An economic simulation set in the late 1700s. Settlements trade, produce, grow, and decline based on the decisions of merchants, the availability of goods, and the limits of human organization across distance.
The focus is commerce, not conquest. War exists in this world but it's expensive, destructive, and rarely the winning move. The interesting decisions are about where to invest, who to trust, what routes to run, and when to break a contract.
Most strategy games model economies as a resource treadmill in service of some other goal - build units, research techs, paint the map. The economy is a means to an end, and it's designed to be legible and controllable. You click a building, it produces 3 iron per turn, you spend 10 iron on a swordsman. The numbers are tuned for balance, not for realism, and the result feels like what it is: a board game with a historical skin.
Manifest is trying to do something different. The economy isn't a means to an end - it is the game. And instead of designing a set of rules that produce predictable outcomes, we're designing a set of systems that produce emergent ones. The goal is a world that feels like it has its own momentum. Settlements should prosper or struggle for reasons that make sense when you look at them - not because a formula said so, but because the grain supply dried up when the port got congested, which drove prices up, which made workers demand higher wages, which squeezed the foundry's margins, which meant fewer tools got produced, which made farming less efficient. A cascade of cause and effect, not a number in a spreadsheet ticking down.
The design philosophy is: build the substrate, not the outcomes. Define how people decide what to buy, how merchants set prices, how workers choose between employment and subsistence, how populations grow when fed and decline when not. Then let the simulation run and see what happens. If the results look like a functioning pre-industrial economy - with regional specialization, trade networks, boom-bust cycles, migration toward opportunity - then the substrate is right. If they don't, the answer is to fix the underlying dynamics, not to patch the symptoms with hardcoded corrections.
This means accepting that the simulation will surprise you. It means tolerating messy intermediate states while the feedback loops find their footing. It means resisting the temptation to clamp a price or force a population to stabilize when the real problem is that the labor market isn't clearing properly. It's harder to build this way, but the result is a world you can trust - one where your intuitions about how economies work actually apply, because the simulation is doing roughly what a real economy does, just faster and smaller.
Populations live in settlements. They work, eat, buy things, and move to better opportunities when they can. Merchants own production facilities and ships. They hire workers, source inputs, produce goods, and sell into local markets. Prices emerge from supply and demand clearing against real budgets and real inventories - nothing is pinned or declared, it all falls out of the interactions.
The world runs on a tick cycle: labor clears first, then production runs, then settlements consume and trade, then people are born or die based on how well they ate. Each phase feeds the next. There's no global optimizer - just a set of coupled feedback loops that either find a stable rhythm or don't.
The simulation's character comes from a handful of interlocking cycles. None of them is complicated alone. Together they produce emergent equilibria that are genuinely hard to predict - which is the point.
The reason these loops exist isn't to create "interesting mechanics." It's because real economies work this way. People maintain pantries. Unemployed workers find odd jobs. Populations grow when times are good. Ports connect local markets to the wider world. We're not inventing game systems - we're modeling the dynamics that made pre-industrial economies tick, then seeing what falls out. When we get it right, the simulation produces recognizable economic behavior without being told to. When we get it wrong, the simulation tells us, loudly, by collapsing or oscillating into nonsense. That's a feature: the system is honest about its own failures.
The stock-buffer cycle. People don't buy exactly what they need each day. They try to maintain a buffer of essentials - a few days' worth of food, some cloth, tools. When their stocks are low they bid aggressively. When they're comfortable they ease off. This exists because it's what real households do, and it turns out to be structurally important: it smooths out supply shocks naturally, prevents single bad ticks from cascading into crises, and creates realistic demand waves when many people restock at once after a shortage clears. Without buffers, every supply hiccup is a catastrophe. With them, the economy has inertia - it takes sustained pressure to move it, not just a one-tick blip.
The subsistence floor. Unemployed people don't just vanish from the economy. They can scrape by on subsistence activity - small-scale farming, foraging, odd jobs. This creates a floor under both wages (why work for less than you can feed yourself?) and food supply (some grain appears even without formal production). The floor erodes with crowding - the hundredth person trying to forage the same land gets less than the first. This matters because it gives the economy a natural basement. Wages can't fall to zero because workers have an outside option. Food can't completely disappear because people will produce some for themselves. But the floor is soft, not hard - overwhelm it with enough people and it crumbles, which is historically accurate and produces the right kind of pressure toward formal employment and production investment.
