The Long Game: What the Land Remembers
- Jun 20
- 8 min read
The prisoner's dilemma says defect — and yet the land isn't a one-time game.
Take a single piece of land. The same plot, the same steward, season after season — sometimes decades...
Every season presents a version of the same choice. Take the yield, or invest in what produces the next yield. Push the soil, or feed it. Extract what's available now, or build the conditions for more later.
Looked at as a single season, these feel like practical decisions about this year. Looked at across many seasons, they reveal something else — a pattern of choices that accumulates into a result far larger than any individual round would suggest.
The last post in this series left something unresolved. Markov Chains described a system that moves between states, where the probability of the next state depends only on the current one. The system has no memory. That was useful however it produced a remainder that didn't quite disappear. A calculation that kept coming up slightly short.
That remainder is what this post is about. But to get there properly, it helps to take a little detour.
A Bar, a Think Tank, and John Nash
Most people encountered game theory — without knowing it — through a single scene in A Beautiful Mind, the 2001 film with Russell Crowe as the mathematician John Nash.
A group of men in a bar. One woman everyone is drawn to, several others who arrive with her. The conventional move — everyone competes for the most attractive option — leads, the film suggests, to everyone blocking each other and all walking away with nothing. The better move is to coordinate, spread attention, produce a better collective outcome.
It's a memorable scene. Mathematicians have since pointed out that it doesn't quite represent the underlying theory with precision — the setup is more illustrative than technically rigorous. But as an introduction to a way of thinking — that the outcome of any choice depends on what others choose simultaneously, and that this changes what the rational move actually is — it did its job. Probably nothing else put game theory into general consciousness as effectively.
The actual work, on the other hand, was happening in a very different room.
RAND — Research ANd Development — was a think tank established after the Second World War to provide analysis to the US military. In 1950, two researchers there, Merrill Flood and Melvin Dresher, were thinking about nuclear strategy. Two superpowers, each facing a choice: cooperate on arms limitation, or develop weapons regardless of what the other side does. The stakes were not abstract.
It was in this context that what would become the prisoner's dilemma was formalized — and where Nash's ideas about strategic equilibrium found their most consequential early application. A bar scene and a cold war think tank, separated by half a century and several layers of cultural translation, pointing at the same underlying question.
The Prisoner's Dilemma
The game itself is simple enough to explain in a few sentences, and counterintuitive enough to stay with you once you see it.
Two players — in the original framing, two people accused of a crime, placed in separate rooms where they cannot communicate with each other. No collusion possible, no ability to commit to a shared strategy, no way of knowing what the other will choose.
Each is asked independently: confess, or stay silent.
In game theory terms these become the two available moves: defect, or cooperate. The outcomes depend on the combination of choices.
If both cooperate — stay silent — both get a good outcome. Call it 3 each.
If both defect — confess — both get a poor outcome. Call it 1 each.
If one cooperates and the other defects — the one who confesses walks away with the best possible outcome (5), while the one who stayed silent gets the worst (0).
Now consider the choice from any single player's perspective.
If the other side cooperates, defecting gets you 5 instead of 3 — better to defect.
If the other side defects, defecting gets you 1 instead of 0 — still better to defect.
Defection is the dominant strategy regardless of what the other player does. And the separate rooms are precisely what makes this inescapable — there's no way to agree, no way to trust, no way to commit.
And yet if both players follow this logic — as rational actors independently will — both end up at (1,1) when they could have had (3,3).
This outcome has a name: Nash equilibrium. The point at which neither player can improve their result by changing their strategy unilaterally, even though both would be better off if they could somehow commit to cooperating.
It is individually rational and collectively terrible.
What makes the prisoner's dilemma so persistent as an idea is that this structure appears everywhere. Two nations deciding whether to arm. Two companies deciding whether to undercut each other's prices. Two farmers deciding whether to draw down a shared water source. In each case, individual rationality leads both parties to a worse outcome than cooperation would have produced.
The flaw in the logic — though it doesn't look like a flaw until you see it — is that it only holds for a single round. Play the game once, with no future interaction, and defection dominates.
But what if the game repeats?
Tit for Tat
In the early 1980s, the political scientist Robert Axelrod ran a series of tournaments. He invited game theorists, economists, mathematicians, and computer scientists to submit strategies for a repeated prisoner's dilemma — the same game played again and again between the same players over many rounds. Scores accumulated across the full series.
Which strategy would perform best?
Dozens were submitted. Some were elaborate, built on long pattern recognition. Some were aggressive, some defensive, some attempted to identify and exploit the other side's tendencies over time.
The winner was the simplest entry in the tournament. Two rules only:
Cooperate on the first move. After that, do whatever the other player did in the previous round.
