Volume VI: Nested Causal Models, Maps, and Interventions
Contains the formal NCM and game-theory research program.
Media / Strategic systems
Nested Causal Modelling and Game Theory
What changes when players and payoffs are placed inside the systems producing them.
Standard game theory remains useful inside its declared enclosure. NCM adds the outer institutions, inner cognitive states, changing resources, and deeper consequences that can reverse a payoff or alter which strategies are available.
Terms in this film
Explore the same term in other films through the transcript glossary.
Continue into the research
Related work
Game Theory Inside the Enclosure
An accessible account of the comparison.
NCM Game Theory Supplement
Technical companion with models and tests.
Read the video transcript
Published captions, with their original wording.
Alright, let's jump right in. Welcome to today's explainer, where we're going to unpack a genuinely fascinating paper by Sid J.A. Hubbard. It's called "Nested Causal Modeling and Game Theory", a ternary enclosure experiment in strategic prediction.
Okay, I know, I know that title is quite a mouthful, but trust me, what Hubbard is doing here is incredibly ambitious, and honestly, really practical. He's basically taking two massive powerhouse frameworks for understanding human behavior and strategic choice,
and running this highly structured experiment to see what happens when we just smash them together. If you've ever looked at a strategic model and felt like, you know, there's something absolutely crucial missing just out of frame. You are going to love this.
So, imagine this for a second. What happens when you suddenly realize the game you're playing is actually trapped inside another game? Game theory is brilliant, right? But it operates inside a very strictly bounded box. It assumes the board is set, the rules are written down, and the players are locked in.
But out here in the real world, nothing is ever truly isolated. We can always zoom out. And that's the seed we're planting today. What if we could model not just the game itself, but the massive, unseen environments that swallow the game whole?
Here is our roadmap for today. We're going to cover the limits of the game, enter nested causal modeling, put NCM inside game theory, look at their parallel predictive power, explore the game formation layer, and finally build a much better strategic map.
Okay, part one, the limits of the game.
We've got to give credit where credit is due. Game theory is arguably one of the most successful formal languages we have for mapping out strategic interaction.
It is just phenomenal at handling strategy, incentives, bargaining, all of that. But it's got a structural blind spot.
Game theory basically asks, what happens after the game is bounded? It just accepts the fixed board and figures out the most rational moves inside it.
Nested causal modeling, or NCM, flips the script entirely. It asks, wait, what boundary made this game appear in the first place?
One framework accepts the limits, the other questions them.
To figure out how these two systems talk to each other, Hubbard brings in a concept called zero infinity.
Now, don't let the name scare you, we're not doing quantum physics today.
For our purposes, zero infinity is just the ultimate blank slate. It's this perfectly balanced, steady state baseline.
Both game theory and NCM depart from this baseline to create their models. Neither one is reality, they both step away from this blank slate to try and describe reality.
The really interesting part is how they take that step.
Which brings us to part two, enter nested causal modeling.
If game theory is a flat chessboard, NCM is more like a giant set of Russian nesting dolls.
Instead of just looking at players making choices on a flat surface, NCM looks at causal descent, how bigger systems contain, squeeze and shape the smaller systems inside them.
To map this out, NCM uses a core equation called the ternary enclosure function. It's got three parts.
E-left is the enclosing condition, basically the outer environment. D middle is the active departure or the relation, and E-right is the enclosed condition, the inner field.
Think of this little equation as our lab equipment. We're going to run three distinct experiments by swapping game theory and NCM into different spots on this formula, just to see what happens.
So how do they actually depart from that zero infinity blank slate? Well, game theory departs by defining players, defining strategies, and defining payoffs.
NCM, on the other hand, departs by defining enclosures, causal flow and boundaries. The big takeaway here? These frameworks are not enemies, they're not competing for the exact same job, they're just entirely different ways of slicing reality, looking through different lenses at the exact same problem.
All right, let's run the first experiment, part three, NCM Inside Game Theory. For this first test, we're taking NCM, shrinking it down, and literally trapping it inside game theory.
