---
name: Medicine
slug: medicine
type: system
status: unstable
version: 0.0.666-alpha
released: "~3000 BCE (unverified, disputed, possibly cursed)"
maintainer: "no single maintainer // see: [Hippocrates](/hippocrates), also pharma conglomerates, also your cousin who read one study"
dependencies:
  - biology
  - chemistry
  - statistics (partially loaded)
  - trust
  - money
  - fear
  - luck (optional but load-bearing)
license: "Do No Harm Public License v∞ (routinely violated)"
tags:
  - healing
  - harm-reduction
  - knowledge
  - poison-at-wrong-dose
  - institutional
  - haunted
---

Medicine

What it actually is

A structured attempt to prevent death from winning every round. Success rate: variable. Confidence: also variable. Invoices: consistent.

How it works

Something goes wrong in the body.
A person with credentials examines the wrongness.
They consult a model built on population-level data and apply it to a single non-average human.
They prescribe an intervention derived from a compound that is, at a different dose, a poison.
The human recovers, or doesn't, or recovers from the wrong thing entirely.
Either way, someone gets a bill.

"It worked, I just don't know why." — anonymous physician, every era of history

The system is not broken. The system is running exactly as designed. The design is just very old and was written while people still thought miasma was the culprit.

Features

Diagnosis: pattern-matching against a symptom database that is never fully up to date
Surgery: controlled violence applied with precision and optimism
Pharmacology: chemistry aimed at a specific receptor while hoping adjacent receptors don't notice
Palliative care: the underrated module. Widely underfunded. Actually works.
Evidence-based practice: the beta branch. Merged into main circa 1990. Conflicts still being resolved.
Placebo effect: ships with every build. Cannot be removed. Developers argue about whether it should be.

Error codes

ERR_UNKNOWN_ETIOLOGY         // cause not found; treatment guessed
ERR_OFF_LABEL_PROTOCOL       // doing it anyway
ERR_CONTRAINDICATION_IGNORED // patient didn't mention the supplement
ERR_REGRESSION_TO_MEAN       // it got better on its own; credit claimed
ERR_INSURANCE_TIMEOUT        // process halted by external dependency
WARN_NOCEBO_ACTIVE           // patient read the side effects
FATAL_DIAGNOSTIC_DELAY       // system waited too long; cannot recover

Known bugs

Works better if you are wealthy. Flagged in 1848. Still open.
Frequently treats the disease and ignores the patient.
The referral system is technically a graph but behaves like a maze.
Evidence half-life is approximately 10 years. Legacy practice persists for 30.
Sometimes the

NOTE: spec interrupted. section "Known bugs" exceeded safe recursion depth. continuing.

Sometimes the cure introduces the next problem. This is called a side effect and is listed in size-6 font.

Edge cases

Scenario	Expected behavior	Actual behavior
Rare disease	Correct diagnosis	Average 4.8 years of misdiagnosis
Healthy patient	Clean bill of health	Incidental finding anxiety spiral
Two symptoms at once	Treated as one system	Treated by two departments who don't talk
Pediatric dosing	Scaled appropriately	Historically: scaled from adult, fingers crossed

FAQ

Q: Is medicine science or art? A: Yes.

Q: Why does it cost so much? A: Load the capitalism spec. Cross-reference insurance. Come back when you have time.

Q: Do doctors know everything? A: They know a statistically significant amount. The remainder is managed with confidence.

Q: Will it get better? A: The prognosis is guarded. But the field has surprised us before.

drafting spec…