I have played this genre for years — Game Dev Tycoon, Startup Company, Big Ambitions, the Devices and Laptop Tycoon line — and then spent eighteen months building one. The thing I did not understand as a player is that the economy is not where you put the numbers. It is where you put the rules the numbers are not allowed to break.
What follows are the actual laws from my simulation. Each one is a test with a name that reads as a sentence, which is deliberate: an economic law written as English survives tuning, and a magic number in a config file does not.
Write the laws first, the numbers second
Every economy I have built badly failed the same way: I tuned a number, then another, then forty more, and six weeks later the game had a different character and no single commit was responsible. The numbers were all defensible. The directions had quietly reversed.
A law is a statement about direction, not magnitude. Here are mine, as they are named in the test suite:
| The law, as its test is named | What it protects |
|---|---|
PricePerTokenOnlyEverFalls | The one-way ratchet the whole economy hangs from. |
DemandGrowsFastThenFlattens | A boom that ends, so early success is not permanent success. |
TheFrontierKeepsClimbingAndNeverWaitsForThePlayer | A world that moves whether or not you do. |
CuttingPriceBuysShareAndNothingElseDoes | Exactly one lever on the most important number. |
ShareIsSplitAgainstRealRivalsAndNeverHandedOver | No free customers, ever. |
AModelNobodyReplacesQuietlyStopsBeingChosen | Nothing you shipped stays good. |
ScarcityPeaksInTheShortageAndEasesAfterwards | Shocks that arrive and then pass. |
ASubscriptionLeftAloneDriftsIntoOverchargingAndThatIsAStory | Standing still has to cost something. |
TheMarketNeverBreaksOverFourteenYears | The edges, where compounding numbers actually fail. |
Read that column again as a design document, because that is what it is. It took me a long time to notice that my clearest statement of what the game is was sitting in a test file.
The best mechanic I have is a number the player never touches
If you take one thing from this article, take this one. It is the cheapest interesting mechanic I have ever built and it is made entirely of a law I had already written.
The setup: the market price of the thing my player sells falls over time — that is law one, and it is true of the real industry the game is about. The player chooses a pricing model early, and a subscription is the obvious choice because it is stable. They set it, it works, they close the screen.
A year later their fee has not moved and the market price has halved. They are now charging twice the market rate, without having done anything at all. The test says it in arithmetic:
And then the assertion that explains why this is a mechanic rather than a simulation detail. The message is what you would read if it ever failed:
The general rule hiding in there
A screen is only interesting if a decision made on it can become wrong on its own. Not through a timer, not through an event that fires at you, but because the world moved and your choice stayed where you left it.
This reframes the usual complaint that a management game “gets boring after an hour”. It rarely needs more content. It needs decay: a reason for the player to revisit a decision they already consider finished.
The counterpart matters just as much. Metered pricing tracks the market down, so it never drifts — and the test says why that is the only reason to choose it:
That is a genuine trade rather than a better option: a subscription pays more while you are paying attention and punishes you when you stop; metered pays less and never bites. A player can explain both in one sentence, which is the bar every economic choice in this genre should clear.
One lever moves share, and only one
Market share is the number players watch, which makes it the number most likely to become incomprehensible. If four systems nudge it, nobody can trace why it moved, and a management game whose central number moves for untraceable reasons is a slot machine with spreadsheets.
So mine has exactly one lever, named as plainly as I could manage:
The companion law is the one people forget, and it is the difference between an economy and
a scoreboard: ShareIsSplitAgainstRealRivalsAndNeverHandedOver. There is no state
in which the market simply gives the player customers. Even a dominant company faces an
assertion that rivals keep some of the market, because an economy where competitors stop
competing is an economy that has quietly ended, and the player will feel that long before
they can name it.
The world must not wait for the player
Most tycoon games are generous here and it costs them their tension. If the competition scales to your progress, nothing you do is urgent, because the gap is constant by construction.
Mine states the opposite as a law:
TheFrontierKeepsClimbingAndNeverWaitsForThePlayer. The best thing in the world
improves on its own schedule. Fall behind for two in-game years and you are behind, not
rubber-banded back.
That sounds harsh and needs a counterweight, which is the balance band from the article on playability tests: an ordinary player must still finish four years within reach of the frontier. The two tests disagree on purpose. One says the world does not wait; the other says the game stays winnable anyway. Balance is the distance between them, and writing both down is what stops you from accidentally softening the first while trying to satisfy the second.
Two laws that make a long game feel alive
DemandGrowsFastThenFlattens gives the economy a boom with an end, so
a company that rode the early wave cannot coast on it. Growth that never flattens is a
game that rewards starting well and then stops asking anything.
ScarcityPeaksInTheShortageAndEasesAfterwards makes shocks temporary.
A shortage that never resolves is just a permanent price change wearing a costume, and
players read it as one.
