2026-11-15
Knowledge does not save you
The single design rule that separates this simulator from a self-improvement game, and the code-level constraint that enforces it.
本文为英文。
In most life games, education is a health upgrade. Study hard, get a degree, and the game quietly lowers your death probability, because educated people live longer in the real world.
That correlation is real. It is also confounded almost to the point of uselessness, and modelling it directly produces a lie with a pleasant shape.
The confound
Educated people do live longer. But education is not randomly assigned. It correlates with household income at birth, with the country you were born in, with whether your parents could afford to keep you in school instead of sending you to work, and with a hundred unmeasured things that also affect health.
When you control for those, the effect of schooling on mortality shrinks. It does not vanish — there is a genuine causal component, through health literacy, through occupational choice, through the ability to navigate a clinic — but it is nowhere near the raw gap.
More importantly for a game: education’s real mechanism is not biological. It is access. Schooling gets you a job that pays enough to live somewhere with clean water. It gets you a diagnosis early enough to matter. It gets you out of a manual occupation that destroys a spine by fifty.
Modelling access instead of biology
So the engine does it the long way round. Knowledge affects which choice nodes are available, and how much money you earn, and therefore what your household can afford. It does not touch the hazard function.
The consequence is a specific, testable behaviour: a person with knowledge 95 and health 20 dies at exactly the same rate as a person with knowledge 20 and health 20. Not approximately. Exactly. The death function does not receive the knowledge value.
That is enforced in the type signature, not in a comment. There is no path by which a high-education character gets a mortality discount.
What this feels like to play
It feels wrong at first, and that is the point.
A player who spends every year on study, gets the degree, gets the good job, and then dies at forty-four of a stroke will look for the mistake they made. There isn’t one. There was a roll, and the roll went badly.
This is the single most common complaint about the original game and it is also the reason people remember it. Every other simulator has trained players to believe that a bad ending means bad play. This one refuses to confirm that belief.
The other side of the trade
It would be easy to overcorrect and make choices meaningless. The engine does not do that either.
Knowledge determines your income ceiling, and income determines your assets, and assets determine whether you can afford the choice nodes that cost money — the move, the clinic, the year off. Health determines whether you survive the year, and health is the one stat that responds immediately to what you do: the health focus adds roughly a point of health per year, and illness events subtract far more.
So the choices are real. They just operate on the inputs, not on the output.
The honest framing
If you are educated, healthy, and live in a rich country, your expected lifespan is long. Every one of those three is partly a result of choices and mostly a result of where the choices were available.
A game that lets you buy your way out of the third one is a fantasy about merit. A game that shows you the third one and then hands you the dice anyway is closer to the truth.
That is what this is trying to be. Not encouraging, not cruel. Just unwilling to lie about which lever does what.