The Great Illusion

“Shall we play a game?” — WOPR

A Century after Norman Angell, wargames keep reaching the WOPR conclusion. The only way to win is not to play.

If you have been following my scribblings here you know I have spent thirty years building machines that let people fight wars without dying. I started in 3D computer game graphics, spent time making military simulations, helped found a game company with Tom Clancy, ran Virtual World Labs for the 100 year old aerospace company Lockheed Martin, and now build AI-infused digital twins and simulation systems for the DOW, the Space Force, NASA and NOAA and healthcare companies. My professional life has been the careful construction of modeled and simulated 1st-4th order consequences of actions. It is all in my book The Simulation Century.

In the back of my mind is Professor James Leutze’s classes on military history at UNC Chapel Hill where he had me read Norman Angell’s The Great Illusion, which led me from there to Paul Kennedy’s The Rise and Fall of the Great Powers. Those books echo in my brain as I watch our investments in kinetic warfare reverberate in our economy and around the globe.

You can be forgiven for not being familiar with this old dusty tome. Angell published his opus in 1909, and by 1913 it had sold something like two million copies in dozens of languages. His argument was not the one he is remembered for. He is remembered for saying that war had become impossible, which he never actually said, and which the Somme made into a permanent joke at his expense. What he wrote was narrower and Professor Leutze’s emphasis to us was that the economies of industrial nations had become so entangled through credit, trade and shared financial machinery that military victory could no longer transfer wealth from the loser to the winner. You could take the territory. You could not take the prosperity, because prosperity (the spoils of yore) was not a chest of gems waiting to be carried off. It lived in contracts, in confidence, in the willingness of strangers to lend and ship and buy. The modern world was endlessly and recursively entangled long before Friedman’s Lexuses and Olive Trees.

The leaders of 1914 read him, admired him, and went to war anyway. Angell got the Nobel Peace Prize in 1933, which is one of history’s better dark jokes, and the century spent the next several decades proving his mechanism correct while insulting his conclusion.

What I have learned through the larger simulations that extend beyond the kinetic is that if you build a good enough model of a modern war, and you run it, Angell shows up in your output whether you invited him or not.

What the runs actually say

The wargames I played on boards in college like Squad Leader, Panzer Blitz and Stalingrad were force-on-force. Blue units, Red units, terrain, weather, attrition tables. You won by holding ground and killing things, and the scoring stopped at the ceasefire. That is not what we build now. A serious contemporary wargame is an economy with a war in it. We model the intricate intertwined DIME and PMESII elements We model the industrial base, the shipping lanes, the port throughput, the machine tool inventory, the labor pool, the energy grid, the semiconductor supply, the sovereign debt curve, the insurance rates on hulls transiting a contested strait and political sentiment. We do a great deal of work on contested logistics, which is a polite term for what happens when the things you assumed would show up do not show up. (Because a wide variety of adversaries, even other than your main adversary, are working hard to prevent them from showing up.) Play those models and the interesting result is never who took the objective. It is what the objective cost. And whether and how long you can hold it.

I cannot remember the last time I saw a run where the winner came out ahead in any accounting of blood and treasure. The mechanisms are boringly consistent. Capital stock burns and has to be rebuilt with borrowed money at wartime rates. Skilled labor is conscripted, killed, or emigrates, and human capital does not respawn. Trade networks that took forty years to weave get cut in an afternoon and reroute around you permanently, because your customers learn that you are a single point of failure and can’t be trusted. Sanctions run in both directions no matter who writes them. The productive economy retools toward output that is consumed at the moment of use and compounds nothing. And every dollar spent on ordnance is a dollar not spent on the things that pay back for fifty years. Certainly individual defense stocks may prosper from the resupply. But that is rarely ever good for the entire economic ecosystem, regardless of how many souls they employ.

The vanquished lose obviously and immediately. The victor loses on a delay, which is why the loss is so easy to mistake for a bill someone else will pay.

