The Low-Hanging Fruit Theory of Innovation Is Only Half Right

A familiar theory of innovation says that progress gets harder because the low-hanging fruit is picked first. The first discoveries are close at hand. Later ones require larger machines, larger teams, deeper specialization, and more money. The economists Nicholas Bloom, Charles I. Jones, John Van Reenen, and Michael Webb made a version of this argument in “Are Ideas Getting Harder to Find?”, showing evidence that research effort has risen substantially while research productivity has declined in several fields. Particle physics needs vast accelerators. Drug discovery becomes slower and more expensive. Space exploration moves from the comparatively accessible parts of the problem into harsher and less forgiving environments.

There is truth in this. Some fields really do become more difficult as they mature. Once the obvious paths have been explored, the remaining work often demands more precision, more infrastructure, and more accumulated knowledge. The easy discoveries are not endlessly renewable inside a fixed field.

But that last phrase matters: inside a fixed field.

The low-hanging fruit metaphor suggests a single tree. It imagines innovation as a stable domain of opportunity where the easiest gains disappear first and everything after that becomes progressively harder. That is sometimes true locally. It is much less true systemically.

Innovation does not only pick fruit. It also grows new trees.

The more useful question is not whether the easy fruit eventually runs out. Of course it does. The question is who creates the conditions in which new easy fruit appears at all.

The problem is not the fruit. It is the next tree.

Innovation Does Not Happen on One Tree

Electricity did not merely solve a set of existing problems. It created new categories of problems that could suddenly be solved. Computing did the same. So did the internet. More recently, advances in molecular biology and artificial intelligence have opened spaces where some kinds of experimentation, simulation, design, and analysis can happen faster than before.

What looks like easy progress from the outside is often not easy in any absolute sense. It is easy because the timing has changed. Tools, infrastructure, knowledge, standards, and institutions have aligned. Once that happens, a wave of opportunities becomes visible. Ideas that would once have been impossible, uneconomic, or simply unintelligible begin to look obvious.

The internet is the cleanest example. Search engines, online marketplaces, social networks, streaming platforms, cloud software, and digital publishing did not emerge because earlier generations lacked imagination. Many of the underlying desires were old: finding information, selling goods, communicating across distance, publishing without permission, coordinating work, storing files, reaching customers.

What changed was the environment. Packet switching, networking protocols, cheaper computing, academic and military research, personal computers, browsers, and later broadband created a field in which those old desires could be turned into new businesses. DARPA’s own account of ARPANET shows how early networking work depended on publicly supported research into packet switching, radio networks, satellite networks, and the standards that allowed separate systems to communicate. The ideas did not become good because founders suddenly became clever. They became actionable because the surrounding world had changed.

In hindsight, some of the resulting companies can look almost inevitable. Of course people wanted better search. Of course they wanted to buy things online. Of course businesses wanted software accessible through a browser. But inevitability is often what possibility looks like after the infrastructure has arrived.

Low-hanging fruit is not just discovered. It is enabled.

Frontiers and Fields

The useful distinction is not simply between easy ideas and hard ideas. It is between unopened domains and opened ones.

At the frontier, the work is expensive, uncertain, and often hard to justify in ordinary commercial terms. The tools may not exist yet. The market may not be legible. The applications may be speculative. Much of the work produces dead ends, partial results, or knowledge whose importance is only visible later.

Once a domain has been opened, the character of progress changes. Experiments become cheaper. Feedback loops shorten. More people can participate. The questions become more concrete. Entrepreneurs, engineers, researchers, and users can begin exploring the space with a speed that was impossible at the frontier stage.

This is why technological progress often feels uneven. For long periods, a field may seem slow, abstract, or wasteful. Then, once enough pieces are in place, it appears to accelerate. The apparent burst of easy innovation is not a refutation of earlier difficulty. It is the result of earlier difficulty being absorbed into tools, institutions, and infrastructure.

Progress, seen this way, is not a single curve. It is a sequence of resets. One field matures and becomes harder. Another opens and suddenly fills with accessible problems. The system does not escape difficulty, but it can periodically change where the difficulty sits.

Where Venture Capital Enters the Story

This is also where venture capital needs to be understood more carefully. The simple version says that venture capital funds innovation. That is true, but incomplete. Venture capital is often very good at exploring a newly opened field. It can identify which opportunities are commercially viable, push them toward scale, and discover business models around technologies whose basic possibility has already been established.

That is not a minor role. Harvesting matters. A domain that has been opened but never explored remains mostly theoretical. Without commercialization, standardization, iteration, and distribution, many technologies would stay trapped in laboratories, universities, corporate research groups, or government programs.

But venture capital is not equally suited to every kind of uncertainty. It is usually strongest once a frontier has become legible enough to form companies around it. It can fund risky commercialization. It can fund product discovery. It can sometimes fund deep technology, especially when the scientific path is becoming clearer and the potential market is large enough. At the same time, research on science-based startups and venture capital funding suggests that companies rooted more deeply in frontier science may face specific frictions in securing VC funding. What venture capital struggles to fund is open-ended research whose applications, timelines, and economic shape are still obscure.

