From AIGPG Wiki
Revision as of 21:34, 17 July 2011 by Steve Rabin (Talk | contribs)
Jump to: navigation, search

Developer: Maxis
Publisher: Brøderbund
Year: 1989
Platforms: Amiga, Atari ST, C64, Macintosh, PC
Genre: Simulation
AI Era: Simple Hard-coded Rules

SimCity is a city-building simulation designed by Will Wright. SimCity was inspired by Will Wright's work on Raid on Bungling Bay that allowed him to build maps for the game as well as a rather complex city/economic simulation (that largely went unnoticed and unappreciated by players).


Description of AI Behavior

As a city-simulator, the game models the evolution of a city (growth and decay) based on design input from the player. The player is able to control zoning (residential, commercial, industrial), place individual buildings (e.g. airports, stadiums, power plants), layout the power grid, route transportation (roads and rail), and control the budget (taxes and expenditures to fire, police, and transportation).

The AI simulation computes how the population reacts in the face of hundreds of inputs, primarily controlled or initiated by the player. The city's population can react with changes to buildings, changes in traffic, and changes in population size. Although the player controls the zoning in a given square, the buildings within that zone change based on the simulation by increasing or decreasing their size and by changing their class (for example slums vs. high-class). Roads can become more or less congested and the population reacts to taxes and other conditions by leaving or moving to the city.


One method of feedback from the simulation is through statistics (public opinion and city statistics). Public opinion includes how well the population thinks the player is doing ("Is the mayor doing a good job?") and "What are the worst problems?" by percentage (for example, 21% taxes, 12% housing costs, 7% pollution, and 2% traffic). City statistics include population, net migration last month, assessed value in dollars, and city classification (such as "TOWN"). The game also gives the city a score from 0 to 1000, along with the score's annual rate of change.

Notable Behaviors

The integrity of the simulation was quite remarkable and it was endless fun to manipulate the city and watch how the city would react.



The entire simulation itself is quite remarkable and was inspired in part by System Dynamics. For example, the wealth of a residential zone is influenced by nearby terrain features (water, parks, trees), low pollution, low crime, and transportation distance to jobs. Since a zone is influenced in part by the statistics of neighboring tiles, it can be said that the SimCity simulation shares characteristics with Cellular Automata. In-depth simulation details are available at the Micropolis open source project, which was built with the original SimCity source code.

Heat Maps

Heat Map: Crime in SimCity (Amiga)
Heat Map: Crime (Amiga version)
The simulation generates and displays graphical statistical data in the form of Heat Maps (with some maps sharing qualities with Influence Maps). The game uses these maps to affect the simulation as well as to give visual feedback to the player. On an interactive map of the city, the player can choose from among several different views, such as police department coverage, fire department coverage, crime hot spots, pollution hot spots, traffic congestion, population density, and wealth. The other views that would not be considered heat maps include a view of which buildings have power ("Power Grid"), where roads and rails lie, and an abstract view of the layout of the city ("City Form").

Procedural Terrain Generation

At the start of the game, the player selects one of 1000 possible maps by entering a number from 0 to 999. While this appears to be a simple index into a particular map, in reality the map is procedurally generated from the seed number (such that a given seed would always produce the same map, presumably by using the number as a seed for a pseudo random number generator). This allowed for a large variety of landforms without having to store each one explicitly on disk. Each landform contained a combination of water and land tiles, with trees populated on some land tiles. The procedural generation created large areas of land, islands, lakes, and forests (groups of trees). Of particular concern to some players was the amount of land on which to build on (which limited the ultimate size of the city). This caused players to prefer particular maps that had the fewest number of water tiles.


What Worked

The simulation was extremely robust and didn't contain any critical flaws that would allow for cheating or quick advancement.

Player Feedback

Critical to manipulating a simulation is understanding the relationships and inputs involved in the simulation. The game did an excellent job of allowing the player to explore the current state of the simulation through visual statistical views and numeral statistics. The player always felt like they could understand why parts of their city were progressing or declining. Additionally, the game would offer alerts and direct feedback when there were critical problems within the city, such as high crime or traffic congestion.

What Didn't Work

As with any simulation, there are certain peculiarities that may not be completely realistic or reasonable. One such example is the use of roads versus rails. As a city grew, the roads would inevitably become congested, no matter how many roads were built. This would lead players to build rails, which were incapable of becoming congested. This lopsided simulation appeared to be overly biased towards rails.


The game included 8 scenarios that could each be won. Unfortunately, particular scenarios could be manipulated because the conditions for success were too simplistic. For example, the Bern scenario required the player to solve the traffic congestion problem, but the scenario could be completed by destroying the city (which would fix the traffic congestion). In rebuilding scenarios, such as San Francisco, lowering the tax rate to zero toward the end of the scenario would spur growth and lead to an easy victory.

Lessons Learned


Reception by Public

SimCity is widely recognized as one of the most successful simulation games of all time. It appeared on 10 different platforms originally and has been rereleased as SimCity Classic on several more, including the web.


SimCity on Wikipedia.

Micropolis open source project (based on the original SimCity code).

External Links

Play SimCity Classic on the web for free.

Personal tools