Rereading Drucker''s Innovation and Entrepreneurship: "Creative Imitation

I · Putting Down the Baggage of “Originality”

Innovation and Entrepreneurship came out in 1985 — more than forty years ago. I first read it in my years as a management consultant. Rereading it now, as a practitioner in the AI era, feels entirely different: we are living through the moment when the line between “imitation” and “creation” is fought over more fiercely than ever. What I am rereading is Chapter 17, titled “Hit Them Where They Ain’t” — the words of a Confederate general of the American Civil War, famous for winning. Drucker borrows the soldier’s maxim to cover two utterly different entrepreneurial strategies: creative imitation and entrepreneurial judo. This essay is about the first; judo can wait for the next one.

Drucker concedes upfront that the term contradicts itself: whatever is creative must be original, and an imitation is by definition not. Yet his definition cuts clean: the entrepreneur who uses this strategy understands what the innovation means better than the people who made it. His heaviest example is IBM, and he uses it twice. First, the mainframe: before the Second World War ended, IBM had already built a true computer — memory, programmable, the works — and then walked away from its own design to follow the ENIAC line out of the University of Pennsylvania. ENIAC’s designers saw a tool for scientific calculation; IBM saw payroll — data processing. It re-engineered the machine for mass production and easy maintenance, shipped it in 1953, and watched it become the standard for commercial mainframes. Second, the personal computer: the PC was Apple’s idea, and everyone inside IBM had judged the small, free-standing computer a mistake. But it succeeded — so IBM moved at once, and within two years it had displaced Apple as the industry standard. Drucker’s timing rule is exact: wait until someone else has created the new thing, but not quite finished the job — then move. That window has a cliff on either side. Move too early and the demand is still unproven — remember, all of IBM had called the small computer a mistake, which tells you nobody could know in advance that the demand existed. Move too late and the originator has already set the standard. Put another way: the most expensive experiment in all of innovation — does anyone actually want this? — is run by the originator, free of charge, for the imitator’s benefit. Apple burned its own money to prove people would buy personal computers; only then did IBM walk in. And here the “soft spot” gets its first precise meaning: the originator has proven the demand but cannot serve it well. It is a soft spot in understanding, not in resources.

My first reaction to this section was suspicion. Look at the protagonists: IBM, Procter & Gamble, Seiko — giants of their industries, every one, and the declared objective is always “the industry standard,” “market leadership.” IBM rolled over Apple in two years not merely because it “understood the PC’s meaning better,” but because it already owned the trust of corporate customers, the distribution, the manufacturing scale. “Hit them where they ain’t” makes you think of guerrillas, of the weak toppling the strong — yet every example Drucker gives has a giant in the lead role. The title promises David; every story is Goliath winning; and Drucker never squares the two. So let it be said plainly: creative imitation is a strategy for the strong, not the weak. Hitting the soft spot presupposes a hard enough fist. And Drucker, writing his field manual entirely from the imitator’s driver’s seat, says not one word about what this strategy does to everyone else — giants suppressing early innovators by copying them, squeezing small players out of existence. I have watched this happen over and over in mainland China: in industry after industry there are de facto monopolists who copy on sight, whose scale means your idea can simply be taken — they have the channels, the capital, the traffic, and genuine innovation is left very little room to live. But this is no Chinese peculiarity; commercial competition works this way everywhere, and the closer a market sits to monopoly, the more often the scene plays out. One and the same act is called creative imitation in the strategy classroom, predatory copying in the ecosystem, and abuse of market dominance once it crosses the legal line. The prettier the name, the more it pays to ask who is using it — and from which seat.

But on a second, closer reading, I revised my opening verdict — “this is just a fig leaf for big-company copying” — and the reason lies in the way the two IBM stories are set side by side. The PC is a standard fast-follower tale. The ENIAC story is far stranger: IBM itself was the originator. It had built the real computer — and then set aside its own creation to follow a competitor’s design. Which reveals that the true threshold of creative imitation is not whether you can imitate, but whether an organization can let go of its pride of ownership in its own inventions. Most companies die of not-invented-here syndrome: they cannot discard their own design however inferior it proves, cannot repudiate their own consensus however wrong. IBM managed it twice — once discarding its own machine, once overturning the unanimous judgment of its own people. Now look at the ones that couldn’t: Kodak invented the digital camera. Xerox invented the graphical interface. Nokia built the earliest touchscreen prototypes. None of them lacked money, talent, or information, and every one of them went down on a road it had personally paved. Resources are necessary, not sufficient; what is scarce is the organizational discipline to put the baggage down.

