What We Mean by Engineering

We borrow the word engineering for the prestige and skip the thing that actually makes it one.

In the 1930s, aircraft designers could calculate a great deal. They could work out the lift a wing would generate and the drag it would suffer, the loads a structure would carry, the power a propeller would deliver. What they could not do, with any of that, was say what made an airplane good to fly. Two aircraft with similar performance on paper could behave completely differently in a pilot’s hands, one steady and forgiving, the other twitchy and exhausting and quietly dangerous, and nothing in the physics of flight told you which you had built until a person took it up and found out. The knowledge of what made an airplane controllable, of how much a control should resist the pilot’s hand and how the machine should respond when disturbed, did not exist. It had to be created, slowly, by engineers who flew the airplanes, recorded what felt wrong, argued about it, and gradually turned a pilot’s vague sense of a bad airplane into specifications a designer could build against.

I find that history clarifying, because it cuts against the picture most of us carry of what engineering is. We tend to imagine that the engineers already had the answer in the physics and merely applied it, that flying qualities were somewhere in the equations waiting to be looked up. They were not. The knowledge of how to make an airplane a pilot could trust was engineering knowledge, made by engineers, and it was no less rigorous for not being physics. Understanding why it was not physics, and what it was instead, turns out to be the key to the whole question of whether go-to-market could ever be an engineering discipline, because the thing the field has borrowed is the word, and the thing it has skipped is exactly the kind of knowledge those aircraft designers had to build.

The picture in our heads

Ask most people what engineering is and you will get some version of the same answer. Engineering is applied science. The scientists discover the laws of nature, and the engineers take those laws and apply them to practical problems, the way you might take a theorem and use it to solve a homework question. On this picture, science is the source of the knowledge and engineering is the delivery, science with a deadline and a budget and a tolerance for getting your hands dirty. The engineer is a kind of skilled translator, fluent in the language of physics, who turns established theory into bridges and engines and aircraft.

The clearest way to see what is wrong with the picture is to notice how often the order runs the other way, with the working machine arriving long before the science that explains it. The steam engine is the great example. Practical steam engines were built, sold, and steadily improved for the better part of a century before there was a science of thermodynamics to say why they worked, and the science, when it finally came, was developed in part by people trying to understand the engines that engineers had already built and were already selling. Sadi Carnot worked out the fundamental limit on an engine’s efficiency by reasoning about machines that were already pumping water out of mines. The engineering did not wait for the physics. The engineering came first and the physics followed it, and that sequence, which would be impossible if engineering were only the application of science, turns out to be the normal sequence far more often than the tidy story admits.

This picture is intuitive, it is taught implicitly in a hundred small ways, and it is wrong in a way that matters. It is not that science is irrelevant to engineering, because it plainly is relevant, and engineers use it constantly. It is that the picture gets the relationship backward and leaves out most of what engineers actually know. The historian Walter Vincenti spent a career documenting this, working through case after case from aeronautical history, and what he showed is that engineering generates its own knowledge, a body of understanding that science does not contain and could not have supplied, and that this knowledge has a structure of its own. The flying-qualities story is one of his cases, and it is typical rather than exceptional. Over and over, engineers turn out to know things that no scientist discovered, because the things were never going to be discovered by studying nature. They had to be worked out by people trying to build something that worked.

The reason this matters here is that when a field borrows the word engineering, it is usually borrowing the prestige of the applied-science picture, the suggestion that what it does is rigorous because it rests on established laws. That is the wrong thing to borrow, because it is not where engineering’s rigor comes from. The rigor comes from the other thing, the hard, specific, accumulated knowledge of how a particular kind of made object behaves, and that knowledge is exactly what a young field does not yet have and has to build for itself.

The picture persists for understandable reasons. Engineers are trained heavily in the sciences, because the sciences are a real part of the toolkit, and a student can come away from that training believing the science was the main event and everything else was a matter of application. The prestige runs the same direction, since pure theory has long carried more intellectual status than the knowledge of how to make things, so the made-knowledge gets quietly recoded as mere know-how, beneath the dignity of the laws it supposedly applies. The effect is that the most distinctive part of engineering, the part that is hardest to build and most valuable to hold, becomes nearly invisible, dismissed as the unglamorous residue left over once the real science has been done. When a new field then borrows the word engineering, it inherits this blind spot, reaching for the laws and the rigor it imagines sitting underneath them, and overlooking the thing it most needs, which is the made-knowledge that no underlying science is going to provide.