Nothing you shipped stays good
AModelNobodyReplacesQuietlyStopsBeingChosen is the law I would add first to
somebody else's tycoon game, because almost every one of them is missing it.
The common design has you build a product, and that product keeps earning forever. It makes the first hours satisfying and the later hours hollow, because the optimal strategy becomes accumulation: never retire anything, let the pile grow. Obsolescence is what turns a pile into a portfolio you have to manage.
Note the word quietly in the test name. The product is not deleted, it does not expire, no notification fires. Customers simply stop choosing it, because the market moved and it did not. The player discovers it by looking — which is the behaviour you want to reward in this genre.
Proving the economy does not break in year fourteen
Compounding numbers fail at the edges. An economy that behaves beautifully for four years can produce a negative price, an infinite demand or a dead market at year nine, and your players include someone who will get there.
So the market is stepped far past the campaign and checked for sanity the whole way:
| Test | What it refuses to allow |
|---|---|
TheMarketNeverBreaksOverFourteenYears | Any value going impossible at the far edge of the curve. |
SomebodyIsActuallyServingTheMarketThroughout | A market that technically exists while nobody is in it. |
EveryModelTypeIsSomethingSomebodyEventuallyBuilds | Content that is reachable on paper and chosen by nobody. |
TheServedMarketChangesShapeOverAWholeGame | A market that is the same shape in year ten as in year one. |
That second one is subtler than it looks. “Somebody is serving the market” is not about the player at all — it asserts the simulated world remains populated. An economy where the rivals all quietly die is technically stable and completely dead, and it is a very easy failure to ship, because nothing errors.
The third is my favourite piece of housekeeping in the whole suite. It walks the content and asserts that every product type is something an actor in the world eventually builds. Content that exists but is never chosen is the economic version of the unreachable feature: finished, correct and invisible.
How much realism, and where to stop
My game is about an industry with real published economics, so the temptation to model everything was strong, and giving in to it was a mistake I made twice.
The rule I settled on: the directions must be honest, the magnitudes must be legible. Price falling over time is true of the real thing and the game would feel wrong without it. The exact shape of the fall exists to make a decision readable, not to match a chart.
The test to apply is whether a player can explain what happened to them in one sentence. “I kept my subscription fee while the market halved, so I was charging double and customers left” is a sentence. If an honest model produces an outcome nobody can narrate, the model is right and the game is wrong, and the game is the thing you are making.
When none of this applies
If your economy is a shop with a handful of upgrades, or the money exists only to pace content you have authored by hand, this is overkill. These laws earn their keep when the economy is the game — when the player spends hours making numerical decisions and expects them to add up.
The honest signal that you need them: you have tuned the same constant more than three times and cannot say what the game is supposed to feel like at hour five.
The order I would write them in again
- One ratchet. Pick the single number that only moves one way and build the rest around it. Mine is the unit price falling.
- One lever on the headline number. Whatever the player watches most, exactly one thing moves it, and that thing is named in a test.
- One decay. Something reasonable must become unreasonable while the player is not looking. This is the mechanic, not the maths.
- No free customers. Share is always contested, including when the player is winning.
- A soak run far past the campaign. Fourteen years in my case; whatever doubles your longest realistic playthrough.
Questions people ask about this
How do you design the economy of a tycoon game?
Write the laws before the numbers. The economy is a small set of statements about which way things may move: the unit price only ever falls, demand grows fast and then flattens, the frontier climbs whether or not the player keeps up. Each is a sentence you can enforce as a test, so it cannot quietly reverse during months of tuning. Numbers are what you tune afterwards; laws are what stop the tuning from producing a different game.
Why does my tycoon game get boring after the first hour?
Usually because standing still stopped costing anything. If a decision made in hour one is still correct in hour five, the screen where it was made is a screen nobody opens again — and a management game is mostly screens. The fix is rarely more content; it is decay, so a reasonable choice becomes unreasonable through the world moving rather than through a timer firing.
Should market share be something the player buys with price?
One lever should move it and the rest should not, or share becomes a number that drifts for untraceable reasons. In mine, cutting price buys share and nothing else does, and that is the name of the test. Just as important: share is split against actual rivals and never handed over, because an economy where competitors stop competing has quietly ended.
How do you stop a simulated economy from breaking late in a long game?
Run it far past where any player will get and assert it is still sane. Mine steps the market over fourteen years and also asserts that somebody is always serving that market. Compounding numbers fail at the edges, and a four-year campaign tested only to four years will eventually meet the player who arrives at year six.
How much economic realism does a tycoon game need?
Enough that the directions are honest, and no more. Magnitudes exist to make decisions legible. The usable test is whether a player can explain what happened to them in one sentence; if an honest model produces an outcome nobody can narrate, the model is right and the game is wrong.
MarketModelTests.cs, BusinessPricingTests.cs and
MarketDriftTests.cs, linked from the sources.