That's it. Start with cooperation. Then mirror.
This strategy — Tit for Tat — won not once but twice, across different tournaments with different participants. It outperformed strategies far more sophisticated than itself, and it did so without ever being the top scorer in any individual pairing. What it did was avoid catastrophic losses, establish cooperation quickly when the other side was willing, retaliate immediately when defected against, and return to cooperation just as quickly once the other side did.
What Axelrod's tournaments demonstrated was something that cuts against a lot of intuition about competition: in a repeated game, reciprocity outperforms both exploitation and unconditional cooperation. Neither the aggressor nor the pushover wins across many rounds. The strategy that holds up is the one that cooperates by default, responds clearly to defection, and doesn't hold grudges.
The shadow of future rounds changes everything about what the rational move looks like in any given one.
Same Land, Next Season
Let's bring this back to the piece of land at the start.
If this season were the only one that mattered — the last round, with no future interaction — a certain kind of extraction makes sense. Take the yield. Don't reinvest in what won't show up until next year. Optimize for what this round returns. As a single-shot decision, it's internally coherent.
But the game doesn't end this season. It's the same land, season after season. A repeated game in the most literal sense — same players, same conditions broadly, same basic choices available, indefinitely.
And in a repeated game, Tit for Tat logic changes the calculation in ways that show up everywhere once you start looking for them.
Take grazing. Push a paddock hard, graze it down to bare ground because the herd needs feeding now, and the immediate yield looks fine — animals fed, no loss this season. But the land registers that as defection. Root reserves are depleted, ground cover that protected the soil from sun and rain is gone, the seed bank that would have regenerated forage thins out. Next season's starting state is worse, and pushing the same paddock again compounds the damage rather than just repeating it. Rotate the herd instead, rest each paddock long enough for recovery, and any single season's grazing looks more conservative — fewer animal-days on that ground this year. But the land cooperates back. Root systems deepen, forage density increases, the paddock can carry more animals next year than it could this one. The conservative move in round one becomes the higher-yield move by round five.
Or take tillage. Plowing breaks up compaction immediately and looks like an unambiguous win this season — easier planting, faster early growth. But it also breaks down soil aggregate structure, accelerates organic matter loss, and exposes biology that took years to build to sun and oxidation. The land defects back slowly — the same field needs more tillage next year to achieve the same effect, not less, because the structure that once held itself together on its own is gone. A no-till or reduced-till approach looks like it's leaving yield on the table in year one. Across a decade, fields under that approach often outperform the tilled ones, because the soil has been allowed to build the structure that tillage keeps tearing down.
Synthetic fertilizer follows a similar arc. It cooperates with this season's yield very effectively — visible, fast, reliable. What it doesn't do is feed the soil biology that would otherwise build fertility on its own. Lean on it exclusively for long enough and the soil's own capacity to hold and cycle nutrients quietly atrophies, so next season needs the same input just to stand still, and the season after that needs slightly more. Compost, cover cropping, and managed inputs build that capacity instead — slower to show results, often less impressive in any single season, but each season's investment becomes part of what the next season can draw on without being asked.
The pattern repeats across each of these: the move that wins the single round is rarely the move that wins the repeated game, and the land doesn't punish or reward instantly. It mirrors back what it was given, on its own timeline, folded into whatever state it hands over next.
This connects to something from the distributions post earlier in this series. Small early investments compound forward multiplicatively. Soil building in year one shapes water retention in year three, which shapes resilience in year five.
Evaluated in isolation, the investment looks modest. Evaluated across the repeated game, it's often the move that makes everything else possible. The single-shot mindset and the repeated-game mindset can look at the exact same season and reach opposite conclusions about what the rational move is.
What the Land Keeps
The current state of the land — soil structure, biological activity, water retention, resilience to stress — isn't a blank condition that exists fresh this season. It's the accumulated residue of every decision before it. The compost applied or skipped. The cover crop planted or not. The rest given or denied. The state doesn't store that history separately. It is that history, compressed.
So a single transition having no memory was true. It just wasn't the right unit to judge memory by. No single round remembers — but a sequence of rounds does, because each round's outcome becomes the next round's starting point. The land was never one transition. It's the accumulation of all of them, and that accumulation keeps score.
What does this season produce now? And what does it leave behind?
The first is what isolated thinking asks. The second is what becomes visible once you notice the game repeats.
The land keeps the score either way. The only real choice is what kind of score it keeps.
Sign up for our newsletter or connect with us on social media to stay up-to-date with our latest posts and permaculture inspiration.
Explore our inspiring series and posts:
Love the post? Share it with your circle, inspire your people.