What engineers actually know

If engineering knowledge is not just applied science, what is it? Vincenti’s answer is that it is many things at once, and that the variety is part of the point. Some of it is theoretical, the mathematical tools and intellectual concepts engineers reason with. Some of it is hard data, the measured properties of materials and components that no theory will give you and that someone simply had to go and determine. Some of it is what he calls practical considerations, the accumulated lore of what tends to go wrong and what tends to work, knowledge that often cannot be written down cleanly at all and lives in the judgment of experienced people. None of these is reducible to the others, and none of them is simply read off from science.

Take the category Vincenti calls practical considerations, because it is the easiest to underrate. These are the things experienced engineers know about how their materials and methods behave in the real world, the quirks that never make it cleanly into a textbook because they resist being written down. A machinist knows how a particular alloy will behave under a particular tool in a way that cannot be fully transmitted except by standing at the machine beside them. An experienced designer knows which elegant solutions tend to cause trouble in manufacturing and which ungainly ones quietly hold up, and much of that knowledge is tacit, carried in judgment rather than in any formula. A discipline holds an enormous amount of this, accumulated across many people and many years, and it is a real part of what separates a design that works on paper from one that works in the world. It is also, tellingly, the kind of knowledge that takes the longest to build, because it can only be earned by doing the thing many times and paying close attention to what went wrong, which is exactly the patient accumulation a young field has not yet had the years to do.

Two of his categories are worth holding onto, because they turn out to be the ones go-to-market most conspicuously lacks. The first is what Vincenti, borrowing a term from the philosopher Michael Polanyi, calls the operational principle of a device, which is the basic idea of how the thing works, the account of how its parts combine to do their job. The operational principle of a propeller is that a rotating blade acts as a wing and produces thrust. The operational principle is not a law of physics. It is a human idea about how to arrange matter so that a purpose is served, and physics will happily describe a propeller once you have invented one, but physics will never hand you the propeller, because the operational principle is a thing engineers conceive rather than a thing scientists find.

The second is the normal configuration, which is the generally agreed shape an operational principle settles into once a field has worked with it for a while. When you picture an airliner, you picture a tube with swept wings, engines slung beneath them, a tail at the back. That arrangement is the normal configuration of a large jet aircraft, not the only one physically possible, only the one the field has converged on, over time, as the sound way to embody the operational principle. A working engineering discipline carries a library of these, the settled forms that experience has shown to work, and an engineer designing a new airplane mostly works within that library, varying and refining rather than reinventing.

These two ideas, the operational principle and the normal configuration, are quiet, and they are the foundation of everything. They are what let an engineer begin a new design already knowing, in broad terms, what the thing should be and how it should work, so that the work becomes the disciplined refinement of a known form rather than a leap into the dark. They are also, when you go looking for them in go-to-market, almost entirely absent, and that absence is the most precise way I know to say what the field is missing.

How the knowledge gets made

It is worth saying something about how this knowledge comes into being, because the methods are as distinctive as the knowledge itself, and they look nothing like the popular image of a scientist at a bench. One of the central ways engineers learn is parameter variation, the patient business of building or testing many versions of a thing with one feature changed at a time, watching how the behavior shifts, until a reliable relationship emerges that no theory predicted and no theory was ever going to supply. The data tables at the heart of many engineering fields were assembled this way, by people who needed numbers that nature was not going to volunteer and who went and measured them, methodically, across the range that actually mattered. Another way is proof testing, deliberately taking a thing to and past its expected limits to find out where it really fails, so that the failure happens on a test rig under careful observation rather than in the world by surprise. And the deepest source, the one that mature disciplines are quietly built on, is the study of failure itself, the forensic examination of the things that broke, because a failure is an experiment the world has already run for you, at its own expense, and it contains knowledge available nowhere else.

