Quality and You..

More With Less

Why quality is always a phenotype — and why we can’t help but seek it

A note before we begin. I was asked to give a talk inside Zebra about materials. What follows grew out of that assignment and then quietly outgrew it. It came together the way good thinking usually does for me — after a solid night’s sleep and a long morning walk with Hero, fed by whatever I’d been reading lately, and sharpened by debating the ideas with several AI partners patient enough to push back. I’m sharing it with anyone who’s curious.

Hero, true to his name, isn’t camera shy — he just has no interest in being seen as a showman.

Phenotype.

If that word just made you stop, or roll your eyes a little — good. You’re in the right frame of mind. A year ago it would have done the same to me. I’d have figured the author was reaching for something clever, dressing up a conversation about product quality to sound smarter than it needed to. Fair reaction. Hold onto it.

So why am I using it?

It came out of a question that has nothing to do with scanners or drop specs: how do we — engineers, designers, teams, people — actually learn and innovate? When I went looking for first principles, I kept landing in the same territory: evolutionary biology, neuroscience, the study of living systems. And the idea that stuck with me is almost embarrassingly simple. The environment is not scenery. It is a participant. It helps decide what survives, what gets selected, and what quietly disappears — in biology, and in everything we build.

Biology already has clean language for this, and it’s worth borrowing. The genotype is the set of instructions — the DNA. In our world, that’s the bill of materials, the spec, the tolerances, the design rules. The phenotype is what those instructions actually become once they meet a real environment. Same instructions, different conditions, different outcome. Genotype is what a thing was told to be. Phenotype is what it became.

Apply that to quality and it does real work right away.

Quality is not a property of the instructions. It is a property of what the instructions become when they meet the world.

The environment gets the last word

I’ve come to treat the things we build as living systems the moment they leave the factory — not as a metaphor I’m fond of, but as the most accurate way I’ve found to think about what happens next. And in living systems, as Denis Noble keeps insisting, there is no privileged level of causation. Cause runs up and down the ladder, not just outward from the genes.

Watch how that plays on the bench. A material is a phenotype of its own chemistry and processing history — and a genotype to the part molded from it. The part is a phenotype to the material and a genotype to the assembly. The assembly is a phenotype to its parts and a genotype to the warehouse, the truck, the hospital floor. Genotype and phenotype aren’t fixed levels. They’re roles, and the same object plays both at once — quality expressed to the layer above, cost paid to the layer below.

It’s turtles all the way down, and the honest response to that is not to panic. You don’t have to model quarks to pass a drop test. The discipline is to stand where the failure mode actually lives, and recurse downward only as far as the evidence demands — far enough that the next build survives the next real environment at lower total cost than the last one. That’s evolution. It’s also just good engineering.

Which is why I say absolute quality doesn’t exist. Not as a philosophical shrug — as a structural fact. A phenotype with no environment is undefined. Quality only shows up when something meets the actual world it was made for.

Overkill dies quietly

The failure mode I worry about most isn’t the dramatic one. It isn’t the cracked housing on the test bench. It’s the slow, invisible death by excess cost.

Over-engineering is a metabolic burden — energy, material, complexity, and validation effort spent on performance the real environment will never reward. The lean design that is also good enough doesn’t beat it on the bad day. It beats it on every ordinary day, by being cheaper to keep alive.

Picture two devices doing the same job in the same place. One carries the accumulated expense of aerospace-grade everything, and tolerances tightened “to be safe.” The other was built to the spec the field data actually justified. In the real environment they survive at the same rate. In the purchasing system — and in the worker’s hand, over thousands of cycles — one of them loses. The phenotype that gets selected is the one whose energy ledger matched reality.

We don’t call this failure. We call it diligence. We praise it in design reviews, and we promote the people who added the extra margin. Only later — when the leaner competitor takes the business, or when the cumulative weight and cost finally surface — do we notice that quality was never the expensive version. Quality was the version the environment, and the human holding it, could afford to keep.

The specification is you choosing a world

Every drop spec is a philosophical act, whether you treat it like one or not.

First, what I mean by that word — distribution — because the whole argument leans on it. A distribution isn’t a single number. It’s the full range of what actually happens to a thing out in the world, and how often each happens.

Think about how you dress in the morning. You don’t dress for the hottest day on record, or the coldest. You dress for the range you’ll realistically meet, with a little margin for a cold snap — and you don’t haul a parka around in July for a blizzard that isn’t coming.

A device lives the same way. Most of its life is ordinary: set down, picked up, carried. Some of the time it’s dropped from waist height onto concrete. Once in a great while it goes off the back of a forklift. That whole spread — the ordinary, the occasional, and the rare — is the distribution. Good design covers the part that actually happens, including the rough days frequent enough to matter, and stops there. It doesn’t pay to survive the once-a-decade catastrophe, and it doesn’t pretend every day is a gentle one.

When you write that a device survives a one-meter drop onto concrete with no functional damage, you’re not describing a test. You’re defining a niche. You’re saying: this is the tail event I’m willing to pay to survive. And you’re saying, by omission: I’m not paying for the meter-and-a-half drop, or the corner impact after three years of chemical exposure, or the fall onto a housing already fatigued by real use. Those are different worlds. Each one asks for a different phenotype.

