The Highest Perfection Looks Imperfect
Loading article…
Pull for previous story
The highest perfection is like imperfection, And its use is never impaired.
Tao Te Ching, Chapter 45
In 1997, 3D Realms announced Duke Nukem Forever, the sequel to one of the most beloved shooters of its era. The studio was rich, talented, and famous, and its answer to every question about the release date was the same: “When it’s done.” A year later, a better game engine came out, and the team switched to it, throwing away much of what they had built. Then graphics moved on again, and they rebuilt again. Every time the game got close, the state of the art moved, and the game had to catch up so it would not look dated on launch day. It was never allowed to be merely good, because it had to be the best.
The game finally shipped in 2011, fourteen years later, after a funding dispute with its publisher forced the original studio to lay off the entire team in 2009, and another company finished the job. Reviewers tore it apart. Chasing the perfect game, they had produced an outdated one, and in the meantime an entire generation of good games had shipped and been loved, made by studios willing to call something done.
Nobody stopped 3D Realms. They had the money, the talent, and the fans. The only thing standing between that game and the world was the studio’s own idea of what it should have been.
That is the trouble with perfectionism. It is not a virtue in disguise. It is an enemy in disguise, and a very effective one, because it wears the uniform of the thing you care about most. It does not tell you to stop caring about your work. It tells you to care about it so much that you never finish, never ship, and in the worst case never start.
The most expensive form of perfectionism is the one that keeps you at the starting line.
You know how it sounds. I will start the blog once I have the right theme. I will start the company once I have the full plan. I will start the side project once I have learned the framework properly. I will start exercising on Monday, when the week is clean. Every one of these sentences sounds like preparation. Every one of them is a way of not starting.
The perfect start is waiting for a moment that does not exist. Information is never complete, conditions are never ideal, and you are never fully ready. The only way to learn what a project actually needs is to be inside it, and the only way to be inside it is to begin badly. Your first draft will show you what the essay is really about. Your first prototype will show you which of your assumptions were wrong. Your first ten users will show you what the product should have been. None of that knowledge is available from the starting line, no matter how long you stand there.
In 1991, a student posted to a Usenet group that he was doing “a (free) operating system (just a hobby, won’t be big and professional like gnu).” The first release could barely do anything. It ran on one kind of machine. It was, by any reasonable standard, rough. That rough start became Linux, which now runs most of the servers on the planet, most of the phones in people’s pockets, and every one of the world’s top supercomputers. Imagine if Linus had waited until it was big and professional before showing it to anyone. There would be nothing to show, because the thing that made it big and professional was thousands of people improving the rough version he was brave enough to release.
This is the pattern almost everywhere. Momentum is worth more than polish at the beginning, because momentum is what produces polish. A rough thing that exists can be improved. A perfect thing that exists only in your head cannot be improved at all; it can only be waited for.
So start rough. Start embarrassingly rough if you must. The roughness is not a failure of the start. It is the start.
If you are building a company, a rough start is not just good advice. It is survival. Waiting for perfect costs real money, because every month spent polishing is a month of runway burned before a single customer tells you whether you are right.
Look at Craigslist. Its website still looks like something straight out of the year 2000: blue links, gray text, no photos on the front page, a design any first-year student could improve in an afternoon. By every rule of modern product design it should have died years ago. Instead it has been profitable for decades, with a famously small staff, while slicker competitors with beautiful interfaces and venture funding came and went. It turns out people were never visiting Craigslist for the design. They were visiting to sell a couch, find an apartment, or hire someone to move them, and Craigslist did those jobs. Everything else was decoration.
Craigslist is not an exception. It is the rule, seen from far enough away.
Amazon launched in 1995 as a plain page of text and links that sold books. Airbnb began as three air mattresses on the floor of the founders’ apartment, rented to people attending a design conference in San Francisco. Zappos started with photos of shoes taken in local stores and posted online; when someone ordered a pair, the founder walked to the store, bought them at full price, and shipped them by hand. Early Twitter crashed so often that its error page, a whale lifted by birds, became more famous than most products. Hacker News still looks roughly the way it did when it launched in 2007, and it remains one of the most influential places in tech.
