Early in my career I wrote an article with a title I still stand behind: good organizational design is destructive. The argument, in its youthful form, went like this — most people are capable and motivated, most organizations are designed as if they aren’t, and the designer’s real job is to remove the accumulated junk standing between good people and good work.
A decade of sitting inside organizations of every size and disposition has not changed my conclusion. It has changed my understanding of why it’s true, why it’s so rare, and what it actually takes to do well. The young version of the argument said: tear it down and watch people thrive. The current version says: subtraction is the most demanding discipline in work design — which is exactly why it’s the most valuable.
Addition is the default state
Start with an observation you can verify in any organization you’ve ever worked in: structure accumulates. Policies, approval steps, recurring meetings, mandatory trainings, reporting templates, sign-off chains, tools layered on tools. Each one arrived for a reason. Almost none ever leaves.
There are at least three forces conspiring here, and it’s worth seeing them clearly because each one feels innocent on its own.
The first is fear, fossilized. A remarkable number of organizational rules are monuments to a single bad incident. Someone, somewhere, once abused an expense policy — so now everyone submits receipts in triplicate. One project went sideways — so now every project passes through a stage-gate review. Each rule is a small fear that got promoted into permanent infrastructure. The incident is forgotten; the rule is immortal.
The second is the deep asymmetry in how organizations perceive risk. Adding a control is legible and defensible: if something goes wrong, you can point to the safeguard you built. Removing a control is the opposite — if anything goes wrong afterward, ever, the removal will be blamed. So the rational move for any individual, in any moment, is to add. Nobody gets fired for installing a checkpoint.
The third is simply human cognition. When people face a problem, they reach for what can be added before they consider what could be taken away — researchers have demonstrated this bias toward additive solutions across all kinds of tasks, and you don’t need a lab to see it. Watch any leadership team respond to any organizational tension and count how long it takes someone to propose a new process, a new role, a new meeting. Now count how long until someone proposes deleting something. You’ll usually be counting forever.
Put these together and you get the default trajectory of every organization: monotonic accumulation. Structure ratchets up and never down, like barnacles on a hull — each one small, each one justified, collectively dragging the whole vessel until enormous energy produces very little speed. Nobody designed the drag. That’s the point. The accumulation is what happens when nobody is designing.
What the accumulation actually costs
The standard complaint about bureaucratic buildup is that it’s slow and annoying. True, but shallow. The real costs run deeper.
The first cost is attention. Every rule, checkpoint, and standing process is a small standing tax on the scarcest resource your organization has. Not just the minutes spent complying, but the cognitive residue: the part of everyone’s mind permanently allocated to remembering how things must be done here, which forms to file, whose sign-off to chase. An organization’s total capacity for focused, creative attention is finite, and accumulated structure quietly consumes it the way background apps drain a battery — individually negligible, collectively decisive.
The second cost is what the design says. Most accumulated controls exist to manage the worst-case employee: the shirker, the cheat, the loose cannon. But here’s the arithmetic nobody runs — those people are a small minority in any healthy organization, while the controls apply to everyone. You are taxing the capable, motivated many to constrain the problematic few. And the message lands. People are exquisite readers of what their environment assumes about them, and they have a well-documented tendency to live down to low expectations just as reliably as they live up to high ones. Design an organization around distrust and you will, with great efficiency, manufacture the untrustworthiness you feared. The control system becomes a self-fulfilling prophecy with a compliance department.
The third cost is the subtlest: accumulated structure crowds out the things that don’t need to be managed at all. People genuinely want to avoid letting down teammates they respect. People will pour discretionary energy into work that connects to something they actually care about. People love having learned, even when learning is uncomfortable. These motivations are free — they come with the humans — but they’re fragile in a specific way: they operate only in the space the structure leaves open. A team drowning in mandated process has no room for the voluntary commitment that makes teams great. A job description drawn as a rigid cookie cutter discards every part of a person that doesn’t fit the shape — and the discarded parts were often where the energy was.
Subtraction is a discipline, not a demolition
Here’s where I have to argue with my younger self, because the early version of this idea had a flaw: it made removal sound easy. Find the cruft, delete the cruft, liberate the people. That framing produces a failure mode I’ve now seen up close — the new leader who torches every process in sight, mistakes the resulting chaos for freedom, and quietly rebuilds the same bureaucracy two years later with their own name on it.
The trouble is that organizational cruft doesn’t come labeled. Every barnacle was once somebody’s solution. Some of those rules are fossilized fears, safe to remove. Others are load-bearing — they encode a lesson the organization paid dearly to learn, and deleting them just schedules a re-purchase of the same lesson at a higher price. From the outside, the two look identical.
This is the old Chesterton’s Fence problem: don’t remove a fence until you understand why it was built. And it’s why genuine subtraction is harder than addition, not easier. Addition only requires a fear and a form. Subtraction requires understanding — an archaeology of why the structure exists, what it actually protects against, and whether that threat still lives — followed by judgment about what to do with what you find.
In practice, the discipline looks something like this. Treat every removal as an experiment rather than a decree: pick a structure whose purpose nobody can articulate, suspend it for a defined period, watch closely, and decide based on what actually happens rather than what anyone feared would happen. Make subtraction a standing practice rather than a purge — a recurring question in your operating rhythm (“what should we stop doing?”) rather than a once-a-decade bonfire. And replace removed control with visible commitment: shared goals, public progress, clear ownership. The point was never zero structure. The point is minimum viable structure — just enough to channel energy, never so much that it substitutes for trust.
That last distinction matters enough to restate. The question is not “how little structure can we get away with?” It’s “does each piece of structure amplify what’s best in our people, or insure against what’s worst?” Amplifying structure earns its keep. Insuring structure should have to fight for its life, annually.
The subtraction stakes just went up
One more thing has changed since I first made this argument, and it tilts the field further in the wrong direction.
Addition used to have at least one natural brake: effort. Writing a new policy, building a new template, standing up a new reporting process took human time, which imposed a modest discipline on proliferation. That brake is now gone. With AI, generating structure costs approximately nothing — comprehensive policies, elaborate process documentation, dashboards, frameworks, all producible in minutes by anyone with a prompt. The marginal cost of organizational addition has collapsed.
The cost of subtraction, meanwhile, hasn’t moved at all — because subtraction was never bottlenecked on production. It’s bottlenecked on understanding, judgment, and the courage to be accountable for a removal. No model supplies those. Which means the gap between how fast organizations can accumulate and how fast they can shed is about to widen dramatically, and the organizations that don’t build a deliberate subtractive practice will find themselves generating their own sediment at machine speed. The discipline this piece describes was always valuable. It’s about to become a survival trait.
The designer’s real job
So yes — a decade on, I’ll double down: good organizational design is destructive. But I’d now put it less romantically. The creative acts of work design — the structures, the roles, the rituals — are the visible ten percent of the job. The invisible ninety percent is curatorial: continuously noticing what has accumulated, understanding what it’s for, and deliberately removing what no longer earns its tax on the organization’s attention and trust.
Organizations don’t drift toward openness. They drift toward accretion, and only deliberate, recurring, well-judged subtraction holds the drift at bay. The people inside your organization are, for the most part, more capable and more motivated than your control systems assume. You don’t need to install that capability. You need to stop burying it — one careful removal at a time.
A note on lineage: an earlier version of this argument appeared in 2016 as “Good Organizational Design is Destructive," written during my decade at The Ready. The subtractive thesis and the positive view of human capability began there; the accumulation mechanics, the Chesterton’s Fence discipline, and the AI argument are where the thinking has gone since.