So game theory is the outer environment, and NCM is stuck inside. Hubbard asks, "Can NCM actually be expressed as a game?" The answer is, yes, sort of.
You get what he calls the enclosure game. In a normal game, you just pick your strategy and play, but in the enclosure game, you can literally move across scale.
You can stay in your current game, sure, but you could also escalate to a larger game, deescalate to a smaller one, or just exit the enclosure entirely.
Players can actively rewrite the game board to escape zero-sum traps. It is wild. It gives game theory a formal way to model changing scales without pretending the boundary is permanently locked.
Let's make this super concrete with this concept of infinite depth. Imagine a company playing what they think is just a localized supply chain game.
Pretty straightforward, but wait, that game is actually nested inside a much bigger logistics game spanning global trade routes.
And guess what? That logistics game is swallowed up by a massive high-stakes political game between rival nations.
Suddenly, the boundary isn't the limit. A player might make a totally irrational move in the supply chain game, but it makes total perfect sense because they're actually trying to make a change.
Because they're actually trying to win the political game three layers up. Moving right along to part four, parallel predictive power.
Okay, experiment number two. This time we're not stuffing one inside the other. We're running them side-by-side as parallel tools.
When do you actually use which? Well, game theory is your absolute best friend when your actors are obvious, the strategies are clear, and finding an equilibrium helps.
It is the master of predicting incentives. But NCM? NCM is your secret weapon when the system is heavily nested.
The boundaries are super blurry, and a shift at a high level triggers a cascade of risk down below. NCM specializes in predicting causal descent.
To see this play out in reality, Hubbard uses the NDAA section 219 case about alliance breakdowns.
When things started falling apart, game theory beautifully nailed exactly why defecting became such an attractive incentive for the players.
But NCM caught something else. NCM showed how the exact same integration pathway totally changed its meaning because the larger enclosing state flipped from a supportive alliance to an adversarial breakdown.
Game theory saw the incentive. NCM saw the causal descent. You literally need both to get the full picture.
Which brings us to part five, the game formation layer.
Let's fire up the third and final experiment, swallowing game theory inside NCM. Now, Hubbard notes that NCM doesn't just eat game theory raw.
There's a middleman in interstitial space called the game formation layer. Think about the invisible setup that happens before a game even begins.
Who even counts as a player? What actions are actually allowed as strategies? What consequences count as payoffs? Who drew the boundary?
Game theory is completely blind here. It totally relies on this game formation layer to even exist.
So in this setup, game theory just receives a nicely packaged, well-formed game and works its magic.
But NCM steps back, looks at the packaging and says, "Who built this?" What outer environment formed this game layer?
What uncounted costs are being quietly dumped outside the payoff matrix?
NCM isn't destroying game theory here. It's essentially performing a boundary audit to make sure the game isn't rigged by hidden, displaced costs before anyone even makes a move.
Finally, part six, a better strategic map.
This is the Grand Finale, the ultimate hybrid system Hubbard proposes. He calls it enclosure-aware game theory.
And guys, this is where the real magic happens. It keeps all the amazing tools of game theory, the players, the strategies, the equilibria.
But it injects the raw power of NCM, nested boundaries, boundary audits, and the ability to dynamically switch enclosures.
You can now simulate the choices your players make inside the game, while simultaneously simulating changes to the very environment that makes the game possible.
I absolutely love this quote from the paper. It just hits the nail on the head.
Game theory takes the player seriously. NCM takes the enclosure of the player seriously.
Merging them isn't about one beating the other. It's about combining them to create a vastly superior map of reality.
So I'll leave you with this one final thought.
The next time you're mapping out a strategy, whether it's for your business, logistics, whatever it is, ask yourself, is the strategic game you're analyzing?
Actually, the right level of reality? Or are you totally missing the larger enclosure that's actually pulling the strings?
Keep questioning the board and always look for the unseen boundaries.
Thanks so much for hanging out with me for this explainer. I hope it blew your mind as much as it did mine.