The Entrepreneurial Bible's Self-Collapse: The More You Know, The Faster You Die

Author: Colossus

Translation: Deep Tide TechFlow

Deep Tide Guide: This article uses U.S. government data to reveal an uncomfortable truth: over the past 30 years, all the bestselling startup methodologies—Lean Startup, Customer Development, Business Model Canvas—have statistically shown no improvement in startup survival rates.

The problem isn’t necessarily that the methodologies are wrong; it’s that once everyone uses the same approach, it loses its advantage.

This argument applies equally to crypto and Web3 entrepreneurs, especially those reading various “Web3 Startup Guides.”

Full text below:

Any method for building a startup, once widely adopted, will lead founders to the same answers. If everyone follows the same bestselling startup techniques, ultimately they will build similar companies without differentiation, and most of these companies will fail. The truth is, whenever someone insists on teaching a way to build successful startups, you should do something different. Once you understand this paradox, it’s obvious, but it also points the way forward.

Before the rise of the new wave of “startup evangelists” twenty-five years ago, the advice they replaced was, frankly, worse than useless. It was a naive mix of Fortune 500 strategies and small business tactics—five-year plans alongside daily operations. But for high-growth startups, long-term planning is pointless—future is unpredictable—and focusing on daily operations exposes founders to faster competitors. The old advice was designed for a gradual improvement world, not for fundamental uncertainty.

The new generation of startup evangelists offers different advice: intuitive, seemingly well-argued, providing founders with a step-by-step process for building companies amid real uncertainty. Steve Blank’s “Four Steps to the Epiphany” (2005) introduced Customer Development, teaching founders to treat business ideas as falsifiable hypotheses: go out, interview potential customers, validate or disprove assumptions before writing code. Eric Ries’s “Lean Startup” (2011) built on this, proposing Build-Measure-Learn cycles: release a Minimum Viable Product, measure real user behavior, iterate quickly instead of wasting time perfecting a product no one wants. Osterwalder’s Business Model Canvas (2008) gives founders a tool to map out nine core components of a business model and pivot quickly when something doesn’t work. Design Thinking—promoted by IDEO and Stanford d.school—emphasizes empathy for end users and rapid prototyping to identify issues early. Saras Sarasvathy’s Effectuation theory suggests starting from the founder’s skills and networks rather than reverse-engineering a plan to achieve grand goals.

These evangelists aim to establish a science of startup success. By 2012, Steve Blank claimed the U.S. National Science Foundation was calling his customer development framework “the scientific method of entrepreneurship,” asserting “we now know how to make startups less likely to fail.” The Lean Startup website states it “provides a scientific approach to creating and managing startups,” and his book’s back cover quotes IDEO CEO Tim Brown, saying Ries “proposed a learnable, replicable scientific process.” Meanwhile, Osterwalder’s PhD thesis claims the Business Model Canvas is based on Design Science (the predecessor of Design Thinking).

Academic entrepreneurship research also studies startups, but their science is closer to anthropology: describing founders’ cultures and startup practices to understand them. The new evangelists have a more pragmatic vision—like Robert Boyle, the early modern scientist who said: “I do not dare to call myself a true naturalist unless my skills can grow better herbs and flowers in my garden.” In other words, science should seek fundamental truths but also be effective.

Its effectiveness, of course, determines whether it deserves to be called science. And on the topic of startup evangelism, one thing is clear: it has not worked.

What have we actually learned?

In science, we judge effectiveness through experiments. When Einstein’s relativity gained acceptance, physicists invested time and money designing experiments to test its predictions. We learn in elementary school that the scientific method is science itself.

But due to a human flaw, we tend to resist the idea that “truth is discovered this way.” Our minds seek evidence, but our hearts need a story. An ancient philosophical stance—discussed brilliantly by Steven Shapin and Simon Schaffer in “Leviathan and the Air-Pump” (1985)—argues that observation cannot give us truth; true knowledge can only be derived logically from what we already accept as true, starting from first principles. While standard in mathematics, in fields with noisy data or shaky axioms, this can lead to seemingly plausible but absurd conclusions.

Before the 16th century, doctors treated patients based on Galen’s writings from 2nd-century Greece. Galen believed disease was caused by an imbalance of four humors—blood, phlegm, yellow bile, black bile—and recommended bloodletting, emetics, and cupping to restore balance. These treatments persisted for over a millennium, not because they were effective, but because the authority of ancient scholars seemed to outweigh current observations. Around 1500, Swiss physician Paracelsus noticed that Galen’s therapies often didn’t improve patients, and some—like mercury for syphilis—were nonsensical within the humoral framework but actually worked. Paracelsus began listening to evidence rather than obeying long-dead authority: “The patient is your textbook, the bedside your study.” In 1527, he publicly burned Galen’s works. His vision took centuries to be accepted—nearly 300 years later, George Washington died after a radical bloodletting—because people preferred the neat, simple stories like Galen’s over the messy reality.