Once that clicks, what Drucker is really driving at comes into focus: quietly, he is rewriting the definition of innovation itself. Inventing a new thing is not innovation; finding the customers and the use the thing truly serves — that is. Creation happens not in the product but in the defining of its meaning. To ENIAC’s designers the machine meant scientific computing; to IBM it meant data processing. To Apple the PC was a hobbyist’s toy; IBM redefined it as standard equipment for the corporate office. Whoever completes the definition of meaning is the innovator in the economic sense. The history books do not list IBM as the computer’s inventor; the market lists it as the setter of standards. The glory of invention and the profit of innovation have always been two different things. This also hands me a touchstone that makes my earlier criticism sharper: look for a redefinition of meaning. When today’s giants copy today’s startups, there is mostly no redefinition of anything — only superior execution grinding down a smaller firm. That does not deserve the name creative imitation. It is predatory copying wearing its clothes.

Hold that touchstone up to Apple — the company that got crushed — and a twist appears. In his Stanford commencement address, Jobs accused Microsoft with real bitterness: the feel for typefaces he had absorbed auditing a calligraphy class after dropping out, which he poured into the Mac’s fonts and typography, was carted off wholesale by Windows. The grievance was sincere — but it tells only the lower half of the chain. The Mac’s graphical interface traced back to Xerox PARC; Xerox had invented the GUI and grasped nothing of its commercial meaning, and Jobs, after one look at a demo, took it and redefined what it was for. He also said, famously, “good artists copy, great artists steal.” Apple is itself a textbook creative imitator. The true picture is never “originator versus thief.” It is a chain of imitation — Xerox → Apple → Microsoft — with every link calling itself the innovator and the next link down a burglar. There are no pure victims on the chain, only originator after originator who failed to hold the right to define what their creation meant. And every transfer of value happens through the same gap: the upstream party not understanding what it has made.

Reread today, the section’s sharpest contrast is right in front of us: in the AI era, every company — giant or startup — stands at both ends of the imitation chain at once. You are imitating someone, and someone is always about to imitate you. Drucker’s law has not moved in forty years: what decides who laughs last is not who invented first, but who saw earlier what the innovation truly means — and who dares, for the sake of that meaning, to put down the baggage of their own “originality.”

II · The Long Game

The first part was about the inability to let go of one’s own inventions. Now Drucker pushes the question a level deeper and lays out the complete mechanism of why this strategy runs so little risk. He adds Procter & Gamble — the same playbook across detergents, soap, toiletries, processed foods — and then comes the watch story. Once the semiconductor arrived, everyone in the watch business knew that a quartz watch would be more accurate, more reliable, and cheaper than anything wound by hand. The Swiss promptly built quartz digital watches — and then, having sunk so much into traditional watchmaking, decided to stretch out the transition, pricing quartz as an expensive luxury and easing it in slowly. Seiko, until then a traditional watchmaker for the Japanese home market, saw the opening and moved at once, making the quartz watch the industry standard. By the time the Swiss woke up it was over: Seiko had the world’s best-selling watches, and the Swiss had been all but pushed out of the market.

On risk, Drucker is blunt. Creative imitation aims exactly where the bet-the-company strategy aims — market leadership, even control of an industry — at a fraction of the danger. By the time the imitator moves, the market exists, the new thing has been accepted, demand is running ahead of what the originator can supply, and the market’s segmentation has formed or is forming. Market research can now find out what customers buy, how they buy, what they consider value. Most of the uncertainty the originator once faced is gone, or at least has become analyzable. In his words: nobody has to explain to customers anymore what a personal computer or a quartz watch is and what it is for. He honestly logs the counterexamples — Hoffmann-La Roche’s vitamins, DuPont’s nylon, Wang’s word processor: get it right the first time and the imitator never gets a chance. But judging by the sheer number of successful imitators, the protection a pioneer buys by getting there first is thinner than advertised. Then comes the example he calls the fullest expression of the strategy: Tylenol. Acetaminophen kills pain without reducing inflammation or thinning the blood — none of aspirin’s stomach-bleeding side effects. The first over-the-counter brand marketed it precisely as “pain relief without aspirin’s side effects,” succeeded hugely, and thereby boxed itself into a small market. Johnson & Johnson saw what the originator had not: the real prize was a painkiller to replace aspirin. It positioned Tylenol from day one as the safe, universal analgesic — and took the whole market within a couple of years.