History has already run the experiment on hardware. Britain led the world in GDP for a few decades, won both world wars and mortgaged an empire doing it, finishing the second as a debtor to the country it had financed a generation earlier, and making the final payment on its American war loans in December of 2006. The Soviet Union won its war and then spent forty years converting an economy into a military and calling the result strength. The United States is the case everyone points to as the exception, the one great power that got rich from a war, and I would argue the causation is backward. America did not prosper because it won. It prospered because of what it did with the twenty years after: the GI Bill, the National Science Foundation, the National Institutes of Health, the interstate highway system, the land grant universities pushed into overdrive, a national commitment to basic research at a scale no one had attempted. The war did not build that. The war interrupted the building.

Kennedy’s ledger

Paul Kennedy made the same point from the other end in The Rise and Fall of the Great Powers in 1987. Read across five centuries, his finding was that military power rests on an economic base, that the base is finite, and that great powers reliably reach a stage where the cost of defending everything they hold exceeds the productive capacity that let them hold it. He called it imperial overstretch. Habsburg Spain drowned in silver and credit defaults. Britain policed a quarter of the planet while its industrial lead evaporated. The pattern is not that these powers lost wars. Several of them won. The pattern is that security spending crowded out the investment that generated the wealth that funded the security spending, and the curve caught up with them.

Kennedy also got his timing wrong, and with the benefit of hindsight I have to acknowledge that he missed on Japan. He wrote in the late 1980s, when Japan looked ascendant and American decline felt like arithmetic. Japan stagnated. The United States had a boom. Critics have used that miss to dismiss the book for almost forty years. I think they are dismissing the wrong part. His prediction was a guess about rates. His mechanism was a description of a machine, and the machine is still running. What saved the United States in the 1990s was precisely the thing his model says saves you; a burst of productivity from a technology base that public science investment (ARPANET) had paid to invent decades earlier, and had kept paying for.

The variable

When we run comparative scenarios, the divergence between a country that ends the century strong and one that ends it managing decline almost never comes from the outcome of wart. It comes from what the country did with the resources that war did not consume. Money into basic research, into education, into power generation, into transportation, into water, into the unglamorous physical layer everything else sits on, produces returns that compound. Money into force structure produces deterrence, which is important and has real value because economies crave stability, but it produces nothing else. Nations that hold that ratio are still there in the endgame. Nations that let it drift do not get conquered. They get overtaken, which is slower and more humiliating.

My company publishes a caveat with every system we ship. A simulation is not a prediction. It cannot be. Our wargaming sims carry a binding claims register that forbids anyone on my team from saying otherwise, and I would apply the same rule to this paper. What a model gives you is not the future. It gives you your own assumptions, made visible and forced into contact with each other, run forward until they say something you did not expect at the 2nd-4th consequence layers. That is the whole value. The 1914 general staffs had models too. Theirs were built by people who had been instructed which answer to produce, and the models obliged, and the assumptions never got tested until they were tested with blood and treasure.

We are in another season of rearmament, and I do not think that is entirely wrong. Deterrence is real, and needful, and weakness invites the thing everyone is trying to avoid. But I would like the countries doing it to run the whole ledger and not just the part that ends at the ceasefire or withdrawal. Ask what the money would have compounded into. Ask what the port, the grid, the lab and the graduating class would have been worth in 2060. Then decide.

Angell’s illusion was never that war would stop. He knew better. The illusion was that anyone could benefit from it economically. A hundred and fifteen years of the US leading the world in GDP and a great many hours of watching my own machines grind through the consequences have not turned up a counterexample I believe. The victor and the vanquished walk off the same field, holding different stories and the same emptier accounts.

Twenty-five years, two strategies

We do not have to wait for a war to test this. The last quarter century is already a natural experiment, and the results are sitting in public data.

In 2000, China’s economy was roughly an eighth the size of America’s, about $1.2 trillion against $10.25 trillion. By 2025 the figures were around $19.5 trillion and $30.8 trillion. Let me immediately complicate the story I am telling, because the reality is more interesting and complicated than my simple thesis. Measured in dollars, the gap between the two economies is now the widest it has been in the entire sixty-five year World Bank series. China peaked at about 77 percent of American size in 2021 and has slid back to the low sixties. Per person, Americans still produce roughly six times what Chinese workers do. Anyone selling you a clean overtaking narrative is selling you something. Measured at purchasing power parity, China passed the United States around 2016 and now sits near $41 trillion, which is the number that matters if what you want to know is how many bridges or missiles or fabs a country can actually build.

