Imagining the first global Simulation War

20.34, Wednesday 4 Aug 2021

I’ve been imagining a future global war where nobody is killed, no city is attacked, and it’s conducted entirely in simulation – unless and until it can become real.

I haven’t played Go much but it feels like a game of frozen anticipations. You anticipate what the other might do, in their attempt to enclose you, and you place a stone to prevent that. And they do the same. So the entire game board becomes this network of cautious concrete counter-plays to imagined threats. It’s like you spend most of the game negotiating the landscape of the game to come.

But if at any point you see a sequence where you can enclose your opponent, and you see that they haven’t anticipated that play, you go for it.

So I imagine that this is what military planning is like.

London is littered with contingencies.

There’s a literal warship parked in the Thames. HMS Belfast. It’s a museum and tourist destination. But the guns still work – one is pointed at a motorway service station 11 miles away. Which is amusing, right? Ha ha. And also a little nod to any country that has contemplated invading London. I bet there’s a freeze-dried plan for how the Belfast would be re-commissioned, and somebody somewhere knows to the hour how long that would take.

Then the big parks, which are beautiful and well maintained. Great for bivouacking large numbers of troops. I bet that’s another plan. In the Second World War, Hyde Park was used to grow vegetables.

I wonder

  • how many contingencies there are like this in London
  • who is responsible for protecting them, ensuring they are fit for purpose, and what the mechanism is to override City Hall planning decisions
  • whether London’s urban planning is studied, secretly, by other countries, and whether the presence of these contingencies has dissuaded any untoward actions.

Possibly not in 2021, right? But London is an ancient city with a long history and a long future. Maybe you have to plan cities for what the world might be like in 100 or 200 or 300 years time.

Rachel Abrams shared a paper with me about urban planning and the drive to suburbia: Galison, Peter. “War against the Center.” Grey Room, no. 4, 2001, pp. 7–33. JSTOR,

Urban planners in the US looked at the devastation of Nagasaki and said, oh this could happen to us. (Having just done it.) So they looked for concentrations of, say, the steel industry, or the new computer industry, and they mandated that major offices were built outside the blast radius if an atomic bomb were to be dropped a major population centre. And so you have suburbia to serve those offices and so on. The strategy was called dispersal.

Here’s the full PDF.

BUT what grabbed my attention in this paper was the bombing campaign on Nazi Germany.

The campaign was guided by the U.S. Strategic Bombing Survey, an immerse affair, employing well over a thousand people – many of them Operations Analysts.

“Operations analysis” was essentially a methodical theoretical reconstruction of the interconnections that held together the German economy and war machine and that asked how it could be blown apart.

And so:

But the operations analysts selecting targets were not just after particular pieces of munitions factories; their goal was to precipitate a collapse of the German economy as a whole. To that end, they directed a series of studies designed to locate just those plants where destruction would cause shortages to ripple through the entire system. Operations followed. Henry “Hap” Arnold, for example, tempted Harry Hopkins with the notion that blasting the German ball bearing industry “would probably wreck all German industry.”

Ball bearings! (And indeed that is what the bombing campaign did.)

But this process is exactly what I mean. Reverse engineer an economy, or a society, or a game-player’s strategy, and figure out the single thread to pull that unravels the whole thing.


I took one of those careers quizzes at school when I was 15 or so – we all had to. Brits of a certain age will remember the DOS quiz with the text-only interface and a hundred multichoice questions.

It gave me two possible future careers at the end: Operations Analyst. And Ceramics. I ended up going to university where I took physics.

My feeling is that it was spot on, but there was no other way for a quiz written in the late 80s/early 90s to say “design strategy.”

I wrote a really-not-very-good short story about this idea years ago. It’s mostly lumps of exposition glued together with minimum viable narrative.

It’s about a fictional board game played on a map of Southampton. The two sides are the Council and the Friends.

The moves are called “counterfactuals” and the starting point is always the actual map of the town.

But it turns out that every play of the game is wargaming an actual, potential future conflict.

‘The Friends haven’t come this close to having certainty since before the new shopping centre was built. The Council stole a march on us then, really changed the board. A great move.’

I’ve never thought of the actual building in Southampton as moves in the game before.

One play-through reveals that there’s a path for the Friends to win – and so it all kicks off.

I don’t know why my head is stuck ploughing this global threat-modelling furrow. But it is: see last year’s post about space, weather, and other novel battlegrounds.

If I were writing that story today, it wouldn’t be a board game, it would be AI. There would be AIs constantly wargaming, constantly running the operations analysis that led to the ball bearing factory target selection.

And maybe that’s part of my fear now? That threat in the 2030s won’t be about somebody realising that social media propaganda can destabilise a society, or some organisation spotting the new ability for a computer worm to infiltrate uranium gas centrifuges and destroy a nuclear program (a decade later and nobody has claimed responsibility for Stuxnet and its cyberattack on Iran).

The discovery process will be automated.

The probing of the attack surface of society will be automated and a thousand times faster than anything we’ve seen to date, whether it’s software engineering or social engineering or knocking out a water treatment plant. Imagine finding a zero day on the economy.

My hope, my wish, is that this finally makes it unthinkable to have enemies because any attack would be unreasonably effective, and so the entire world community embarks on a giant exercises of potlatch and soft propaganda and diplomacy – aggression and self-defence both become questions of: how to make friends.

But actually where my head goes is to a future Simulation War, 2030–2070.

A Simulation War conducted entirely virtually, at hyper speed. The arms race will be measured in an ever-escalating TWPS, trillions of wargames per second, the computational capacity of a nation devoted to hunting for sequences in possible futures that lead to a win state before uncertainty takes over.

We won’t know the virtual cold war is happening aside from the real world moves to change the board itself, the starting conditions. We’ll see weird urban planning decisions, or bizarre industrial strategy capital allocation decisions, or modifications to university curricula, or manipulations of the atmospheric carbon concentration, none of them making sense except in the context of being moves in the game, anticipated defences in a numerically critical proportion of future mirrorworlds.

If you enjoyed this post, please consider sharing it by email or on social media. Here’s the link. Thanks, —Matt.