How the Swiss lost deserves to be copied into the notebook of everyone who has already made money. Note the detail in the text: the Swiss were not blind, and they were not empty-handed — they “promptly produced quartz digital watches.” What did them in was an active decision, perfectly rational-looking at the time: too much invested in traditional watchmaking, therefore price quartz high, as a luxury, and take the transition slowly. In the language of the previous section: the Swiss defined the quartz watch’s meaning as “an expensive variant of the traditional luxury watch,” while Seiko defined it as “a more accurate, more reliable, cheaper watch for everyone.” Another contest over the definition of meaning — except the Swiss definition was not written by their judgment. It was written by their balance sheet. This drives “putting down the baggage” a level deeper. The psychological attachment to one’s own invention is the curable kind of baggage. The hard kind is the structure of interest: your existing assets will price the new thing for you. For anyone who has already extracted position, profit, and vested interest from a field, dismantling that structure is a different order of difficulty from admitting your invention belongs in the bin. In passing: this is precisely the pathology at the heart of Christensen’s The Innovator’s Dilemma of 1997. Drucker laid out the complete case file in 1985 — twelve years early.

Further down, the passage on risk is the chapter’s real theoretical contribution: Drucker dismantles “first-mover advantage,” a myth still being sold today. Every item on his list is a cost borne by the first mover — the uncertainty, the expense of educating a market, supply that cannot keep up — while the imitator enters when “market research can find out what customers are buying.” Beneath that sentence lies a hard little truth: a market cannot be researched before it has been created. The originator must decide in the dark, without data; the imitator decides in daylight, with it. So the two are not even carrying the same kind of load. In the economist Frank Knight’s classic distinction: the originator carries uncertainty, which no probability can be attached to; the imitator carries only risk, which can be calculated. And that phrase from part one — “not quite finished the job” — only now shows its exact meaning: that moment is the threshold where uncertainty has just collapsed into calculable risk. The “low risk” of creative imitation is not timidity. It is entering the market precisely at that threshold. Tylenol pushes the logic to its limit. IBM at least re-engineered ENIAC for mass production; Seiko at least had to build the watches. Tylenol’s molecule was identical to its rival’s — acetaminophen is acetaminophen. Zero difference in the product; one difference in positioning — “a substitute for people who can’t take aspirin” versus “the safe painkiller for everyone” — decided the ownership of an entire market. A clean controlled experiment, proving that the “creative” in creative imitation can live one hundred percent in the meaning and zero percent in the product, and the strategy still works.

But finish these pages and a paradox sits on the table. Both strategies aim at the same prize, and the imitator’s risk is plainly lower — so by any rational calculus, everyone should queue up to imitate. Who volunteers to originate? If the whole world understood this chapter, nobody would take innovation’s first step. The market has visibly not collapsed. Where is the answer? Not inside Drucker’s framework, I think — he is doing the arithmetic of business, and a human life runs on more than business arithmetic. The economist’s answer would be: someone always overrates their own originality and underrates the uncertainty, so the supply of originals never dries up. That answer smuggles in a premise — that originators and imitators are chasing the same stakes, money and market share. That is not the reality I see. Humanity keeps pursuing the new, and you do not need “somebody miscounted” to explain it: the thing that actually makes people happy is creation itself. Here is the most concrete line I know between carbon-based and silicon-based life — feeling and meaning are themselves our reward. A company that can truly gather people never does it on “our risk profile is the lowest”; it does it on originality, on creativity, on a vision people want to chase together. At the level of one person it is even plainer: repetitive, mechanical work will never hold the joy of creative work. That is the engine of human originality, and of progress. So the paradox dissolves not because someone miscalculates, but because originators and imitators are not after the same thing at all: the imitator can take the near-term money, and cannot take the joy, the meaning, the vision — which are precisely the rewards that matter to people whose eyes are set further out. Originality is, at bottom, a form of the long game. Drucker himself would take this side — his whole book stands against reducing the entrepreneur’s motive to profit maximization.

Following this thread, there is a corollary — its warm face first. Precisely because the rewards creators care about include things money cannot denominate, they will happily give the money away — cheap, or free. The open-source movement is that impulse, institutionalized. Writing this, I can’t help saying it plainly: open source is one of humanity’s lights. From Unix and Linux, through public papers and open datasets, to today’s open-weight models, generation after generation of creators have laid their best work into the commons and asked nothing for it — because another kind of reward, the joy of making, the respect of peers, the quiet certainty of having moved the world an inch, weighed more. The productivity revolution now underway — call it generative AI, call it the road to AGI — stands, layer upon layer, on exactly this commons: open frameworks, open methods, open models; none of it grew out of thin air. Without the open-source tradition, there is no “today” at all. To advocate for open source is not sentimentality; it is refilling the spring that progress drinks from. The corollary has its colder face too: a commons is open to all comers, imitators included — the joy of creation subsidizes the imitators of the entire world. As for the economist’s story, I need not discard it; it slots in underneath. The joy of creation explains why originals never stop coming; the overestimation of one’s own originality explains why they come in perpetual oversupply — enough surplus to feed a whole population of imitators. Remove the second and originals would still exist, only far scarcer.

