← Back to all modules
MODULE 7

Algorithmic Harm Simulator

Experience how "objective" models create discriminatory feedback loops. You'll build a creditworthiness algorithm using only "neutral" variables—and watch it recreate redlining.

How this simulation works
You're a lender building a credit risk model. You want to be fair, so you'll avoid obvious discrimination—no race, no criminal justice variables. Just "neutral" predictors. The simulation will show you what happens when these variables are correlated with incarceration history.
Step 1: Select your model variables
Choose which variables to include in your creditworthiness model:
Annual income
Higher income = more ability to repay
Employment history
Stable employment = reliable income stream
Zip code
Neighborhood economic indicators
Credit history length
Longer history = more data points
Bank account ownership
Existing banking relationship
The impossibility of "neutral" algorithms with unjust data
You cannot build a fair algorithm from data generated by an unfair system. Every variable you used—employment, zip code, banking—is corrupted by mass incarceration's legacy. The solution isn't better algorithms; it's transforming the systems that generate the data.
Continue to Module 8: Ethics Documentation Generator →