During the last couple of decades working in quality, I’ve lost count of how many times I’ve seen the same pattern play out: A strong launch. Tight focus. Great early results. People doing the right things for the right reasons. Controls are followed. Issues are surfaced quickly. Leaders are engaged. The system works. And then—slowly—cracks in the armor begin to show.

A failure here. A miss there. At first, they seem isolated, even explainable. But over time, a pattern emerges—and when you look closely enough, you realize those failures all trace back to the same place: the basics.

Tool wear that went a little too long. Preventive maintenance that slipped. Checks that became routine instead of intentional. Shortcuts that felt harmless in the moment. Tribal knowledge used to stretch equipment and tolerances. Competing priorities pulling attention elsewhere.

Nothing dramatic. Nothing malicious. Just drift.

Eventually, you find yourself staring at what was once a best-in-class launch and asking the most uncomfortable question in quality:

How did we get here?

This isn’t incompetence. It isn’t laziness. And it certainly isn’t a lack of commitment from the people doing the work.

It’s entropy.

Not the thermodynamics textbook definition but the operational reality that systems, left alone, naturally move from order to disorder. Quality Digest contributor Harish Jose has written directly about entropy in manufacturing, cautioning against simplistic definitions of “order” and “disorder,” while reinforcing the idea that systems degrade unless they are actively managed.

The point is practical: Quality rarely collapses overnight. More often, it fades quietly—predictably. Unless we deliberately design systems to fight that drift, entropy always wins.

Entropy is the default setting

Every organization knows how to inject energy into a system—at least temporarily.

We do it during launches, escalations, customer crises, and recalls. We do it when leadership attention is high and consequences are visible. In those moments, quality performance improves rapidly because focus is intense. Standards matter again. Controls are followed. Problems surface early. Leaders show up. The system tightens.

But energy is finite. People rotate to new programs. Engineers move on. Leaders shift attention. Budgets tighten. New priorities emerge. And as attention moves, discipline begins to soften.

That’s when familiar warning signs appear:
• Control plans turn into paperwork exercises
• Audits become compliance theater
• SPC charts get filled out but never reviewed
• Reaction plans are bypassed under pressure
• Training becomes something people “used to do”
• “Good enough” quietly replaces “do it right”

Entropy doesn’t arrive with a press release. It slips in gradually, through comfort and familiarity.

W. Edwards Deming’s work is often remembered for tools and slogans, but the deeper message is structural: Systems don’t self-manage. A system requires management and coordinated leadership to sustain performance.

The uncomfortable truth is that you don’t lose quality because people stop caring. You lose quality because you stop actively protecting it.

The real enemy: ‘Set it and forget it’

One of the most dangerous phrases in quality is, “We already fixed that.”

There’s no such thing as a permanent fix. There are only fixes you’re still maintaining.

This is where many organizations unintentionally sabotage themselves. We celebrate the improvement, document the change, and move on, assuming the system will hold. But improvement that isn’t protected becomes yesterday’s story.

This is also where the language of continuous (or continual) improvement matters. Deming’s community has been careful to emphasize that “support” and “commitment” aren’t enough; leaders must understand what to do and actively carry obligations that can’t be delegated.

The meaning is simple:
• Every system needs reinforcement
• Every process needs revalidation
• Every standard needs protection
• Every control needs ownership

When improvement is treated like a one-time event rather than a living system, decay is inevitable.

What anti-entropy really means

Anti-entropy isn’t motivation. It’s not slogans. It’s not telling people to “try harder.”

Anti-entropy is system design.

It’s anything that continuously re-injects attention, discipline, learning, and accountability back into the work—especially after the spotlight has moved on.

This aligns directly with the Shingo Institute’s model: Beliefs and systems drive behavior, and the systems that govern how people work shape outcomes over time.

Anti-entropy systems are the difference between hoping people remember and building systems that don’t rely on memory. They’re the difference between local heroics and structural reliability. They’re the difference between reacting to failures and preventing drift.

Anti-entropy isn’t about eliminating problems. It’s about ensuring that when problems appear, they lead back to stronger systems, not just temporary fixes.

Where entropy shows up first

Entropy rarely announces itself as a major failure. It shows up in small, excusable decisions:
“We’ll do the PM next week.”
“Just this once, skip the check.”
“The chart looks fine—file it.”
“Everyone already knows how to do this.”
“We don’t need to escalate yet.”

Each decision feels reasonable in isolation. But entropy compounds.

Over time, standards become flexible. Controls become optional. The system becomes dependent on experience and judgment rather than structure and discipline.