Paracelsus started from what worked and traced back to the cause. First-principles thinkers, on the other hand, assume a cause and then claim it’s effective regardless of the outcome. Are modern startup thinkers more like Paracelsus—evidence-driven? Or more like Galen—relying on elegant, self-consistent stories? Let’s look at the evidence in the name of science.

Here is official U.S. government data on startup survival rates. Each line shows the probability that a company founded in a given year survives to a certain point. The first line tracks one-year survival, the second two-year, and so on. The chart shows that from 1995 to today, the proportion of companies surviving one year has remained roughly unchanged. The same applies to two, five, and ten-year survival rates.

The new evangelists have been around long enough and are well-known enough—books have sold millions, and nearly every university entrepreneurship course covers these methods. If they worked, the data would reflect it. Yet, over the past three decades, there has been zero systemic improvement in making startups more likely to survive.

The government data includes all U.S. startups—restaurants, dry cleaners, law firms, landscaping companies—not just high-growth tech startups supported by venture capital. Evangelists do not claim their methods are only for Silicon Valley-type companies, but these techniques are most often tailored for environments where founders are willing to accept extreme uncertainty for the chance of high returns. Therefore, we adopt a more targeted metric: the proportion of venture-backed startups that, after completing their initial funding round, go on to raise subsequent rounds. Given how venture capital works, we can reasonably assume that most companies that fail to raise follow-up rounds do not survive.

The solid line shows raw data; the dashed line adjusts for recent seed-stage companies that might still raise Series A.

The sharp decline in the proportion of seed-funded companies that go on to raise further rounds does not support the idea that venture-backed startups have become more successful over the past 15 years. If anything, they seem to fail more often. Of course, venture deployment is influenced by factors beyond startup quality: COVID-19 shocks, the end of the zero-interest-rate era, the capital-intensive nature of AI, and more.

Some might argue that the growth in total venture funding has flooded the market with less qualified entrepreneurs, offsetting any gains in success rates. But in the chart below, success rates decline both during periods of funding growth and contraction. If an oversupply of unskilled founders was dragging down the average, success rates should rebound after 2021 when funding slowed. But they did not.

But isn’t an increase in founders itself a kind of success? Try telling that to entrepreneurs who followed evangelist advice and still failed. These are real people risking their time, savings, and reputation; they deserve to know what they’re facing. Top VCs may have made more money—more unicorns than before—but that’s partly because exits take longer, and partly because the power-law distribution of exits means more startups lead to huge successes as more are launched. For founders, that’s cold comfort. The system may produce more big wins, but it doesn’t improve individual odds.

We must face a stark fact: the new evangelists have failed to make startups more likely to succeed. Data shows that, at best, they have had no effect. We have spent countless hours and billions of dollars on a fundamentally flawed framework.

Moving Toward a Science of Entrepreneurship

Evangelists claim they are giving us a science of entrepreneurship, but by their own standards, we have made no progress: we still don’t know how to make startups more successful. Boyle would say that if your garden hasn’t grown better herbs and flowers, there’s no science. It’s disappointing and confusing. Given the time invested, widespread adoption, and the apparent intellectual caliber behind these ideas, it’s hard to imagine they’re useless. Yet, the data shows we’ve learned nothing.

If we want to build a real science of entrepreneurship, we need to understand why. There are three possibilities. First, perhaps these theories are fundamentally wrong. Second, perhaps they are so obvious that systematizing them is pointless. Third, perhaps once everyone uses the same theories, they no longer confer any advantage. After all, strategy is about doing something different from your competitors.

Maybe the theories are wrong

If these theories are fundamentally wrong, their spread should have decreased overall startup success rates. Our data shows that this isn’t the case; in fact, failure rates for venture-backed companies seem to have increased for other reasons. Setting aside the data, these theories don’t look obviously wrong. Talking to customers, running experiments, iterating—these all seem obviously beneficial. But Galen’s theories in 1600 also didn’t seem wrong to doctors then. Unless we test these frameworks as we do other scientific hypotheses, we can’t be sure.

That’s the standard Karl Popper set in “The Logic of Scientific Discovery”: a theory is scientific only if it can, in principle, be proven false. You have a theory; you test it. If experiments don’t support it, you discard it and try something else. An unfalsifiable theory isn’t a theory at all, but a belief.

Few attempt to apply this standard to startup research. There are some randomized controlled trials, but they often lack statistical power and define “effectiveness” as something other than real startup success. Given that venture capital invests billions annually, and founders spend years trying their ideas, it’s strange that no one seriously tests whether the techniques taught to startups are truly effective.