None of these founders waited for perfect. Each one shipped something that would embarrass a perfectionist, and each one learned from real users what to build next. Reid Hoffman put it bluntly: if you are not embarrassed by the first version of your product, you have launched too late.
The lesson for founders is not that design does not matter. It is that your users are judging one thing above all: does this solve my problem? Get that right with an ugly product and they will forgive almost everything else. Get it wrong with a beautiful product and no amount of polish will save you. Spend your scarce early months on the first question. The pixels can wait, and for some companies, as Craigslist shows, they can wait forever.
None of this is an argument for sloppiness. I care a great deal about quality, and I think you should too. Code that is correct, writing that is clear, products that respect their users: these are worth fighting for, and I have little patience for people who use “done is better than perfect” as an excuse to never make anything good.
The difference between seeking quality and seeking perfection is not a difference of degree. It is a difference of kind.
Quality is a direction. It asks: is this better than the last version? Is it good enough for the people who will use it? What is the next most important thing to improve? Those questions always have answers, and every answer moves you forward. Quality is compatible with shipping, because quality accepts that improvement continues after release. Seeking quality is a practice you can do every day.
Perfection is a wall. It asks: is this flawless? That question never has a satisfying answer, because there is always another edge case, another sentence that could be sharper, another pixel slightly out of place. So the perfectionist keeps polishing, and the returns keep shrinking. The first ten hours fix the big problems. The next ten fix the small ones. The next hundred fix things nobody else will ever notice, while the work sits unshipped and helps no one.
Here is the test I use on myself. When I am about to delay something, I ask: who, specifically, will be harmed by the flaw I am fixing? If I can name a real user and a real harm, that is quality, and the delay is worth it. If the only person who will ever notice is me, that is perfection, and the delay is just fear wearing a nicer coat.
Because that is what perfectionism mostly is underneath: fear. Fear of being judged, fear of being seen as less capable than you want to be seen. Shipping means being evaluated, and as long as the work is unfinished, nobody can evaluate it. I wrote about ego recently, and this is one of its quieter tricks. The perfectionist is not protecting the work. The perfectionist is protecting a self-image that cannot survive a visible flaw.
The deepest problem with perfection is not that it is hard to reach. It is that it does not exist.
Perfect relative to what? Every real piece of work is a set of trade-offs. A faster algorithm uses more memory. A simpler interface hides power from experts. A more thorough essay loses the readers who wanted something short. A more secure system is less convenient to use. You can make something perfect along one axis only by making it worse along another, and there is no arrangement of trade-offs that is best for everyone at once. What looks perfect to one user looks broken to the next.
And even if you found that arrangement, it would not stay perfect. Requirements change. Hardware changes. Users change. The code I was most proud of ten years ago now looks naive to me, not because it got worse, but because the world around it moved and I learned more. Perfection, if it existed, would be a snapshot of a moving target, already out of date by the time the shutter closed.
Nature knows this better than we do. Evolution never produces perfect organisms. It produces organisms that are good enough to survive in the environment they happen to be in, with every compromise and leftover that implies. Our own bodies are full of awkward designs: a blind spot in each eye, a spine adapted badly to walking upright, a throat shared by food and air. And yet here we are, the most capable species the planet has produced, built entirely out of good enough.
Once you see that perfection is not a place you can arrive at, the chase loses its grip. You are not falling short of a real standard. You are failing to reach a mirage, and the only sane response to a mirage is to stop walking toward it.
Laozi saw all of this two and a half thousand years ago, and said it in two lines: “The highest perfection is like imperfection, and its use is never impaired.”
The first time I read that, it sounded like a paradox. After enough years of building things, it sounds like an engineering principle.
