Why Your Leaders Aren't Failing - Your Structure Is

The Uncomfortable Truth Behind Manufacturing's Leadership Crisis

Every year, companies spend billions on leadership development: executive coaches, management training, accountability systems, values workshops. Yet, in facilities across the country, the same problems continue to occur: simple repeat errors, accountability complaints, machine downtime, poor morale and unfollowed processes.

Looking at the outcomes, business leaders reach the same conclusion: leadership problem. That’s wrong. At least, it reflects the results of asking the wrong question.

The Law Nobody in Operations Talks About

In 1958, cybernetician W. Ross Ashby formalized what became known as the Law of Requisite Variety: only variety can destroy variety. A controller, any controller, must possess at least as much variety (complexity, adaptive capacity, response options) as the system it seeks to regulate.

In plain terms: if the environment can produce 1,000 different failure states, and you as a leader can only respond to 200 of them, the other 800 will eat you alive. Not because you're a bad leader but because your capabilities are inadequate for the problem.

Ashby's Law doesn't care about your work ethic, the number of certifications you hold, your personality or how many times you walk the floor. If the complexity of what you're managing exceeds the complexity of what you can perceive and respond to, the system fails. Every time.

Where Hierarchy Works and Why

The traditional management hierarchy, the kind baked into military doctrine and simple service organizations, is genuinely effective in one specific context: relatively homogeneous teams executing well-understood tasks.

A rifle platoon is a good example. Every soldier trains on the same weapons, the same doctrines, the same physical standards. The officer commanding that platoon has done what the soldiers do. He's had extensive training in it. He knows the equipment, understands the failure modes, and has the experiential variety to supervise effectively. The hierarchy functions because the leader's variety reasonably matches the environment's requisite variety.

It's not a perfect system but it works.

Now walk into a factory.

The Factory Floor Breaks the Model

In a modern manufacturing facility, a front-line supervisor might oversee a dozen operators each running a different machine. On the surface this looks manageable. Twelve people, one supervisor, it should work.

But look closer at what those twelve machines actually represent.

Each one has 20 to 30 critical process settings (temperatures, pressures, speeds, feed rates, physical adjustments and cycle times) that govern how well the line runs. These settings also determine whether the line produces good product, whether it creates conditions that put people at risk, and whether the machine is being slowly destroyed by parameters nobody thought to question.

Safety, quality, productivity, and equipment longevity, the four things every plant manager loses sleep over, all flow directly from critical setup steps that nobody controls, documents, or verifies.

At this critical level of detail, in many organizations, there are no standards and there is little documentation. There is only fragile, unverified and inconsistent institutional knowledge that walks out the door every time someone quits, transfers, or calls in sick.

Now put a supervisor over all twelve of those operators. He's looking at 20-30 standards per person across 12 people with multiple SKUs. He may know that a standard process is needed. Someone may even have written one. But how do you get everyone to remember and follow all of those standards?

That supervisor cannot effectively check 240 to 360 critical process variables across a dozen machines multiplied by a number of SKUs while firefighting. Especially when he hasn't done those jobs and doesn't know the critical points.

And that's before turnover hits. Before someone puts the wrong product in a machine or runs half a shift with the coder printing the wrong stuff.

What does the supervisor do in that environment? He triages. The supervisor's best play is to run toward the loudest alarm and deal with today's crisis visibly enough to avoid punishment.

The whole team just hopes tomorrow's catastrophes don't land before they catch their collective breath.

At the senior levels we call this poor leadership. We write this supervisor up for not being proactive. Maybe we send this supervisor to a management seminar.

But Ashby would call it something else entirely, a variety mismatch. The environment's requisite variety (the sum of every undocumented setting, every undertrained operator, every process that exists only in someone's memory, etc.) vastly exceeds what any single supervisor can perceive and respond to.

This isn't a character flaw. It's a problem with the structure of your system. And while leadership development may increase the "variety" the leader can handle in this environment, that strategy will not change the underlying math.

Then we need to talk about that training.

You Can't Train Your Way Out of This

The instinctive response to a knowledge gap is a training program. It's also, largely, a waste of time.

I'm sure I'm not the only one who's seen supervisors put on machines, usually in the wake of a disaster, under the guise of training them. This may well give them some knowledge, but it doesn't create standards. And, for the most part, it doesn't change their capabilities.

