independently; they exist in relationship. Improving a non-constraint doesn’t help the system; it just creates excess inventory, wasted motion, or frustrated capacity waiting for work that the constraint can’t yet process.
This is a profound recognition. It means that understanding a system requires understanding relationships, not just components. It means that the whole has properties that can’t be predicted from the parts. It means that intervention requires wisdom about where leverage actually lies, not just effort applied indiscriminately.
Buffer Management as Relational Wisdom
One of the most sophisticated developments within Theory of Constraints is buffer management. Rather than trying to predict exactly how long each task will take, or precisely when materials will arrive, buffer management acknowledges uncertainty by building protective capacity into the system at strategic points.
A time buffer before a critical delivery date. A stock buffer before the constraint. A capacity buffer in resources that feed the constraint. These aren’t inefficiencies to be eliminated; they’re wisdom about the nature of reality.
What does a buffer actually represent? It’s humility. It’s an acknowledgment that we’re participating in flows we don’t fully control. It’s capacity held open for what we cannot predict. When we maintain a buffer, we’re saying: “I know that reality will surprise me. I’m not pretending to certainty I don’t possess.”
This is relational thinking smuggled into a control framework. The buffer exists because the system-as-lived is never identical to the system-as-planned. It’s a practical recognition that our models are maps, not territory, and that the territory will always exceed our mapping.
Moreover, buffer management works by attention to what’s actually happening. When a buffer penetrates into its red zone, that’s a signal—not a failure, but information. The system is telling us something. Effective buffer management means listening to what reality is saying and responding appropriately, not defending our predictions against inconvenient facts.
Where Theory of Constraints Stopped Short
Despite these profound insights, Theory of Constraints has mostly been applied within a throughput-maximisation frame. The goal is typically defined as making money—now and in the future. The constraint is something to be exploited and eventually elevated. The system is something to be optimised, even if that optimisation is wiser than local-efficiency thinking.
This framing still assumes separate agents optimising a system from outside it. Managers stand apart from the production flow, identifying constraints, making decisions, implementing changes. The system is an object to be acted upon. The managers are subjects doing the acting. This separation seems so natural that it’s rarely questioned.
But what if this separation is itself the deepest constraint?
Consider what happens when we try to predict the future. We gather data, build models, generate forecasts. We produce single-number estimates: this project will take 18 months, this product will sell 50,000 units, this investment will return 15%. These numbers create an illusion of certainty—what Schragenheim and Dettmer, in a recent article, aptly call “the illusion of certainty.”
Their proposed improvement is to replace single numbers with “reasonable ranges”—acknowledging that reality might fall anywhere within a spread of possible outcomes. This is better. But it’s still prediction. It’s still standing outside the flow, trying to anticipate what will happen, then judging after the fact whether reality fell within our predicted range.
Interestingly, their own buffer management is wiser than their reasonable ranges. The buffer doesn’t try to predict; it prepares. It doesn’t judge whether reality matched expectations; it responds to what’s actually happening. Buffer management is participation-thinking expressed in control-language.
The Fallacy of Misplaced Concreteness
The philosopher Alfred North Whitehead identified something he called “the fallacy of misplaced concreteness”—mistaking abstractions for the concrete realities from which they were derived. A map is an abstraction from territory. A model is an abstraction from the system it represents. A forecast is an abstraction from the living, unfolding, uncertain future.
The fallacy occurs when we forget the abstraction is an abstraction. When we treat the map as if it were the territory. When we defend our model against reality rather than updating it. When we judge people for “missing” a forecast that was never more than a guess dressed in numerical clothing.
Single-number forecasts commit this fallacy crudely. Reasonable ranges commit it more sophisticatedly. But both remain substitutions of abstraction for reality. The question “did reality fall within our predicted range?” still positions us as judges standing outside the flow, assessing after the fact.
What lies beyond substitution?
Living Systems and Participation
All living systems exist in uncertainty. A tree doesn’t predict its environment and then act—it responds, adapts, participates in ongoing exchange with soil, weather, and neighbouring trees. The feedback is continuous, not periodic assessment against prior forecasts.
Suzanne Simard’s research on forest ecology reveals that trees share resources through underground fungal networks—what she calls the “wood wide web.” A “mother tree” doesn’t optimise her own throughput; she participates in the flourishing of the whole forest, sending carbon and nutrients to seedlings, even to trees of different species. The forest’s intelligence is distributed, relational, responsive.
What would it mean to design organisations that operate this way? Organisations that expect not to know? That treat ongoing responsiveness as primary rather than better initial prediction?
My colleague Mark Winter’s forthcoming book Deciding Matters explores precisely this territory, drawing on John Dewey, Geoffrey Vickers, and Peter Checkland. The title plays on both meanings: matters are always deciding (settling, taking form through processes we participate in), and deciding matters (the quality of our participation shapes what emerges). “Decision” abstracted from flow is already a distortion.