These methods share a character. They are ways of extracting reliable, usable knowledge from the made world by interrogating made things, and they belong to engineering rather than to science, because their purpose is to build objects that work rather than to understand nature. A field that wanted to become an engineering discipline would have to take up methods like these in earnest, would have to vary its parameters on purpose and study its failures forensically and tabulate what it found so the next person did not have to rediscover it. It is worth noticing how little of this go-to-market does deliberately. It rarely runs a true parameter variation, changing one thing and holding the rest still long enough to learn a relationship. It almost never conducts a real post-mortem on the efforts that failed, preferring a reassuring story about why this particular case was unusual and therefore safe to forget. The knowledge that would accumulate from doing these things never accumulates, because the things are not done, and so each effort starts roughly where the last one started, which is the opposite of what a discipline feels like from the inside.

Normal design and the leap in the dark

There is a distinction that follows from this, which the historian Edward Constant drew and Vincenti adopted, between normal and radical design. A design is normal when the engineer knows both the operational principle and the normal configuration, when they are working within a settled understanding of what the thing is and how it works, and the task is to produce a particular version of it, longer-ranged or cheaper or more efficient than the last. A design is radical when one or both of those are missing, when the engineer does not have an agreed account of how the thing should work or what shape it should take, and has to find out while building.

Almost all of mature engineering is normal design. This sounds like a deflation, as though engineering were mostly routine, and in a sense it is, but the routine is exactly the achievement. The reason a new airliner can be designed and built and carry passengers safely on its first commercial flight is that almost nothing about it is being discovered for the first time. The operational principle is known, the configuration is known, the behaviors are tabulated, the failure modes are understood, and the enormous accumulated structure of normal design knowledge means the engineers are refining a deeply understood object rather than gambling. Radical design, by contrast, is rare and slow and failure-prone, because the people doing it are building the knowledge at the same time as the thing, and they do not yet have the operational principle or the configuration to lean on. The early decades of flight were radical design, and they were dangerous and wasteful and full of crashes, which is what radical design looks like.

Now hold go-to-market against that distinction, and the situation becomes uncomfortably clear. Go-to-market has no agreed operational principle for the thing it builds. Ask a room of skilled practitioners how the object actually works, the system that turns a product and a market into durable demand, and you will get a dozen different accounts, most of them lists of tactics rather than principles. It has no normal configuration, no settled library of forms that experience has shown to be sound, only fashions that rotate with the tools. Which means that nearly every go-to-market effort is radical design, undertaken without the knowledge that radical design requires, and conducted as though it were routine. People launch into new markets believing they are doing normal design, refining a known approach, when in fact they are improvising the operational principle and the configuration on the spot, every time, and then they are surprised by the variance in outcomes, by the crashes, which are simply what radical design produces. The field experiences as bad luck what is actually the predictable result of doing radical design without admitting it.

You can see the confusion in how launches are planned. A team treats entering a new segment as a known motion, allocates a budget, and projects a funnel using conversion rates borrowed from a different segment where they happened to hold, then presents the whole thing with the confidence of normal design, as though the operational principle for the new market were settled and the only open question were execution. What actually happens is that the operational principle was never established for this market, the borrowed rates do not transfer, and the team discovers somewhere in the middle that it has been improvising the entire structure under the impression that it was following a recipe. The plan had the form of normal design and the substance of radical design, and the distance between those two is where the budget quietly goes. None of this means the people are not skilled. It means skilled people are being asked to do radical design with tools and expectations built for normal design, which is a setup for exactly the kind of failure that gets misread afterward as a problem of effort or talent.

Knowledge built under constraint

There is one more feature of engineering knowledge that the applied-science picture hides, and it is the one that gives the whole thing its character. Engineering knowledge is made under constraint, by people who have to act before they have certainty, and who are answerable for what they build. This is the condition that shapes the knowledge, not a footnote to it.