Carry “the long game” into today’s AI industry and things get much clearer than any purely technical talk. Liang Wenfeng has been the sample I find most interesting these two years, and the reason is exactly those words: the structure of interest. He comes out of High-Flyer, a quantitative fund — managing other people’s money in the financial industry, as clear and self-consistent a structure of interest as exists anywhere. Yet inside that structure he grew a temperament unlike his industry’s and unlike the AI industry’s either. Not wanting to argue from impressions, I went through the transcript of his July 2026 investor meeting — DeepSeek’s first outside equity round, roughly RMB 51 billion. Across nearly four hours, what he laid out was: no profit maximization; the strongest models stay open-source; no super-app; no genius mythology. API prices are set to recover hardware costs in ten months, with room to raise them that he explicitly declines to use. He called the company’s founding impulse “goodwill toward the world” — “something beyond money” — and said the first few dozen people who joined did not put IPOs and personal wealth first. Note the setting. An investor meeting is the one room where the structure of interest is supposed to do the talking, and that is precisely where he refused to let it price his future. It is the exact inverse of what the Swiss watchmakers did — and it is the long game at its plainest: knowing the money is there to be taken, and trading it for something further away. Of course these words come at a discount: a man describing himself, and “not chasing profit” can itself be positioning aimed at investors. But keeping the strongest models open, holding API prices down, declining the super-app: those are things that have already happened. The deeds match the words. So I will grant it: this sample really is different.

The same transcript contains the most Druckerian sentence of all, in which he owns the structure of creative imitation outright: the old mode was “one to two years behind, doing comparable work with one-twentieth the compute of the American companies.” That is the timing doctrine confessing itself — late entry, low cost, wait for uncertainty to collapse into risk, then move. The second half of his sentence is the part worth chewing on: the next phase is to compress the gap to six months, three months, and to reach a higher level first in a few chosen directions. That is an announcement of changing seats — from imitator to originator. And the seat change has a price. The costliest invoice — does anyone want this? — used to be paid by the originator, free, on the imitator’s behalf. Compress the gap to three months and that free lunch is gone: three months is nowhere near enough for uncertainty to collapse into risk. From here on, he pays the originator’s invoice himself. So yes, DeepSeek is a household-name exemplar of creative imitation — but the label needs a tense. It is an exemplar in the past tense, and its principal is walking out of the strategy on purpose. R1 versus o1 was the closing masterpiece of that phase: OpenAI hid the reasoning chain as a trade secret; DeepSeek published its chain of thought, opened the weights, cut the price by an order of magnitude — and thereby redefined the meaning of “reasoning model” from a guarded commodity into a public methodology. The product followed; the meaning was original. Creative imitation, by the book.

The open-source move, though, has already walked out of Drucker’s 1985 frame. Drucker’s imitator attacks other people’s soft spots; DeepSeek, by open-sourcing its strongest models, deliberately exposes its own — it volunteers to be the whole industry’s most imitated company. This is not naivety. The business logic from that meeting is plain: the AI market is too big for any one company to hold, and a rival willing to live on one percent margin can always undercut you. Translated into this essay’s terms: he is trading away monopoly profit at the product layer to hold the right to define meaning over the long term. Once an open standard becomes the industry default, the meaning you defined becomes everyone else’s starting point. The best defense of the right to define meaning turns out to be inviting the entire world to imitate you. Drucker never wrote this move — in 1985, no industry’s cost of distribution could fall this close to zero.

Now look at the other end of the chain, and imitation stops being one-directional at all. OpenAI originated the ChatGPT paradigm; Anthropic was the late-arriving pursuer. But once Claude Code had defined the shape of agentic coding — AI working continuously inside a codebase on its own — OpenAI turned around and imitated the shape back with Codex, nearly feature for feature. The Xerox → Apple → Microsoft chain in part one looked one-way. In the AI industry the chain is already a two-way loop, and its period has compressed from Drucker-era years to months. Which explains the blizzard of distillation accusations: the chain now turns so fast that no link has time to establish its own “I am the originator” story before the next link has already redefined it. At the limit, when the imitation cycle runs shorter than the product cycle, the defensible lifetime of meaning-definition approaches zero, and the whole industry becomes high-frequency trading in meaning — everyone an originator, everyone an imitator, accuser and accused rotating monthly.

So this part closes on the long game. The short-term arithmetic the imitator will always do more sharply than the originator — Drucker’s chapter teaches it plainly, and it still works forty years on. But some things the imitator cannot take at all: the joy of making, the vision that gathers people, that click of certainty when you are the first to say clearly what a new thing means. And AI itself is the most extreme footnote to this truth: everything today’s large models can do was learned from what humans created — books, code, paintings, conversation. Imitation is AI’s origin story; it is the largest act of creative imitation in human history, and “the joy of creation subsidizes the imitators of the world” has come literally true in the AI era. AI has pushed the cost of imitation toward zero — I use these tools every day myself. And precisely in such an era, the untakeable things become the last scarce goods. The glory of invention goes to the history books; the profit of innovation goes to the market. But what actually decides whether a person, or a company, will take the first step in the dark is neither — it is the longer account they keep inside.

