TycoonGlobal

Economy

The Closed Faucet Law

Only city retail sales mint new currency. Everything else is a transfer or a burn. Why one faucet beats a dozen, and what it costs to enforce.

Economy7 min readPublished 9 Oct 2026#economy#invariants

There is a specific way that persistent-economy games die, and it is almost always the same way. Not a bug, not a server fire — a slow, compounding leak.

It starts reasonably. A daily login bonus to improve retention. A quest reward so new players have something to spend. A rewarded ad that grants a small cash bundle. Each is defensible on its own. Together they are a dozen uncoordinated money printers running against sinks that nobody is tracking, and six months later a loaf of bread costs nine thousand of whatever the currency is called.

One printer, by construction

TycoonGlobal's second invariant states it plainly: only City NPC Retail Sales mint new M0.

When a simulated city buys a finished good from a player, the cash credited to that player is created by the system. That is the only event in the entire game that increases the money supply. Everything else — spot trades, B2B contracts, loan principal, VC tranches, stock trades, insurance payouts — is a strictly zero-sum transfer between existing accounts, minus whatever permanent sinks apply.

Only seven of the thirty Phase-1 SKUs can trigger it, and all seven are finished consumer goods:

SKUReference priceTier
FRESH_BREAD$16.50T1
BOTTLED_WATER$10.20T2
ROAST_COFFEE$16.50T2
READY_MEAL$23.50T2
SMART_SPEAKER$125.00T3
SMARTPHONE$285.00T3
LAPTOP$490.00T3

Notice what is absent. Wheat is not on this list. Nor is steel, nor copper wire, nor microchips. You cannot mine ore and sell it to the world for new money. You can only sell it to another player — which means the entire upstream economy is funded by somebody, somewhere, successfully putting a finished product in front of a consumer.

What this buys you

The supply chain has to be real. If the only exit is retail, then every intermediate product exists because something downstream needs it. There is no NPC vendor who buys infinite iron ore at a fixed price, so there is no infinite-ore strategy. Mining is profitable precisely when smelters are profitable, which is when construction is profitable, which is when somebody is expanding.

Market crashes mean something. When consumer electronics demand falls in a recession phase (0.75x multiplier), it propagates upstream: fewer laptops means fewer microchips means fewer wafers means less silicon sand. That is not scripted. It falls out of the faucet rule.

Inflation has a single dial. When the money supply grows too fast, there is exactly one place to look. The central bank's algorithmic base rate reads net money-supply growth directly:

r_weekly = clamp(0.035 + 1.5 x (dM_7d / M_total - 0.015), 0.02, 0.12)

Credit gets more expensive automatically when the world is inflating. With a dozen faucets, that feedback loop would be guesswork.

What it costs

Enforcing one faucet is genuinely restrictive, and the restrictions are the interesting part.

No cash rewards, ever. Not for logging in, not for watching an ad, not for winning an event. The reward for winning a Parity Sprint is a collectible emblem, a title, analytics days and event-pass tokens — never a permanent output multiplier and never cash. Every monetisation surface had to be redesigned around this; see why monetisation never mints cash.

New players need a real path to income within minutes. If you cannot hand someone a starting bonus, the opening loop has to actually work. A founder begins with $15,000 and starter plots, and must reach a retail sale quickly enough that the game feels alive. Unlocking a plot slot, building a Tier-1 mill and taking it to level 2 costs $5,500, leaving $9,500 of working capital on day one.

Every leak is a bug, not a balance issue. If a code path credits cash without a corresponding debit, the ledger invariant fails and the test suite goes red. That is a much better failure mode than discovering it six weeks later in a price chart.

Proving it holds

The simulator runs 1,000 agents for 90 simulated days and asserts the money-supply identity at the end:

M_end = M_start + faucet - sinks

Across 2,160 hourly ticks and roughly 2.9 million transactions, drift was exactly $0. It is asserted as a test, which means a future change that quietly mints a cent cannot ship.

The corollary matters as much as the rule: a faucet this narrow is easy to starve. Early simulator runs measured 92% of the money supply disappearing over 90 days, because wages were being charged on idle plants and retail demand was modelled far too small. Both were fixed, and retention recovered to 1.037 of starting supply. The faucet and sink equilibrium tells that story in full.