This is why the “if it ain’t broke, don’t fix it” mindset is so dangerous in quality. Toyota thinking challenges the assumption that the absence of problems is a sign of health. In fact, when abnormalities stop surfacing, it often means the system has lost its ability to detect drift.

Donald Wheeler’s recent Quality Digest piece reinforces this idea by challenging the notion that operating OK is a reason to leave a process alone. Stability without visibility creates false confidence, and false confidence accelerates entropy.

Eventually, people start saying: “We used to be really good at this.”

That statement is the clearest signal that entropy has taken hold.

Why controls fail over time

Control plans, in-process checks, SPC, audits, and layered process audits are all legitimate anti-entropy tools. They exist to verify discipline, catch drift early, and stabilize performance. In theory, they should protect the system long after launch.

Toyota’s view of standardized work helps explain why they often fall short in practice. Standardized work isn’t the goal. It’s the baseline. Its purpose isn’t to preserve a process indefinitely, but to make abnormality visible so improvement can occur. When standards are no longer questioned, reinforced, or evolve, they quietly degrade into ritual.

They are necessary, but they are often not sufficient.

One reason is that most controls are designed as local defenses, not sustaining systems.

A local control answers the question, “Did we check the part?”

A sustaining system asks a different question. “Is the checking system itself still healthy, relevant, and sufficient for the risk it’s meant to control?”

That second question is where many organizations struggle.

Controls themselves decay over time. Checks become routine rather than intentional. Reaction plans become theoretical. Charts become decorations. Audits become formalities. When no one owns the ongoing health of the control system, entropy quietly takes over.

This weakness becomes even more dangerous when combined with a second, less visible failure mode: undeclared key characteristics.

One of the most significant sources of latent quality risk isn’t a broken control but a missing one—specifically, when a characteristic that should be designated as key or critical is never formally identified as such.

Toyota’s concept of jidoka is fundamentally anti-entropy. Jidoka forces processes to stop at the first sign of abnormality, preventing deviation from flowing downstream and accumulating quietly over time. When a characteristic isn’t elevated as critical, the system has no obligation to stop, signal, or escalate when drift begins.

Early in a launch, everything may appear stable. Capability looks acceptable. Output meets requirements. Because nothing is failing, the characteristic continues to be treated as routine rather than protected as managed risk.

At launch, nothing breaks. But the risk is already embedded.

Without formal designation, the system doesn’t actively protect that characteristic. It doesn’t receive heightened attention. It’s not monitored with increased rigor. It’s not defended when pressure increases or priorities shift.

Over time, drift caused by tool wear, maintenance variation, subtle process changes, or undocumented tribal knowledge accumulates quietly. The failure doesn’t announce itself. It matures.

When it finally surfaces, it often does so in the most damaging way: in the field, across entire model-year populations, and at a specific point in product life. At that moment, the discussion is no longer about internal quality metrics. It becomes about warranty exposure, recall risk, regulatory scrutiny, and reputational damage.

In these cases, organizations often discover that control plans were technically followed, audits were passed, and checks were performed. And yet the system still failed.

This is the uncomfortable truth: Controls can’t protect what the system never explicitly chose to protect.

Audits are intended to function as anti-entropy mechanisms, but they suffer from the same vulnerability. ISO, IATF, internal audits, and layered process audits fail when they degrade into compliance rituals rather than learning mechanisms.

In organizations where audits are treated as check-the-box exercises, they reinforce the illusion of control while allowing drift to continue. Findings are closed administratively. Root causes remain shallow. The system learns how to pass audits rather than how to improve.

In contrast, organizations with a strong quality mindset use audits as early warning systems. Leadership treats audit results with the same seriousness as financial metrics. Findings trigger system-level discussion. Corrective actions focus on strengthening controls and reassessing risk, not simply closing paperwork.

The difference isn’t the audit standard. It’s the organizational mindset.

This is where Joseph Juran’s thinking remains highly relevant. The Juran Trilogy—planning, control, and improvement—makes it clear that improvement without ongoing control is temporary. Control isn’t documentation. It’s the active protection of gains over time.

Put differently, improvement without a living control system is simply borrowing time from the future.

Anti-entropy thinking forces a harder and earlier question: What are we choosing to protect, and what are we implicitly leaving exposed?

If that question isn’t being asked deliberately, the organization is relying on hope rather than design.

The anti-entropy framework: Four reinforcing layers

To fight entropy at scale, quality must be designed as an operating architecture, not a collection of tools.