Sully on the Hudson
Consider Captain Chesley Sullenberger landing US Airways Flight 1549 in the Hudson River. The subsequent inquiry tried to reconstruct his “decision,” comparing it against alternatives, asking whether different choices might have produced better outcomes. They were treating the situation as a problem requiring optimal selection from a decision tree.
But Sully wasn’t selecting from options. He was recognising what the situation required and responding from decades of formation that had shaped his perceptual and motor capacities. His maps had become him through embodied practice—they weren’t representations sitting between him and reality. The 208 seconds between bird strike and water landing weren’t filled with deliberation; they were filled with participation.
The inquiry’s criticism exemplifies what happens when symbolic intelligence, confronted with successful intuitive response, tries to capture it in decision-logic and finds it wanting by criteria that were never relevant. They were asking: “Was this the optimal choice?” The question itself misunderstands what happened.
Sully’s preparation wasn’t prediction—it was formation. He had spent decades studying aircraft incidents, not to predict which emergency he would face, but to become the kind of person who could respond appropriately to whatever emerged. This is a fundamentally different relationship with uncertainty.
The Deepest Constraint
Here is what I’ve come to understand, thirty-five years after reading Goldratt on that beach in Mauritius:
The deepest constraint isn’t within the system. It’s the belief that we stand outside the system.
When we believe ourselves separate from the systems we’re trying to manage, we naturally adopt a stance of prediction and control. We try to model the system accurately, forecast its behaviour, and intervene to produce desired outcomes. This stance generates the illusion of certainty as a psychological necessity—because if we’re separate from the system, we can only act on it through our representations of it, and uncertain representations don’t tell us what to do.
But we’re not separate. The manager is part of the organisation. The consultant is part of the intervention. The observer is part of the observation. Separation is appearance, not reality.
Once we recognise this, the whole apparatus of prediction and control reveals itself as a sophisticated response to a problem that doesn’t exist in the way we’ve framed it. We don’t need to predict the future if we can learn to participate in the present with sufficient skill and attention. We don’t need to control outcomes if we can learn to contribute appropriately to processes that have their own integrity.
Recognition Rather Than Prediction
I call this alternative stance “Recognition Theory.” Recognition, in this sense, means coming to know through relationship rather than detached analysis. It means perceiving what a situation requires and responding appropriately—like Sully on the Hudson, like a mother tree sharing resources with seedlings, like a skilled facilitator sensing what a group needs next.
Recognition can’t be reduced to algorithm. It emerges from formation—the slow work of becoming someone who can perceive and respond in this way. It requires presence, attention, and willingness to be changed by what we encounter. It’s what happens when our maps have become us, when knowledge has been metabolised into capacity.
This doesn’t mean abandoning analytical tools or formal methods. It means understanding their proper place. Models are useful when held lightly, as provisional guides rather than authoritative representations. Forecasts are useful when understood as expressions of current understanding, not predictions of what will happen. Plans are useful when treated as starting points for conversation with reality, not blueprints to be defended against inconvenient facts.
Theory of Constraints, followed to its roots, leads here. The constraint is relational. The system is interconnected. The buffer is humility. The response to buffer penetration is attention. What remains is to recognise that we ourselves are part of what we’re trying to understand—and that this recognition changes everything.
An Invitation
I offer these reflections to the Theory of Constraints community not as criticism but as development. Goldratt saw something important. His successors have developed sophisticated tools and methods. But there’s further to go.
What would it mean to apply Theory of Constraints to our own thinking about Theory of Constraints? To ask: what is the constraint that limits our understanding of constraints? To recognise that the answer might be the very framework of separation and control within which we’ve been operating?
Buffer management already knows something that the rest of the methodology hasn’t fully absorbed. It knows that we’re participating in flows we don’t control. It knows that reality will surprise us. It knows that the appropriate response is attention and adjustment, not better prediction.
What would it look like to build organisations—and lives—on this foundation? Not the illusion of certainty, but the practice of trust. Not prediction and control, but recognition and participation. Not standing outside systems trying to optimise them, but growing into ever-more-skillful participation in the living processes of which we are inescapably part.
This is the territory that opened for me on that beach in Mauritius, though it’s taken thirty-five years to find words for it. I suspect Goldratt glimpsed it too. Perhaps it’s time for the community he founded to follow where he was pointing.
—
Terry Cooke-Davies is a Distinguished Fellow at The Schumacher Institute, developing Recognition Theory. He blogs at The Pond and the Pulse on Substack.
This reflection was shaped in dialogue with Claude, and gently composed with the help of Aiden Cinnamon Tea.
Neither of them requires attribution. Both offered presence.