The population-production balance. More people means more labor supply, more consumption demand, and more pressure on food. If production can feed everyone, population grows, which creates more workers, which enables more production - a virtuous cycle. If it can't, population declines, which reduces demand, which lets the remainder stabilize - a harsh but self-correcting one. This is the slowest loop and the one that determines whether a settlement thrives or withers over the long term. It's also the one that connects everything else: labor markets, food prices, production capacity, and subsistence all feed into whether the population is growing or shrinking, and that in turn reshapes all of those markets. Get this loop right and the simulation has a long-term arc. Get it wrong and settlements either explode to infinity or collapse to zero.
The external anchor. Settlements don't exist in a vacuum. Port settlements can trade with the outside world - importing when local prices spike, exporting when there's surplus. This acts as a soft stabilizer: it dampens price extremes without fixing them. The outside world offers finite liquidity at frictional prices, not a bottomless reservoir. Lean on it too hard and it stops helping. The anchor exists because closed economies are brutally unforgiving - a single bad harvest with no outside relief means death spirals. Real economies had trade connections that provided exactly this kind of pressure relief valve, and modeling it explicitly (with real costs and limits) produces much more realistic price behavior than either a fully closed system or a magical price floor.
The key insight across all of these: none of them is a "game mechanic" in the traditional sense. Nobody designed a stock-buffer system to create interesting gameplay. It's there because households maintain pantries, and it turns out that modeling pantries produces better economic dynamics than not modeling them. The subsistence floor isn't a balance lever - it's a description of what unemployed people actually do, and it happens to prevent degenerate edge cases. Each loop earns its place by being true to the underlying reality, and the emergent behavior is interesting because it's grounded, not in spite of it.
The simulation core is written in Rust. There's no frontend yet — the simulation is exercised and observed through its test suite and instrumentation layer.
The project has an unusually strong testing and observability layer. Invariant tests verify that currency is conserved, stocks never go negative, and accounting stays consistent. Convergence tests run scenarios from various starting conditions and check that the economy stabilizes rather than exploding or collapsing. Stress scenarios push the system into uncomfortable regimes to map out where it breaks. This infrastructure matters because when you're building on emergent dynamics rather than designed outcomes, you need to know quickly when something has gone wrong - and "wrong" is often subtle, a slow drift rather than a crash.
The economic core is the foundation. On top of it, the plan is to layer:
Organizations and hierarchies. Right now there are merchants and pops. Eventually there will be a richer graph of organizations - guilds, trading companies, family enterprises - that can own assets, employ people, and form hierarchical or contractual relationships with each other. Control degrades over distance and hierarchy depth. Managing ten outposts isn't impossible, just increasingly unreliable. This is transaction cost economics as a game mechanic - organizations exist because coordinating internally has different costs than coordinating through markets, and those tradeoffs shift with scale and distance.
Contracts and arbitration. Organizations will be able to sign supply agreements, exclusive trade rights, and partnerships. Enforcement comes not from a central authority but from powerful organizations willing to act as arbiters - and punish breach by restricting market access. Burn a well-connected partner and you lose standing with everyone they know. Burn an isolated one and nobody notices. Relationship tiers (favored, neutral, disfavored, excluded) replace fragile numeric reputation scores. The result should feel like the Hanseatic League or Venetian merchant republics - governance that emerges from commercial networks rather than being imposed from above.
Information as a physical thing. In the late 1700s, you don't know what grain costs in a port three weeks away unless someone physically brings you the news. Market snapshots will travel with ships and caravans. Your trading decisions will be limited by what you actually know, not what the simulation knows. Stale information is dangerous - act on last month's prices and you might sail into a glut. This turns information networks into strategic assets without needing a separate espionage system. A ship isn't just cargo capacity - it's a node in your knowledge network, and losing it hurts twice.
Explicit transport. Ships and caravans as first-class entities with capacity, speed, and upkeep. Standing trade routes for efficiency, one-off missions for opportunism. Inland settlements will only access the outside world through merchant logistics, making route capacity and chokepoint control genuinely important. A river junction isn't valuable because the game says it is - it's valuable because all the cargo between two regions physically passes through it.
A richer production graph. More goods, cross-linked supply chains, shared bottlenecks. Lumber needed for both tools and ships. Iron competing for smithy capacity. Enough complexity to force real allocation decisions without becoming an inventory management nightmare.
The guiding constraint across all of this: every new system must work through the existing economic feedback loops, not around them. Organizations change who can buy and sell under what terms. Information changes what decisions are possible. Transport changes delivery cost and delay. Nothing directly overrides prices or conjures goods out of thin air. The accounting stays clean, the feedback loops stay honest, and the emergent behavior stays trustworthy.