Think about what actually lasts. The tools that survive for decades are almost never the ones that were flawless at birth. They are the ones that were left open enough to grow. Unix was full of rough edges and odd conventions, and it outlived nearly every carefully engineered system of its era, because its simplicity left room for others to extend it. The internet was built on protocols that are, by design, best effort: packets can be dropped, delayed, reordered. A perfectionist would never have shipped that. And precisely because it tolerates imperfection, it has never stopped working.
That is what Laozi means. Something that looks finished and flawless has nowhere left to go; any change can only damage it. Something that looks a little incomplete still has room. It can adapt, absorb new needs, be repaired and extended. Its use is never impaired, because it was never sealed shut. People work the same way: a person who must appear perfect is brittle, and a person at peace with being imperfect is resilient.
Laozi goes further than saying imperfection is useful. He warns that the pursuit of perfection actively turns against you.
The whole idea rests on one short line in Chapter 40: “Reversal is the movement of the Tao.” Everything that moves toward an extreme begins, at that extreme, to move back. Later Chinese thinkers compressed this into a proverb: when a thing reaches its extreme, it turns into its opposite. An older text, the Lüshi Chunqiu, put it even more directly: “What is complete must become deficient; what reaches the extreme must reverse.”
Chapter 9 of the Tao Te Ching makes it concrete:
Stretch a bow to the very full, And you will wish you had stopped in time. Temper a sword-edge to its very sharpest, And you will find it soon grows dull.
This is the most precise description of perfectionism I know. A bow drawn to its very limit cannot be held there; it either snaps or your arm gives out. A sword-edge tempered to its very sharpest is the one that dulls fastest. And the chapter ends with the cure: when the work is done, stop. The last bit of effort spent chasing the extreme does not just fail to help. It is exactly the part that makes the thing fragile.
Engineers rediscover this all the time. A machine learning model pushed too hard to score perfectly on its training data can overfit: it memorizes the quirks and noise of the examples it has seen, and does worse on data it has never seen. A system tuned to squeeze out the last percent of performance becomes so tightly coupled to today’s workload that tomorrow’s workload breaks it. A product designed to satisfy every possible request becomes so bloated that it satisfies no one. In each case the pursuit of the perfect score is precisely what produces the failure. Push past good, and you do not arrive at perfect. You come out the other side, into worse.
Chapter 15 offers the way out: “Whoever holds to the Tao does not seek fullness. Because they are never full, they can wear out and be renewed.” This is the strongest reason to stop chasing perfection. Not because it is hard, and not only because it does not exist, but because a thing that never tries to be full keeps the one property that matters most over time: the ability to be renewed. A bow not drawn to the full can still be drawn further. The imperfect product can still be improved. The finished, perfect one has only one direction left to go.
Duke Nukem Forever had fourteen years, a talented team, and all the money it needed to become perfect, and it still never got there, because there was nowhere to get to. Every time it caught up with the target, the target had moved, and by the time it finally shipped, the chase had worn it into something worse than a good game shipped years earlier would have been.
Letting go of perfectionism changed how I work, but the biggest change in me was in how I see my life. I used to hold my life to the same impossible standard I held my work to. A career was supposed to be a straight line, each step a little higher than the last, and every detour felt like a failure. A mistake was not something to learn from but a stain to hide. Without ever saying it out loud, I was treating my own life like Duke Nukem Forever: something that had to be flawless before it was allowed to count.
I do not live that way anymore. I allow myself to make mistakes, because a life without mistakes is a life without experiments, and a life without experiments never finds out what it could have been. My career does not have to follow a linear path; some of the most valuable things I know, I learned on what looked like detours at the time. My life does not have to be perfect. It never could have been. It only has to be lived, honestly and fully, and it turns out that is much easier to do once you stop grading it.
So here is what I try to do now. Start before you are ready. Pursue quality with everything you have, but measure it against the people you serve, not against a flawless image in your head. Remember that perfection is a mirage, that every real thing is a trade-off, and that pushing past good leads not to perfect but to worse. And when you finally ship something rough, flawed, and useful, do not apologize for it.
The highest perfection is like imperfection. That is true of a product, and it is true of a life. Ship the imperfect thing, live the imperfect life, and keep making both better.