This does not mean that public research creates and private capital merely decorates. That would be too neat. Private companies have built new domains. Corporate laboratories have made fundamental contributions. Startups can push science and engineering forward, especially when new tools lower the cost of experimentation. The boundary between frontier creation and commercial exploration is porous.

Still, the distinction matters because incentives differ. Venture capital usually needs a plausible path to ownership, scale, exit, and return. Basic research can generate enormous value while offering none of those things in a form an investor can capture. A new protocol, a scientific instrument, a mathematical method, a database, a materials breakthrough, or a publicly funded research ecosystem may create the ground on which later businesses stand, without itself looking like a business.

In that sense, venture capital is often better at exploring the orchard than planting the first tree.

Who Plants the Tree?

New domains usually emerge from mixed systems rather than from one heroic source. Public research matters. Universities matter. Corporate laboratories matter. Military and space programs have mattered. Standards bodies, infrastructure spending, procurement, regulation, and patient private capital can all play a role.

The point is not that the state creates everything and markets merely harvest it. The point is that different institutions tolerate different forms of uncertainty.

Markets are powerful once value can be priced. They are less comfortable when the value is unclear, the timeline is long, and the eventual application may belong to someone else. Public and institutional research can sometimes absorb that uncertainty because it is not judged only by immediate product-market fit. In pharmaceutical innovation, for example, Andrew Toole’s study of public basic research and industrial innovation found that NIH-funded basic research had a significant effect on the entry of new drugs, especially at the earliest stage of discovery. Corporate laboratories, when protected from short-term pressure, can do something similar. Universities can preserve questions before they become business plans.

None of this makes such institutions automatically efficient. They can be bureaucratic, slow, politically distorted, or wasteful. Public money can be spent badly. Universities can become status machines. Corporate laboratories can be cut apart when the strategic mood changes. Long-term research is not virtuous simply because it is long-term.

But these institutions are capable of a different kind of work. They can expand the space of what is possible before that space has obvious commercial value. That is the work a healthy innovation system cannot afford to neglect.

The Real Slowdown Risk

If all progress depended on one tree, innovation would naturally slow as the fruit became harder to reach. The remaining discoveries would require more effort for smaller returns. In some mature fields, that may be exactly what happens.

But the broader system depends on renewal. New domains reset the difficulty curve. They create fresh layers of accessible problems. They make new forms of experimentation cheap enough for more people to attempt. They turn impossibility into engineering, and engineering into ordinary business.

The risk, then, is not simply that we run out of ideas. Human beings are rarely short of ideas. The risk is that we stop creating the environments where ideas become testable, useful, and cheap enough to matter.

A society can still look innovative while this happens. There may be constant product releases, new apps, better interfaces, improved logistics, more efficient advertising, and endless recombinations of existing tools. The surface can remain lively. Capital can flow. Companies can form. Presentations can be filled with the language of disruption.

But if most activity happens inside domains opened by earlier generations, the system may be living off inherited possibility. It can harvest brilliantly for a time while slowly narrowing the future.

Harvesting Is Not Planting

This is why the debate about whether innovation is becoming more expensive can become misleading. It treats progress as if it were a single process, when it is really several processes stacked on top of each other.

There is the work of opening new domains. There is the work of exploring them. There is the work of scaling what works. There is the work of optimization, recombination, and distribution. These stages overlap, but they are not the same. They require different institutions, incentives, time horizons, and tolerances for failure.

Confusing them produces bad expectations. We start asking venture capital to behave like basic research. We ask universities to behave like startups. We ask public institutions to justify frontier work with near-term commercial metrics. We ask markets to fund things whose value cannot yet be seen by markets. Then we are surprised when the system becomes better at refinement than renewal.

The visible part of innovation is often the easiest to measure: new companies, new products, rapid iteration, rising valuations, adoption curves. The less visible part is slower and harder to quantify. It is the work that changes what can be attempted in the first place.

Both are necessary, but they are not interchangeable. One explores a newly opened landscape. The other opens the landscape.

The Next Tree

The low-hanging fruit metaphor is useful because it captures something real. Fields mature. Easy opportunities disappear. More effort is often needed to achieve the same visible gain. Any serious account of progress has to acknowledge that.

But the metaphor becomes dangerous when it makes innovation sound like the gradual exhaustion of a fixed world. The history of technology is also a history of new worlds becoming available: electricity, aviation, antibiotics, nuclear physics, computing, the internet, molecular biology, machine learning. Each began with hard, uncertain, often expensive work. Each later created zones where new ideas suddenly became easier to find.

The question is not only whether we are picking the remaining fruit efficiently. It is whether we are still doing the slower, stranger, less immediately defensible work that makes new fruit grow.

A society can keep harvesting for a long time from domains opened by earlier generations. It can look energetic while drawing down inherited possibility. The narrowing only becomes visible later, when iteration remains lively but the horizon stops moving.

The danger is not that we run out of fruit. It is that we forget how much work it takes to plant the next tree.