Over-specify, and you have paid for a world that does not exist. Under-specify, and you get selected against the day the real one arrives.

The whole art is setting the energy budget — mass, gravity, height, and how that energy gets managed once it lands — so the thing clears the actual distribution, rare bad day included, without carrying the cost of events that almost never happen. That’s not compromise. That’s precision.

Every other requirement works the same way: chemical resistance, vibration, thermal cycling, the abuse a device takes from people who don’t read manuals. Each is a claim about which slice of the environment’s distribution you’ve decided to treat as real. The phenotype you get is the one selected for the world you actually described — not the world you were afraid of.

Materials tell you what they’ve been through

At the bottom of the ladder, where we pay for quality in chemistry and processing history, materials aren’t passive either. They’re phenotypes too. A polymer housing is the expressed behavior of its molecular weight, its additives, its molding conditions — judged by the drops and chemicals and stresses it will actually meet.

I’ve spent enough time reading failed parts to trust them as witnesses. Stress whitening, crazing, discoloration — those aren’t only defects. They’re the part telling you its history, the loads the genotype couldn’t fully anticipate, written right into the surface. Choose a material that clears the real distribution with minimal over-design, and you aren’t just buying chemistry. You’re choosing a story the part can keep telling without failing.

This is why the long game matters, and why the cheapest line on the bill of materials is so often a lie. A material that eliminates a field failure for a decade isn’t a cost center; it’s proof that the energy you spent at the genotype level produced a phenotype the environment kept. A cheaper resin that quietly reintroduces an old failure mode isn’t a saving. It’s a phenotype the market will reject later — usually after the warranty has expired, and the reputational damage has already started.

Tools that conserve judgment — or consume it

The same rule that governs the device governs the tools we use to design it.

Simulation, generative design, AI assistants — these are powerful genotype modifiers. They explore spaces no human could enumerate by hand, and they surface failure modes early. They can also become their own kind of overkill: judgment spent on ever-finer models of events that will never happen, while the one habit that actually matters — asking whether the model is still tethered to the real distribution — quietly weakens.

The discipline doesn’t change. Minimal energy, just good enough, even in the analysis. Run the tool until it has cleared the failure modes that matter, then stop. The goal was never a perfect digital twin of reality. It was a decision good enough to let the next physical build survive the next real environment at lower cost. When a tool starts optimizing for its own metrics — mesh density, simulation count, confidence intervals detached from field data — it has stopped conserving judgment and started consuming it. The phenotype of an engineering culture that lets that happen is impressive diligence nobody can afford.

We don’t model quarks to pass a drop test. We stand where the challenge lives, and we go only as deep as the failure mode requires. That holds for the tools we build, and the tools we lean on — because leaning on a tool is exactly the move: it carries part of the load so you don’t have to. The only question worth asking is how much of your own load you can hand off before you’ve forgotten how to carry it at all.

Why we can’t help but seek it

Here’s the part I find hard to shake. The drive toward quality isn’t a professional virtue we bolt on top of competence. It looks like the same drive that runs through every living thing: spend as little energy as you can get away with, while still clearing the challenges that actually matter.

There’s a strand of neuroscience — Karl Friston’s free-energy principle is its most ambitious version — that says the brain itself works this way, building models just good enough to minimize surprise without burning more resource than the world warrants. I’ll be honest that it’s a contested idea, and I’m not resting the argument on it. You don’t need the grand theory to feel the pull. We are the kind of creature that can look at a design and sense, in the body, that it’s carrying too much. We can watch a process and feel it curdle from selection into ritual. That sensitivity isn’t mystical — under real competition and real scarcity, the lineages that wasted energy tended to lose to the ones that didn’t, and we are the descendants of the ones that didn’t. Engineering is just that ancient pressure, made deliberate.

That’s why quality feels human even when the object is a scanner, or a medical housing. We’re not only optimizing a product. We’re taking part in the same selection that made us — choosing, on purpose, which phenotypes get to continue, and which get quietly discontinued. That’s not only a technical responsibility. It’s closer to an existential one.

The only quality that matters

There is no absolute quality. There is only the phenotype that emerges when a particular set of instructions meets a particular world, and is judged fit enough to continue. The art — in engineering, and in the older human work of making things — is to set the energy budget so the thing clears the real distribution without paying for a distribution that never arrives. Materials are where you pay. The field is where you learn whether the payment was wise. The tools are how you explore the space between, without exhausting yourself or your judgment.

The worker who reaches for a scanner at the start of a shift is doing something very old, in a modern form: picking the tool whose cost matches the world it will actually meet. One of those tools will still be worth making in ten years. The other will have been replaced by something leaner, cheaper, and just as sufficient — and that replacement won’t be a loss. It’ll be quality, doing what quality always does: conserving energy while still reaching the goal.

That’s the whole of it. Quality is how we take conscious part in the ongoing selection of what deserves to continue. Not perfection — the fit. The honest, disciplined fit between what we build and the world that will judge it, and the refusal to pay for worlds that will never come.

Joseph P. McFadden Sr.

Stratford, Connecticut  ·  2026