An effective anti-entropy system consists of four reinforcing layers.

Layer 1: Local controls; discipline at the point of work

This is where most quality systems begin:
• Control plans and reaction plans
• In-process inspections
• SPC or process behavior charts (when data support it)
• Poka-yoke and mistake-proofing
• Measurement system discipline

This layer fights entropy where it originates—at the process, the machine, and the operator interface.

But local controls are fragile. They depend on stable conditions, human attention, and consistent leadership—all of which entropy erodes over time.

Philip Crosby’s philosophy offers a solid anchor here: Quality comes from prevention, not detection. That prevention mindset is what control plans and in-process controls are supposed to enable, when they are treated as living mechanisms rather than forms.

Layer 2: Visibility; making drift impossible to ignore

Entropy thrives in darkness. If drift is invisible, it will continue.

Visibility means designing data and feedback so risk becomes obvious early:
• Metrics that highlight distance-to-spec, not just pass/fail
• Trends that show direction, not just today’s result
• Dashboards that prioritize risk instead of displaying everything
• Comparability across processes, suppliers, and plants

Visibility doesn’t fix problems. It exposes them. And exposure is often the first—and hardest—step toward improvement.

This is also where “digitization” earns its keep when done correctly—not as a prettier report, but as a mechanism that makes drift and chronic instability visible enough to trigger action.

Layer 3: Governance; designing attention on purpose

Governance is the most overlooked anti-entropy mechanism. It answers questions like:
• Who owns stability after launch?
• Who reviews drift and how often?
• What thresholds trigger action or escalation?
• Who has authority to strengthen controls?

Governance isn’t bureaucracy. It’s attention by design.

Toyota places clear responsibility for system stability on leadership. Leaders aren’t responsible for reacting to problems alone, but for designing and protecting the systems that prevent them. That responsibility can’t be delegated. When leadership ownership fades, governance decays into reporting, escalation becomes political, and entropy fills the gap.

With governance, trends are reviewed on cadence, ownership is explicit, and decisions are made early, before problems become failures.

Layer 4: Consequence and reward; making sustainment matter

This layer determines whether entropy wins or loses. If nothing happens when discipline fades, discipline will fade.

Sustainment must have consequences—and rewards. These can include:
• Recognition tied to sustained performance, not heroics
• Business advantage tied to stability and prevention
• Clear consequences for chronic drift
• Evidence-based certification or recertification
• Leadership attention tied to behavior, not just results

This layer is where prevention becomes real, because it starts to affect opportunity, resources, and trust—echoing Crosby’s point that the cost is in nonconformance, not in doing quality right.

Problems are signals, not just failures

A mature quality system asks more than, “What went wrong?” It also asks, “What was supposed to prevent this, and why didn’t it?”

Every meaningful failure is evidence that the system needs strengthening. Not just correction. Strengthening. That means fixing:
• The defect
• The process
• The control
• The governance
• The incentive

This is also consistent with the lean/Toyota view that standardized work isn’t the end point—it’s the baseline for the next improvement. The Lean Enterprise Institute has stated it plainly: Today’s standardized work is the basis for tomorrow’s kaizen.

If problems don’t feed back into system design, entropy continues unchecked.

What anti-entropy looks like in practice

Anti-entropy shows up in everyday decisions:
• Ultratight tolerances are treated as managed risk, not just print notes
• High-risk processes get stronger controls, not hope
• Low-frequency data are visualized as risk, not noise
• Control plans are audited for health, not existence
• Drift triggers escalation automatically
• Stability earns trust, opportunity, and advantage

It also changes the questions leaders ask:
• “Is this process still stable?”
• “Is this control still working?”
• “Who owns protecting this?”
• “What would cause this to decay?”

These questions shift quality from reaction to prevention. They also align with a practical kaizen mindset: Continuous improvement assumes the system is never “done.” Taiichi Ohno’s kaizen framing is blunt: If you assume things are all right as they are, you can’t do kaizen.

There is no permanent improvement

The most dangerous belief in quality is that improvement is finished. To the contrary:
• Every system decays without maintenance.
• Every standard softens without reinforcement.
• Every control weakens without ownership.

There is no finish line. There is only sustainment or decay.

Entropy is natural. Drift is predictable. Quality loss isn’t mysterious. What’s rare is designing systems that actively fight that drift—systems that assume attention will fade and compensate for it by design.

Excellence isn’t accidental. It’s designed, maintained, and defended, not by slogans, but by systems. Entropy is natural. Excellence is intentional.