Every repair is a fork in the road: restore it, or improve it.

Why this book still matters

In the late 1980s, Winston P. Ledet was part of a corporate team at DuPont trying to answer a blunt question: why does maintenance cost so much and deliver so little? Instead of another benchmarking binder, the team built a system-dynamics model of how maintenance actually behaves, fed with experience from roughly 2,000 DuPont sites and production units. The model kept spitting out an answer nobody in the maintenance department wanted to hear: plants that get faster at repairs stay on the same treadmill. Repair speed optimizes the loop; it never shrinks it. The only lever that kept paying, year after simulated year, was eliminating the defects that create failures in the first place.

That modeling work became The Manufacturing Game, a hands-on simulation workshop that has since put tens of thousands of operators, craftspeople, engineers and managers through the experience of running a plant proactively instead of reactively. The 2009 book by Winston P. Ledet, his son Winston J. Ledet and Sherri M. Abshire distills what those decades taught: there are three stable ways to run maintenance, and the road between them runs through behavior, not budget.

Reactive is where the crusher runs to failure and the crew that saves the shift at 2 a.m. gets the coffee-room glory. It feels busy and heroic, and it is stable precisely because heroism is rewarded and nobody has time to prevent anything.

Planned is the world Doc Palmer's handbook builds: planners, schedules, kitted jobs, wrench time up. It delivers a genuine step change - and then it plateaus, because planning organizes the response to defects without reducing their arrival rate. If you are building this domain, start with our Palmer summary.

Precision is where the defects themselves get hunted: couplings laser-aligned every time, lubrication clean and right, procurement specs that reject the cheap seal, operators who run equipment inside its limits. Failures become rare enough that the remaining ones get real engineering attention.

The uncomfortable part: you don't drift between domains, you jump. Each one is a stable attractor - its habits, rewards and daily pressures pull you back. Buying a CMMS does not make you Planned, and buying vibration sensors does not make you Precision. Shelfware is what you get when a site buys the tools of the next domain while keeping the behaviors of the current one.

Defects enter the plant through roughly four doors: operating practices (the pump dead-headed against a closed valve, the crusher fed oversized ore), maintenance-induced defects (the bearing hammered onto its shaft, the coupling aligned by eyeball), parts and procurement quality (the bargain seal that lasts a third as long, the wrong elastomer in stores), and original design (the compressor suction that was always too close to the knockout drum). No single department controls all four doors - which is why the book insists defect elimination belongs to operators, craftspeople, buyers and engineers together.

The delivery mechanism is the cross-functional Action Team: a small mixed crew that picks one defect type, kills it, banks the win and picks the next. One team traces conveyor idler seizures to a greasing practice; another ends a family of compressor valve failures by fixing suction filtration; a third rewrites the spec so the cheap seal can never be purchased again. None of these wins is spectacular alone. Compounded across a year, they are the difference between domains.

The moment a machine fails is the cheapest moment it will ever be to improve it. The machine is already open. The failed parts are in your hands, telling you exactly what killed them. Production has already conceded the downtime. The pain is fresh enough that people will actually change a procedure. Wait three weeks and all four advantages are gone.

So treat every corrective work order as an upgrade opportunity, not just a restoration. The seal failed on the slurry pump? While it's open: fit the better seal and flush plan, correct the pipe strain that was loading the casing, and write the minimum-flow limit into the operating procedure so the defect source itself is gone. The conveyor pulley bearing failed? Align and balance to spec on reassembly instead of "good enough", and fix the wash-down practice that was flooding the labyrinth seal. Same wrench time, completely different future.

This is also why precision beats detection. Condition monitoring catches defects sliding down the P-F curve; precision means fewer defects ever start the slide. Detection is a safety net - elimination shrinks what the net has to catch. For the detection side of the story, see our P-F & D-I-P-F summary.

Nobody firefighting a plant believes they have time for proactive work - and inside their daily experience, they are right. Every hour spent on defect elimination is an hour taken from a crisis that is real today, for a benefit that arrives in months. A presentation cannot beat that lived experience. A simulation can, because it compresses time: in The Manufacturing Game, a team runs a simulated plant for about three years in a day. They firefight, fall behind, then try eliminating defects - and they feel the workload collapse as failures stop arriving. The payoff that takes eighteen patient months in real life takes one round at the table.

The numbers from the DuPont modeling point the same way: eliminating defects cut failures dramatically in the model, while pure repair-speed optimization barely moved availability at all. Treat that as a modeling insight rather than an audited plant result - but three decades of Game workshops and defect elimination waves at real sites keep confirming the direction. People who have experienced the loop go home and start Action Teams. People who have seen a slide deck go home and forward it.