III · The Winner’s Blind Spot, and Who Gets to Define “Distillation”

We have covered the originals people cannot let go of, then the structures of interest they cannot let go of. Here Drucker names this strategy’s actual prey — and demolishes an easy misreading on the way. He could not be plainer: creative imitation does not exploit the pioneer’s failure. On the contrary, the pioneer must succeed. Apple was a spectacular success; so, before Tylenol, was the first company to brand acetaminophen. “But the original innovators did not understand what their success meant.” Apple cared about the product, not the customer: when users needed programs and software, it served up more hardware. Then comes the most concrete verdict in the chapter: technically, IBM’s personal computer was not significantly different from Apple’s — but IBM had the programs and software customers wanted ready from day one, and it broke its own decades-old tradition of direct sales, selling through dealers, through mass retailers like Sears (then America’s largest), through its own stores, making the machine easier to buy and easier to use. “These policies — not hardware features — were IBM’s true innovation.” In one sentence: creative imitation starts from the market, not the product; from the customer, not the maker.

He also states the precondition and the risks. The precondition: a fast-growing market. The creative imitator does not steal the originator’s customers; it serves the market the pioneer created but has not served properly — satisfying demand that already exists rather than conjuring new demand. The risks are real: hedging can scatter your forces; you can misread the trend and imitate something with no future. And then he does something few authors will — he takes apart his own best example. IBM successfully imitated every major advance in office automation, led in every category, and yet, precisely because every product was an imitation, the line was sprawling and mutually incompatible, and assembling an integrated office system from it was next to impossible. Being “too clever,” he says, is a risk built into the strategy. Finally he notes where the strategy works best — high tech, for a humble reason: high-tech innovators are the least market-minded of all, fixated on technology and product, and therefore the likeliest to misread their own success and fumble the demand they themselves created. His list of what the imitator needs: alertness, flexibility, willingness to listen to the market — plus hard work and massive effort.

“The pioneer must succeed” rewrites the character of the whole chapter. Up to here you might read the strategy as kicking people when they’re down; Drucker rules that out. Creative imitation does not pick over the fallen. It feeds on the blind spots of winners. And notice this cuts deeper than part one. There the originator failed to understand what their creation meant; here they fail to understand why they succeeded — and that failure is almost impossible to detect from the inside, because you are winning, and nothing in the signal stream tells you your story about yourself is wrong. Success always writes itself an explanation — we win because of this mechanic, this hardware, this integrity — and that explanation is the victor’s after-the-fact narrative, not necessarily the true cause. Success is the most effective anesthetic of the mind, and what the creative imitator does, at bottom, is pocket the spread on a winner’s mistaken story of himself. This is the soft spot’s third layer. First layer: not understanding the meaning. Second: strapped down by the structure of interest. Third: blindfolded by your own success story.

The IBM PC passage also settles a doubt I opened with. On first reading the definition, I leaned toward calling IBM’s copy of Apple pure imitation. Drucker’s verdict here is more precise than both my suspicion and my rebuttal: at the product layer there was indeed no significant difference — the “pure imitation” instinct was right; but IBM’s innovation lay outside the product, in the software ecosystem customers wanted and in the sales channels that broke IBM’s own direct-sales religion. That advances the whole framework a step. Part one moved “creation” from the product to the meaning; this moves it again — meaning must be cashed out into actual delivery: who can buy it, how they buy it, whether it works for them once bought. Tylenol might tempt you to think defining meaning is a matter of finding the right slogan; this corrects that. Positioning that never lands in channels and ecosystems is nothing at all. Note, too, that this was IBM’s third act of self-negation — first the discarded in-house design, then the overturned consensus, now the direct-sales system itself, root of its mainframe hegemony and core of its identity. Three times, the same motion: reworking the very thing it was proudest of. Set beside the Swiss watchmakers, it makes a controlled pair: both saw the new thing; the Swiss could not bring themselves to touch traditional watchmaking, IBM could. The threshold of this strategy was never “imitating others.” It is negating yourself.