In the 1880s, German psychologist Hermann Ebbinghaus documented what he called the Forgetting Curve: the rate at which the human brain discards information it doesn't actively use. His findings were stark. Without reinforcement, people forget roughly half of new information within a day. Within a week, retention drops to around 20%. After a month, most people retain almost nothing from a training event they haven't applied in the field.

Now apply that to the problem we've been describing. How does a manager defeat the "Forgetting Curve" to reinforce hundreds of critical checks across multiple supervisors while building new checks every time something goes wrong?

The supervisor's environment is a firestorm of problems, complaints, and stress. How can a supervisor make tasks that he has never performed appear important to the operator?

All of that complexity needs to be checked and repetitively reinforced. But the supervisor is already overtasked, how do you create time for repeated application? The forgetting curve doesn't bend in the supervisor's favor, it runs its full course, every time, because the conditions that would interrupt it simply don't exist for someone in his role.

Training doesn't fix the problem.

This isn't a failure of the training program. It's a failure to understand what training can and cannot do. Formal training transfers declarative knowledge: facts, frameworks and written procedures. It does not transfer the tacit competence that actually drives reliable operation on a factory floor. That knowledge lives in operator hands and habits, built over thousands of hours of direct engagement with the equipment and people.

Expecting a supervisor to acquire equipment-level competence through training is like expecting someone to become a combat pilot because they read the flight manual. The manual is real. The knowledge in it is real. But without the hours in the cockpit, it doesn't transfer and the aircraft will punish you for pretending otherwise.

This isn't the solution to the problem.

The Elevation Trap

Another common poor strategy is to elevate decisions to leaders.

When something goes wrong the natural organizational reflex is to kick the decision upstairs. Get the supervisor. Call the manager. Loop in engineering.

All this actually does is route additional complexity into the node that is already the bottleneck: the leader who is already holding more variety than any one person can process. Every escalated decision that could have been resolved at the floor level is another brick on a structure that is already overburdened. You haven't solved the variety problem. You've exacerbated it.

And the perverse outcome is this: the more you train your workforce to escalate, the more dependent the system becomes on a single point of failure. The leader becomes not a force multiplier but a cork in a bottle, the thing that every decision has to pass through before anything moves. The leader gets buried. The floor stops developing judgment. And when the leader is out, on vacation, or just overwhelmed at the wrong moment, the system has no secondary capacity to fall back on.

Pushing Variety Down, Not Up

The real lever for reducing requisite variety at the leadership level is to distribute decision-making. Push authority, knowledge, and judgment downward to the people closest to the work.

This is not the same as abdication. This is a critical practice to handle excess complexity.

Standard processes. If the right way to handle (or prevent) a common failure mode is documented, trained, and internalized, that decision no longer needs a leader. The decision has already been made. The variety has been absorbed by the system itself, not by the supervisor.

Clear decision boundaries. Operators and technicians need to know what they own and what requires escalation. Not everything should be escalated, escalation triggers should be the exception, not the rule. Clarity at the boundary is what allows authority to live at the floor level without creating chaos.

Distributed competence. A workforce that understands its equipment, knows its processes, and has been trained to recognize and respond to failure modes is a workforce that generates its own variety. Each capable person at the floor level is, in Ashby's terms, a regulator. They absorb the environmental complexity before it ever reaches the leader.

Verification systems, not supervisory surveillance. Without follow-up, your training system will not be effective. Repetition and verification of critical knowledge, settings, and practices are not optional. That supervisor isn't going to successfully bridge the gap of those 20-30 things across those 12 people. Systems can and that's what differentiates poor organizations from great ones.

Together, these don't just reduce the burden on the leader. They change the nature of leadership in a complex environment. The leader's job becomes simple enough to where someone can actually do it.

What This Means for Your Organization

If you're a plant manager, an operations director, or an executive looking at chronic failure patterns in your facilities, the first question to ask isn’t, "who dropped the ball?" It’s "does our leadership structure have the variety to govern the complexity of this environment and are we routing complexity in the right direction?"

If decisions habitually flow upward, if your supervisors are perpetually firefighting, if institutional knowledge lives in individual heads rather than in systems, you don't have a leadership problem. You have a structural one.

The fix isn't a better leader. It's a structure that is capable of success: variety pushed down to the people who can absorb it, verification systems that let leaders govern without omniscience, and a clear-eyed understanding that accountability is only meaningful when the person being held accountable had a realistic path to success.

Leaders deserve to be held accountable but they also deserve to be placed in structures where accountability is achievable. The rest is just blame.