The Gun Pointed at the Problem

You are sitting in a government briefing room, somewhere in the late 1950s, and someone has just placed a single sheet of paper in front of you. The Soviets have put an object in orbit. It is beeping. The United States, with a defence budget that dwarfs Moscow's, cannot do the same thing. Not yet. Not because of money. Because no American contractor yet knows how to build what just crossed the sky. That gap, concrete and humiliating, is what actually changes history.

The pattern repeats often enough to feel like a law. Two powers lock horns, and within a generation some technology that would have taken decades to mature arrives in a handful of years. The space race gave the world integrated circuits and satellite communications as side effects. The Second World War compressed radar, pressurised aviation, and the foundations of computing into roughly six years. The Cold War, sustained over four decades, produced the internet, GPS, and nuclear power generation, none of which were the intended product.

And yet rivalry alone explains nothing.

The Iran-Iraq war lasted eight years and produced no meaningful technological acceleration on either side. The decades-long standoff between India and Pakistan generated a nuclear deterrent but left both countries importing most of their advanced military hardware. Plenty of rivalries have ended with both parties poorer, more isolated, and no closer to any technological frontier than when they started. So what actually separates the rivalries that produce breakthroughs from those that produce stagnation? The answer is structural, not motivational, and it rewards careful unpacking.

The Four Conditions That Actually Matter

The first condition is the presence of a specific, measurable performance gap that neither side can close by purchasing its way out. When the United States confronted the reality of Sputnik, the problem was not a production shortfall. It was a technical frontier that no amount of procurement spending could buy across. That kind of gap forces investment into research rather than manufacturing. It redirects resources toward the unknown, which is the only place genuine breakthroughs live.

The second condition is institutional continuity. A rivalry that lasts long enough, and remains stable enough in its basic structure, allows organisations to accumulate knowledge across multiple funding cycles. DARPA, founded in direct response to Sputnik, has funded research for decades on the premise that some investments take fifteen years to pay off. That institutional patience is not the default state of government. It requires the sustained pressure of a credible rival to justify it politically. Rivalries that flare and cool unpredictably produce bursts of spending followed by defunding, which is the most reliable way to destroy a research programme before it matures.

The third condition is what you might call productive adjacency: the technologies being developed have to share components, knowledge, or talent pipelines with civilian or commercial applications. The US semiconductor industry did not grow because defence contractors wanted to sell chips to consumers. It grew because the same miniaturisation problems that made a guidance computer fit inside a missile also made a calculator fit in a pocket. The knowledge leaked, as knowledge always does. Rivalries that produce highly classified, deeply siloed research, certain Soviet biological programmes being the obvious example, generate less spillover and therefore less compounding acceleration.

The fourth condition, and the one most analyses skip entirely, is genuine uncertainty about who will win. This is, frankly, the actual engine. A rivalry in which one side is so obviously dominant that the outcome is not in doubt tends to produce complacency on top and despair, or simple imitation, below. The most productive rivalries are those in which both sides believe, with reasonable justification, that the other might pull ahead. Remove that mutual uncertainty and you remove most of the pressure that makes the whole mechanism work.

Where Rivalries Go Wrong: The Stagnation Trap

Consider a plausible scenario. Two regional powers share a long border and a bitter territorial dispute. Both spend heavily on military hardware. Both buy that hardware from the same three international suppliers, because neither has invested in a domestic defence-industrial base capable of producing it. The rivalry generates procurement spending, not research spending. It funds foreign engineers, not domestic ones. Each crisis produces a new arms purchase rather than a new capability. After thirty years, both countries have more sophisticated equipment than they did, but the equipment was designed somewhere else. The technological frontier has not moved for either of them.

This is the stagnation trap, and it is more common than the acceleration story. It springs from a specific absence: no capability that cannot be bought off the shelf, no institutional depth to sustain multi-year research, no spillover potential to compound gains over time. Remove any one of those conditions and rivalry tends to produce inflation in military budgets rather than innovation in military technology.

There is a related failure mode on the other side of the ledger. A dominant power, facing a rival it genuinely fears, can over-classify its research to the point where the knowledge never escapes into the broader economy. Some historians argue this is precisely what happened with certain Soviet space-programme advances. The capabilities existed, locked inside institutions with no mechanism for transferring knowledge to civilian engineers. Secrecy, taken too far, becomes its own form of stagnation. A library no one can enter is not a library.

The Spillover Is the Point

Here is the wrinkle that should change how you think about all of this. The technologies most people associate with Cold War acceleration, GPS, the internet, microwave ovens (yes, those too: a radar engineer named Percy Spencer noticed a chocolate bar melting in his pocket near an active magnetron in the 1940s), were not the intended outputs of the programmes that produced them. They were accidents of adjacency. The intended outputs were weapons systems and communications infrastructure. The civilian world got the offcuts.

The social return on a rivalry-driven research programme is often far larger than the military return, and is almost impossible to predict in advance. The people funding ARPANET were not trying to build a global communications network for civilians. They were trying to solve a specific military routing problem. The fact that the underlying protocol turned out to be the most consequential communications technology since the printing press was not in any project brief. Not remotely.

This creates a genuinely uncomfortable observation for anyone who wants to plan for technological acceleration: the most productive rivalries are the ones in which the research is broad enough, and the institutions permeable enough, that knowledge escapes in directions no one anticipated. That is almost the opposite of how most governments think about directing research spending. Most governments want legible returns on legible investments. History suggests the best returns are neither.

What Stagnation Looks Like from the Inside

Take two engineers, Mara and Dmitri, both hired by rival state aerospace agencies in the same year, both working on propulsion problems of similar technical difficulty. Mara's agency publishes findings through a university partnership. Her work is cited by a commercial rocket startup within four years, and the startup's eventual success pulls three more engineers into the field, each of them building on what she established. Dmitri's agency classifies everything at the highest level. His findings sit in an archive, cited by no one, and when his programme is defunded in a budget cycle, the knowledge disappears with the team.

Same rivalry. Same investment level. Radically different compounding.

Ask yourself: how many Dmitris does it take to cancel out one Mara?

The lesson is not that secrecy is always counterproductive. It clearly is not. The lesson is that the structural choices made around a research programme, who can see it, who can build on it, what adjacent fields it touches, matter at least as much as the size of the budget or the intensity of the geopolitical pressure driving it. Rivalry supplies the urgency. Structure determines whether that urgency produces anything worth having, or simply vanishes into a classified archive when the money runs out.

Rivalry is the gun pointed at the problem. The question worth asking is whether anyone bothered to aim it.