In 1994, Paul Starr watched his children play SimCity and started worrying about who had decided what happened when they raised taxes.
That question became a long essay about simulation, public policy, education, and black boxes.
The Princeton-hosted page preserves Starr’s article from The American Prospect, Spring 1994.
It arrives at an unusually good moment in computer history.
SimCity had already sold millions of copies. SimEarth, SimLife, SimFarm, A-Train, and SimHealth were pushing simulation into homes. Multimedia PCs were becoming powerful enough to make complicated models feel like entertainment rather than spreadsheets.
Starr was fascinated by that—and suspicious of it.
What assumptions are hiding in the toy?
The essay traces simulation from military and planning exercises through academic modeling and into commercial games.
Starr interviews Will Wright and notes that SimCity’s model deliberately exaggerates effects because feedback has to be visible enough to make the game playable. Wright describes the game as a caricature rather than a prediction system.
That distinction becomes the essay’s central concern.
A simulation can teach players to think about systems, tradeoffs, feedback loops, budgets, land use, traffic, utilities, and unintended consequences. But the model also makes assumptions about how those systems work.
If the assumptions are hidden, the player can start confusing “the way the game works” with the way reality works.
Starr connects that problem to government policy models, where forecasts can carry enormous authority even when most participants cannot inspect the machinery underneath them.
Thirty-plus years later, this reads less like an old game review than an early warning about algorithmic authority.
CacheRat’s 1,967 Ancient Web Domains research list includes surviving academic pages because old HTML sometimes preserves arguments that have become newly relevant.
The graphics are not the interesting artifact here.
The question is.
Who wrote the model?