A scientist studying a phenomenon can wait. They can decline to draw a conclusion until the evidence is sufficient, and the cost of waiting is usually only time. An engineer cannot wait, because the airplane has to be designed this year with the knowledge available this year, and the choice is never between acting and waiting, it is between acting on incomplete knowledge and not acting at all. This pressure produces a distinctive kind of knowing, organized around acting well under uncertainty rather than around establishing truth. It is why engineering cares so much about margins, the deliberate gap between what a structure is expected to face and what it is built to withstand, which is a way of converting uncertainty into safety. It is why engineering is obsessed with tolerances, the bands within which a thing will still work, and with failure, the study of how things break, because a discipline answerable for what it builds cannot afford to learn about failure only by accident. The applied-science picture misses all of this, because it imagines engineering as the calm application of settled law, when the reality is disciplined action in the permanent absence of complete information, which is a harder and more interesting thing.

The safety factor is the clearest emblem of this. An engineer building a structure does not build it to withstand exactly the load expected, because the expected load is uncertain and the cost of being wrong is a collapse, so they build it to withstand some multiple of that load, and the multiple encodes everything the field has learned about its own uncertainty. It is humility made quantitative, a number that says, in effect, that we know we do not know exactly, so we leave room. There is no equivalent habit in go-to-market, no convention of building in margin against the uncertainty of one’s own assumptions, which is part of why a plan that turns out to be slightly wrong about its central numbers breaks instead of bending, and the people running it are left explaining a collapse they had no language to see coming.

This is the kind of knowledge those aircraft designers were building when they turned a pilot’s discomfort into a specification. They could not wait for a science of flying qualities to be completed, because they had airplanes to build, so they made the knowledge they needed in a form they could act on, with margins and tolerances and a growing catalogue of what went wrong. That is the texture of real engineering, and it is worth comparing plainly against what go-to-market does when it reaches for the same word. The comparison is not flattering, and it is not meant to be, because the gap it reveals is the work to be done.

Knowing it the way an engineer knows a bridge

So here is the question worth sitting with. What would it mean to know go-to-market the way an engineer knows a bridge?

It would mean having an operational principle for the object, an account of how durable demand is actually created and held, agreed clearly enough that practitioners could reason from it rather than trading anecdotes. It would mean having normal configurations, settled forms that experience had shown to be sound, so that most go-to-market work could become the disciplined refinement of understood objects rather than improvisation dressed as method. It would mean knowing the behaviors and the failure modes well enough to design with margins, to say in advance where a given approach would hold and where it would break, and to treat the breaking as something studied rather than something suffered. It would mean, above all, recognizing how much of what the field now does is radical design, and either building the knowledge that would make it normal or being honest that a leap in the dark is what is happening.

There is something freeing in saying it this plainly. A field that admits it is doing radical design can at least do radical design well, with the care and the margins and the humility the situation calls for, instead of pretending to a confidence it has not earned. The aircraft pioneers were not reckless because they lacked a science of flight. The reckless ones were the ones who behaved as though they already had such a science when they did not. The serious ones knew they were working at the edge of what was known and built accordingly, leaving themselves room to be wrong, and that is part of why some of them survived to assemble the knowledge the rest of the field eventually inherited.

None of that exists yet, and saying so is a description of an opening, not a complaint. The aircraft designers did not have a science of flying qualities either, and they built one, because they were unwilling to keep producing airplanes that were dangerous for reasons no one could name. Go-to-market is roughly where they were, producing outcomes it cannot explain, attributing to talent or luck what is really the absence of a body of knowledge, calling itself engineering while skipping the part that would earn the word. The part it skipped is the part worth building, and it begins with the thing the field has never pinned down, the object itself, because you cannot state an operational principle for a thing you have not yet been able to name.

---

References

• Walter G. Vincenti, What Engineers Know and How They Know It: Analytical Studies from Aeronautical History (Johns Hopkins University Press, 1990), especially the treatment of engineering knowledge as distinct from applied science, the categories of engineering knowledge, and the case of flying-qualities specifications.
• Edward W. Constant II, on the distinction between normal and radical design, adopted and developed in Vincenti.
• Michael Polanyi, on the idea of the operational principle of a device, Personal Knowledge (1958).