The fast-growing-market precondition straightened out a kink in my own thinking. I had been condemning giants for copy-crushing small players while conceding that Drucker’s playbook is genuinely brilliant — two attitudes that looked inconsistent, until I saw they are one principle under two market conditions. In a fast-growing market, demand outruns anyone’s capacity to serve it; the imitator needn’t steal a single customer — catching the overflow is a good living, and imitator and originator are not locked in zero-sum. Once growth tops out, the identical behavior degenerates into a pure fight over share. Today’s AI industry is the former — compute shortages, rate-limited APIs, enterprises in line; Drucker’s “demand tends to run far ahead of the original innovator’s capacity to supply” is literally true here, which is also why the spread of open-source models has not killed the closed vendors' revenue. The giant-copies-startup cases I have watched in mature industries are mostly the latter. One variable — the market’s growth rate — decides whether creative imitation is symbiosis or predation.

The office-automation self-demolition is the passage in this chapter most worth rereading. The same IBM, in the same chapter, is best case and cautionary tale. The mechanism is simple once spoken: imitation proceeds item by item, and each item’s meaning was defined by somebody else, so the sum has no unified meaning of its own. Imitation can copy products; it cannot copy the picture of the whole. You can win every local battle and still fail to assemble the whole thing the customer actually wants. This is a key amendment to “the right to define meaning”: meaning has layers, and winning it at the single-product layer does not hand you the system layer. And there is an ending the Drucker of 1985 could not yet see: IBM won the PC war by creative imitation — and then, inside the open architecture it had built, surrendered the definition of meaning to Microsoft and Intel. The suppliers it elevated became the true standard-holders of the PC age. So the imitation chain has a vertical dimension too: it passes not only sideways between rivals but up and down the supply chain. The one who wins the market and the one who holds the meaning need not be the same party.

The most legendary sequel to this thread was written by Apple — the crushed party of part one. After Jobs returned, Apple fought a textbook comeback in mobile. The iPhone, in 2007, redefined the phone’s meaning from a communication device into a personal computing platform in your pocket; the App Store, in 2008, cashed that meaning out into delivery — the world’s developers writing its software while Apple kept the gate. This time Apple held both the meaning and the delivery: the Android camp imitated every feature of the product and never dislodged Apple’s hold on what “the premium smartphone” means. Nearly twenty years on, the company IBM once rolled over in two years is the most profitable company in the world. And the future? No one holds a meaning forever — that is this chapter’s lesson. How long you hold it depends on holding meaning and delivery together — that is Apple’s.

One more detail, as a footnote to Liang Wenfeng. Drucker’s capability list for the imitator — alertness, flexibility, willingness to listen to the market — is, item for item, a list of abilities to receive external signals. Not one item is the ability to hold an internal conviction. The originator’s profile is the exact opposite: what the originator needs is the strength to resist external signals — to persist when no one believes and there is no data — while the imitator needs sensitivity in absorbing them. These are two fundamentally different organizational temperaments, and it is rare for one culture to be good at both. So when I said Liang is moving from the imitator’s seat to the originator’s, add the heavier clause: what has to change is not just strategy and compute, but an entire organization switching from listening to the market to not listening to it. Drucker’s line that high-tech innovators are the least market-minded shows both faces here — it is the originator’s soft spot, and it is also the originator’s job requirement.

The theory is dense enough by now, and a living specimen happens to be sitting right in front of us: Palworld. Lately I have been building a player-tool site for the game — Palworld LudBase (palworld.ludbase.com). Its promise in one line: it remembers what you can’t and calculates what you’d rather not — the cheapest breeding route, whether your tech and production are falling behind, a full Paldeck, a complete interactive world map, and a progress checkup that reads your own save file right in the browser (parsed locally, never uploaded). Building the site means watching the game’s every move. On July 10, Palworld’s 1.0 release set off another wave of attention — and pushed the 2024 lawsuit back into the trending lists along with it. The background is simple enough. The game comes from a small Tokyo studio called Pocketpair, a project that started with three or four people. Early access launched on Steam on January 19, 2024: eight million copies in six days, a concurrent peak of 2.1 million players — the second-highest in Steam history — second that year only to Black Myth: Wukong, with total players later passing forty million. That September, Nintendo and The Pokémon Company took it to court in Tokyo.

Most people assume Nintendo sued over “copying.” It didn’t — it sued over patents: throwing a ball-like item to catch creatures, summoning a captured creature to fight, riding one to travel. More telling still, one of those patents was filed six months after Palworld blew up, registered within a month under accelerated examination, and then used to sue. The damages sought came to only about thirty thousand US dollars — the point was never money; it was an injunction, a warning shot at whoever comes next. What followed was rather deflating: Pocketpair shipped a few patches that re-implemented summoning and riding, and the game stayed just as much fun; the Japan Patent Office rejected Nintendo’s related applications on the grounds that these mechanics had long been in general use — at one point citing a fan-made Pokémon video from 2013 as evidence; Nintendo narrowed its claims to older versions sold only in Japan, 1.0 sits entirely out of reach, and the verdict will not come until this November.

