Comment by larsiusprime

Simulations and models, regardless of whether they're designed for fun or for realism, can still teach you plenty of things about the real world; sometimes not from the simulation itself but explicitly by its omissions. The fact that SimCity famously removed all the parking lots tells you something interesting right away. The fact that City Skylines bites the bullet of traffic and renders it in fairly sophisticated detail -- to the degree that the game often devolves into "Traffic Management Simulator" instead of "City Builder Simulator" -- is another such interesting lesson.

But in terms of "how directly transferable are lessons learned directly from the model" is always going to depend on the model's assumptions.

Fun fact -- have you ever heard of the "Monocentric city model?" It comes up in a lot of econ papers, and a lot of urban economics, and therefore a lot of real world policy is based on it.

The "Monocentric city model" is about as simplistic as it gets -- it models the city as a LINE. IE, you have a bunch of agents, and they live some distance from the city, and they have different travel/commuting costs proportional to their distance from the city center; let's run that and see what kinds of consequences fall out of it.

Turns out it's actually a pretty good model and generates outcomes that are pretty analogous to what we see in the real world! Is it perfectly representative of the real world? Of course, not, it models the city as a friggin line. But if we're talking about sheer complexity alone, SimCity is already orders of magnitude more complex than this famous economic model. But that in and of itself doesn't mean that a successful SimCity city would make for a successful real world city, because it's not just about sheer complexity, it's about the appropriateness of the assumptions.