But evangelists have little incentive to test their theories: they make money and influence through book sales. Accelerators profit by funneling many entrepreneurs into the power-law distribution, celebrating a few outsized successes. Academic researchers face their own distortions: proving their theories wrong risks losing funding, with no immediate reward. The entire industry resembles what physicist Richard Feynman called “cargo cult science”: a facade mimicking scientific form without substance—deriving rules from anecdotes rather than establishing causal relationships. Just because some successful startups have done customer interviews doesn’t mean your startup will succeed if you do the same.

But unless we admit that current answers are insufficient, we won’t be motivated to seek better ones. We need experiments to discover what works and what doesn’t. This will be costly, because startups are poor test subjects. It’s hard to force a startup to do or not do something (can you stop founders from iterating, talking to customers, or asking users’ preferences?), and maintaining rigorous records is usually low priority when fighting for survival. Each theory has many nuances to test, and in practice, these experiments may be impossible to conduct properly. But if that’s the case, then we must admit that for any unfalsifiable theory, it’s not science but pseudoscience.

Maybe the theories are too obvious

To some extent, founders don’t need formal training in these techniques. Even before Blank’s “Customer Development,” founders were talking to customers. Before Ries named it “Lean Startup,” they were building MVPs and iterating. Before design thinking was popular, they were designing products for users. Business practices often emerge naturally, driven by market forces; millions of entrepreneurs independently reinvented these methods to solve daily problems. Perhaps these theories are simply obvious, and evangelists are just repackaging old wine in new bottles.

That’s not necessarily bad. Having effective theories—even if obvious—is a first step toward better theories. Unlike Popper, scientists don’t abandon promising theories at the first falsification; they try to improve or extend them. Thomas Kuhn, in “The Structure of Scientific Revolutions,” powerfully illustrated this: after Newton’s gravity theory, predictions about the moon’s motion were wrong for over 60 years until mathematician Alexis Clairaut recognized it as a three-body problem and corrected it. Popper’s standard would have led us to discard Newton. But that didn’t happen because the theory was well-supported elsewhere. Kuhn argued that scientists are often stubborn within a paradigm—what he called a “paradigm”—which provides a framework for building and improving theories. They don’t abandon a paradigm easily unless forced, because it offers a path forward.

There is no single paradigm in startup research. Or rather, there are too many, and none sufficiently compelling to unify the field. This means that those trying to treat entrepreneurship as a science lack a shared guide on which questions matter, what observations mean, or how to improve imperfect theories. Without a paradigm, researchers are just spinning their wheels, each speaking a different language. To make entrepreneurship a science, we need a dominant paradigm—a sufficiently convincing, collectively guiding framework. This is a harder problem than simply deciding which theories to test, because a paradigm must answer pressing open questions. We can’t create one from nothing, but we should encourage more efforts.

Maybe the theories are self-undermining

Economics teaches us that if you do the same thing as everyone else—selling to the same customers, using the same processes, sourcing from the same suppliers—competition will drive profits to zero. This is the cornerstone of business strategy: from George Soros’s “reflexivity”—market participants’ beliefs change the market itself, eroding their advantages—to Peter Thiel’s Schumpeterian “competition is for losers.” Michael Porter’s “Competitive Strategy” codifies this as the need to find uncontested market space. Kim and Mauborgne’s “Blue Ocean Strategy” takes it further, advocating for creating entirely new markets rather than fighting over existing ones.

But if everyone is using the same methods to build their companies, they tend to compete directly. If every founder interviews customers, they’ll converge on similar answers. If every team releases MVPs and iterates, they’ll end up with similar products. Success in a competitive market must be relative, meaning effective practices must differ from what everyone else is doing.

Reductio ad absurdum makes this clear: if there were a foolproof startup success process, people would be mass-producing successful startups all day long. It would be a perpetual money machine. But in a competitive environment, the flood of new companies leads to most failing. The false premise is that such a process could exist.

An exact analogy comes from evolutionary theory. In 1973, biologist Leigh Van Valen proposed the “Red Queen Hypothesis”: in any ecosystem, when one species evolves advantages at the expense of another, the disadvantaged species evolve to counteract those advantages. The name comes from Lewis Carroll’s “Through the Looking-Glass,” where the Red Queen tells Alice: “It takes all the running you can do, to keep in the same place.” Species must continually innovate with diverse strategies just to survive against competitors’ innovations.

Similarly, when new startup methods are rapidly adopted by everyone, no one gains a relative advantage, and success rates stay flat. To win, startups must develop novel, differentiated strategies and establish barriers to imitation before competitors catch up. This often means that winning strategies are either developed internally (not found in any open publication) or are so unconventional that no one would think to copy them.

This makes building a scientific understanding seem very difficult…