Suing on patents rather than copyright is itself proof that Nintendo misread the cause of its own success. Suing on patents means the company went down its own asset inventory — mechanics, hardware, exclusive systems — hunting for which item had been taken. But what Palworld took was never on that list. What it took was the demand Nintendo had chosen not to serve: mature-rated tone, guns, open-world survival and base-building, playing on a PC with your friends. Drucker’s line in this section might have been written for it — serving the market the pioneer created but did not serve properly. Nintendo has won for so many generations that it naturally narrates its success as “our exclusive console, our exclusive lineup, our distinctive style, our distinctive integrity.” Its actual success was spreading the joy of catching creatures to the entire world — a demand far larger than the boundary it was willing to serve within. Filing a patent after the fact is a winner discovering a hole in the moat and firing a brick into it too late; and the Patent Office’s rejections amount to an official stamp: the asset Nintendo took to court is its least valuable one, while its genuinely priceless asset — generations of feeling for Pokémon — Palworld never touched and could not touch. It staked its least valuable asset on a war to defend its most valuable one, because it had misidentified which was which.

There is a lesson about clocks here too. From filing to verdict runs more than two years, and in that same stretch Palworld completed its entire evolution from early access to 1.0 — the suit is aimed at a version of the product that no longer exists, and the product has moved beyond the suit’s reach. Whatever a patent can protect is precisely what is easiest to engineer around; what cannot be protected — the whole experience, the feel of the systems working together — is the real asset. When legal time runs slower than product time, property protection fails in practice — not because the law is wrong, but because it is slow. In AI the effect is more extreme still: distillation accusations everywhere, and not one resolved within a model generation.

Structurally, Palworld is to Nintendo what the IBM PC was to Apple. Apple sold only through specialty computer dealers; Nintendo sells only through its own console. IBM went to Sears; Palworld went to Steam. Drucker said IBM’s real innovation was not hardware but making the thing easier to buy and use; Palworld’s real innovation is not the creature designs but putting “catching monsters” into a container anyone can open for thirty dollars on a PC, with friends. Nintendo’s exclusive hardware is both its moat and its ceiling — the same “your assets price the new thing for you,” in a different denomination of sunk cost: the Swiss were bound by sunk cost on a balance sheet, Nintendo is bound by sunk cost in a brand. It cannot ship a gun-toting, violent, mature-rated Pokémon; that would break its most valuable asset, the family-friendly name. Its soft spot is not lack of ability — its own success assets forbid it to serve that demand. Pocketpair, three or four people at the start, was under no such prohibition. Brand sunk cost is harder to put down than financial sunk cost, because it is identity, not a number. And in passing: this is the judgment my own work on Palworld LudBase verifies daily. What players lack was never “another Pokémon” — it is someone to remember the breeding charts for them, work out the production chains, make sense of their own save files. The demand a pioneer creates always exceeds the boundary any single supplier is willing to serve, and the overflow is the living soil of third-party tool sites like LudBase. Drucker’s theorem in this section holds all the way down to a tool site.

I will not whitewash Palworld into a pure innovator. The controversy over art and creature resemblance is real, and Pocketpair’s earlier games drew criticism as collages of whatever was popular. By the touchstone from part one: at the layer of art and characters it is heavy imitation — the uncomfortable half of the truth; at the layer of delivery and experience it is genuinely original. Both are true at once, the same double verdict we reached on IBM. But here is the difference: IBM was a strong player crushing a weak one; Palworld is a weak player flipping the strong player’s table. Same strategy, same moves — invert the balance of power and the optics invert with it. That is the thing I most wanted to say. Whether creative imitation is symbiosis or predation turns on two conditions: the market’s growth rate, and the balance of power. The strong using it on the weak in a topped-out market is suppression; the weak using it on the strong in a growing market is a breakout. My distaste when giants copy startups, and my instinctive sympathy for the Palworld side, are not a double standard. They are one principle under two sets of conditions.

One step further and you reach the hardest part of all: who has the power to name an act “creative imitation,” and whose act only ever gets named “distillation.” When Western companies judge the same behavior in the East, the word of choice used to be shameless copying — the fashionable version today is “distillation” — while the same act on their own résumé is narrated as creative imitation, fast following, ecosystem integration. The most ironic live example at this moment is Anthropic. As noted above, it was itself the late-arriving pursuer of the ChatGPT paradigm, and after its Claude Code defined the shape of agentic coding, OpenAI imitated the shape right back with Codex — the company sits at both the upstream and downstream ends of the imitation chain simultaneously. Yet this is the company that repeatedly accuses Chinese AI firms of advancing by imitation and distillation, and that cut off Chinese-majority-owned companies' access to its models by corporate policy. One company, occupying both ends of the chain, and holding the naming rights too. DeepSeek is the named end: by the standards of this very chapter it is a textbook — one is tempted to say Western-management-textbook — exemplar of creative imitation: late entry, low cost, turning a rival’s trade-secret reasoning process into a public methodology, redefining what the technology means. The name it received all the same was the one word: “distillation.” Naming rights follow power, not behavior. When Drucker coined “creative imitation” in 1985, he meant to hand entrepreneurs a serviceable strategic tool; forty years later the term doubles as a narrative weapon — applied to oneself it is strategy, applied to a rival it is swapped for the other word. That is not Drucker’s fault. But reading him today, you have to see this layer.