The level is not the interesting part. Kennedy’s question is the interesting part. What did each country do with the twenty-five years?

China spent it buying interdependence instead of taking it. Cumulative Belt and Road engagement since 2013 has reached about $1.4 trillion across 150 countries, and 2025 was the largest year on record at $213.5 billion. By late 2025 the China-Europe rail freight service had run 120,000 trips. Read that through Angell and the strategy is almost too on the nose. You do not have to conquer a port if you financed it, built it, and hold the operating contract. These investments will pay dividends in the future.

Domestically, the model came in part from an American. Premier Li Keqiang read Jeremy Rifkin’s The Third Industrial Revolution, half a million copies of which were printed in China, and made its logic central to the 13th Five-Year Plan: a communications internet, an energy internet and a mobility internet converging on one platform, with the Internet of Things stitched through all of it. I have my disagreements with Rifkin as a forecaster that I mention in my book The Simulation Century. That is beside the point. The point is that a government read a book about infrastructure convergence and then poured concrete. China now operates roughly 48,000 kilometers of high-speed rail, more than 70 percent of the world’s total. Its R&D spending hit about $550 billion in 2025, 2.8 percent of GDP, above the OECD average for the first time. (US spend is3.4%) It holds more than five million valid domestic invention patents. Nine of the top ten institutions in the 2025 Nature Index are Chinese, up from one in 2016. The Australian Strategic Policy Institute finds China leading high-quality research in 66 of 74 strategically significant technologies. In 2024, by the OECD’s purchasing-power measure, Chinese R&D spending passed American R&D spending for the first time.

None of that is an endorsement. The same record Belt and Road year was simultaneously the greenest and the dirtiest in the program’s history, with fossil fuels accounting for more than three quarters of China’s overseas energy work. The property sector is a slow-motion accident, youth unemployment sits above 16 percent, the demography is brutal, and I would not trade the American political system for theirs at any price. I am not arguing about regimes. I am arguing about where the money went.

Which brings me to my own country, and to the part of this essay I would rather not have to write. The administration’s fiscal 2026 budget request sought to cut the National Science Foundation by 57 percent, the National Institutes of Health by 41 percent and NASA by 24 percent, in the same document that raised defense spending toward $1.5 trillion, an increase of 44 percent. Congress refused most of it, and the fiscal 2027 request came back asking for a 55 percent NSF cut anyway. Meanwhile the damage moved into the plumbing, where budget headlines do not reach. The NSF is on pace to issue the fewest new grants in four decades. Something like 8,000 grants were canceled across NIH and NSF. Federal science agencies shed nearly 118,000 employees between September 2024 and February 2026. In a Nature survey of more than 1,600 scientists, three quarters said they were considering leaving the country. As I am writing this we are hearing about new breakthroughs in MRNA cancer vaccines. Why are we not doubling down on that? If elected, I promise to…

I would rather not reach for the phrase anti-science, because slogans end arguments instead of settling them. In Paul Kennedy’s vocabulary the description is more precise and more damning. This is a deliberate reallocation from the compounding side of the national ledger to the consuming side. Note which category is being cut hardest. Basic research is the one area where the United States still holds a decisive lead over China, roughly half a percent of GDP against a fifth of a percent. That is the seed corn, and it is the thing on the block.

Here is the part my wargame sim experiences taught me to watch for. Nothing bad happens in the year you cut a research budget. The lab keeps its lights on. The graduate student finds another advisor. The paper gets written somewhere else. The line item disappears and the country feels exactly the same on the day after, which is precisely why the decision is so easy to make. The cost lands ten to fifteen years out, as a cancer cure that was never invented, a company that incorporated in Shenzhen or Geneva instead of San Diego, a cohort of engineers who trained abroad and stayed there. It is the victor’s delay again, applied to peacetime. Human capital does not respawn.

The countries that win the century are the ones that build something while everyone else is squabbling.