With the whole chapter read, one takeaway belongs to reading itself, best said with the old proverb: better no books than blind faith in them. Drucker’s framework is a field manual written from the imitator’s driver’s seat. Its insight into timing, into the customer’s view, into delivery, is still sharp forty years on and worth learning line by line. But it is silent on the balance of power, silent on naming rights, silent on how the optics of one identical act flip when strong and weak trade places. To see the thing clearly you have to change seats several times. From the imitator’s seat you see blind spots and opportunity. From the originator’s seat you see your own success being arbitraged away. From the ecosystem’s seat you see symbiosis while the market grows and predation once it peaks. And only from the narrator’s seat do you finally notice that “creative imitation” and “distillation” very often describe the same act — the difference is only in who is telling the story.

But that is a reader’s takeaway, not where I want to end. My closing position returns to the open-source spirit. Creation is the greatest emotional reward a human being can have; the willingness to share what you create is the spring that progress drinks from — this whole productivity revolution, generative AI, the road to AGI, rests layer by layer on things someone first created and then made public. The imitator’s arithmetic will always be sharper; Drucker settled that account forty years ago. But the world moves forward not on sharp arithmetic — it moves on the people who know they will be imitated and put their best work out anyway. So if this essay leaves one line behind, let it be this: create, and share; and don’t let any name — imitation, distillation, or whatever comes next — frighten you out of the first step.

Note 1: The Chinese translation I read attributes ENIAC to “Pennsylvania State University” — a mistranslation for the University of Pennsylvania. Drucker’s historical dates are themselves loose in places; in this chapter, precision of detail yields to the structure of the argument.

Note 2: Drucker lists Wang’s word processor among the cases that “got it right the first time and left imitators no chance.” True in 1985 — and a few years later Wang Laboratories was annihilated by the PC running word-processing software. Getting it right the first time seals off imitators only within a category; it cannot stop the category itself from being redefined. Defending a product market and defending the right to define meaning are two different wars. This is the particular dividend of rereading: time hands us data Drucker could not have had.

Note 3: Two passages in the Chinese translation of the Tylenol section need logical reconstruction: a sentence reading “recently the US classified it as a prescription drug” has the direction reversed — context indicates “until recently it had been a prescription drug”; and “aspirin’s market was very small” should, in context, refer to the small market for acetaminophen under its original substitute positioning, not to aspirin’s market. Exact wording should be checked against the English original.

Note 4: Drucker’s “when users needed programs and software, Apple brought out more hardware” is directionally right but simplified — VisiCalc, the first spreadsheet and the killer app of the Apple II era, is proof that software decided hardware’s fate. Apple’s real failure was not ignorance of software but of open software ecosystems and channels. A garbled sentence in the Chinese translation of this section should read, per context, “the high-tech innovators who misread the causes of their own success are not few.”

References

1. Peter F. Drucker, Innovation and Entrepreneurship: Practice and Principles, Harper & Row, 1985. Quotations follow the Chinese translation of Chapter 17.
2. Clayton M. Christensen, The Innovator’s Dilemma, Harvard Business School Press, 1997.
3. Frank H. Knight, Risk, Uncertainty and Profit, 1921.
4. Steve Jobs, Stanford Commencement Address, 2005.
5. What Liang Wenfeng told investors — full-transcript digest (Chinese) — TechWalker
6. DeepSeek won’t be “a more profitable company”: Liang Wenfeng’s four-hour investor meeting (Chinese) — HK01
7. DeepSeek founder Liang Wenfeng says the company does not pursue profit maximization (Chinese) — Sina Tech
8. Palworld 1.0 release date confirmed for July 10 — Steam official news
9. Palworld total players surpass 40 million (Chinese) — GCORES
10. Nintendo Filed a Lawsuit against Palworld for Pokémon Games Patents — Shiga International Patent Office
11. Nintendo Palworld Lawsuit Update: Full Timeline & Latest News — Palpedia
12. The Japanese Patent Office rejects a patent attached to the Palworld lawsuit — GamesRadar+
13. Palworld Just Passed 2 Million Concurrent Players On Steam — GameSpot
14. Palworld dev talks about game’s influences and original assets — AUTOMATON WEST