He reflected on the design of simulators and user interfaces in SimCity, SimEarth, and SimAnt. He demonstrated several of his games, including his current project, Dollhouse.
Here are some important points he's made, at this and other talks. I've elaborated on some of his ideas with my own comments, based on my experiences playing lots of SimCity, talking with Will, studying the source code and porting it to Unix, reworking the user interface, and adding multi player support.
In order for a game to be realizable, all of those different parts must be tractable. There are games that might have a great user interface, be fun to play, easy to understand, but involve processes that are currently impossible to simulate on a computer. There are also games that are possible to simulate, fun to play, easy to understand, but that don't afford a useable interface: Will has designed a great game called "Sim Thunder Storm", but he hasn't been able to think of a user interface that would make any sense.
The actual simulation is much less idealisticly general purpose that I would have thought, epitomizing the Nike "just do it" slogan. In SimCity classic, the representation of the city is low level and distilled down compactly enough that a small home computer can push it around. The city is represented by tiles, indexed by numbers that are literally scattered throughout the code, which is hardly general purpose or modular, but runs fast. It sacrifices expandability and modularity for speed and size, just the right trade-off for the wonderful game that it is.
Some educators have asked Maxis to make SimCity expose more about the actual simulation itself, instead of hiding its inner workings from the user. They want to see how it works and what it depends on, so it is less of a game, and more educational. But what's really going on inside is not as realistic as they would want to believe: because of its nature as a game, and the constraint that it must run on low end home computers, it tries to fool people into thinking it's doing more than it really is, by taking advantage of the knowledge and expectations people already have about how a city is supposed to work. Implication is more efficient than simulation.
People naturally attribute cause and effect relationships to events in SimCity that Will as the programmer knows are not actually related. Perhaps it is more educational for SimCity players to integrate what they already know to fill in the gaps, than letting them in on the secret of how simple and discrete it really is. As an educational game, SimCity stimulates students to learn more about the real world, without revealing the internals of its artificial simulation. The implementation details of SimCity are quite interesting for a programmer or game designer to study, but not your average high school social studies class.
Educators who want to expose the internals of SimCity to students may not realize how brittle and shallow it really is. I don't mean that as criticism of Will, SimCity, or the educators who are seeking open, realistic, general purpose simulators for use in teaching. SimCity does what it was designed to and much more, but it's not that. Their goals are noble, but the software's not there yet. Once kids master SimCity, they could learn Logo, or some high level visual programming language like KidSim, and write their own simulations and games!
Other people wanted to use SimCity for the less noble goal of teaching people what to think, instead of just teaching them to think.
Everyone notices the obvious built-in political bias, whatever that is. But everyone sees it from a different perspective, so nobody agrees what its real political agenda actually is. I don't think it's all that important, since SimCity's political agenda pales in comparison to the political agenda in the eye of the beholder.
Some muckety-muck architecture magazine was interviewing Will Wright about SimCity, and they asked him a question something like "which ontological urban paridigm most influenced your design of the simulator, the Exo-Hamiltonian Pattern Language Movement, or the Intra-Urban Deconstructionist Sub-Culture Hypothesis?" He replied, "I just kind of optimized for game play."
Then there was the oil company who wanted "Sim Refinery", so you could use it to lay out oil tanker ports and petrolium storage and piping systems, because they thought that it would give their employees useful experience in toxic waste disaster management, in the same way SimCity gives kids useful experience in being the mayor of a city. They didn't realize that the real lessons of SimCity are much more subtle than teaching people how to be good mayors. But the oil company hoped they could use it to teach any other lessons on their agenda just by plugging in a new set of graphics, a few rules, and a bunch of disasters.
And there was the X-Terminal vendor who wanted to adapt the simulator in SimCity into a game called "Sim MIS", that they would distribute for free to Managers of Information Systems, whose job it is to decide what hardware to buy! The idea was that the poor overworked MIS would have fun playing this game in which they could build networks with PCs, X-Terminals, and servers (instead of roads with residential, commercial, and industrial buildings), that had disasters like "viruses" infecting the network of PC's, and "upgrades" forcing you to reinstall Windows on every PC, and business charts that would graphically highlight the high maintanence cost of PCs versus X-Terminals. Their idea was to use a fun game to subtly influence people into buying their product, by making them lose if they didn't. Unlike the oil company, they certainly realized the potential to exploit the indirect ways in which a game like SimCity can influence the user's mind, but they had no grip on the concept of subtlety or game design.
The design of the game play has a lot to do with the user's model of the system, and SimCity elegently supports a number of different user models, games, and toys in one program. You can use the terraforming tools and natural features to play with it like a sandbox or landscaping toy, without even starting the city simulation phase of the game. You can even use it as a painting tool, drawing colorful designs and cartoons with land, water, roads and buildings. SimCity comes with several scenarios with different conflicts and goals, and has a menu of disasters you can invoke to destroy your city, or challenge yourself to recover. You can start your own city from scratch, and develop it in any direction you want. A satisfying feature of SimCity 2000 is the ability to put signs in your city, to name roads and buildings and parts of town. How else could you personalize a simulated city?
There was some interesting discussion about using SimCity as a medium for story telling: encouraging people to imagine far beyond the bounds of what the computer is able to simulate. You can build cities to empathise with, and tell stories about them, about their people, culture, buildings, and history. A class of students could label different parts of a city, and each person could tell a story about a different part, that interacted with the stories going on in neighboring parts of the city. Then they could make a web site with the downloadable city, and an image map of the whole city, linking to all the stories on web pages, with screen snapshots of their neighborhoods, and lots of hypertext links between each story. This way each student could colaborate with several others to write a web of interconnected stories, all about the same city!
The time scale slows down as the game progresses, from geological time, to when life appears, to when intelligence appears, to when technology is developed. There was some trouble conveying this to the users. One thing that supported the notion of time scale is how the view controls along the bottom of the global map were ordered in a temporal progression, in the order you'd need to use them, from the continental drift display, to the technology display.
Will showed me Dollhouse several years ago, and it was amazing then, and even more so now. It's not a product yet, but he's been working on solving some very hard problems. He's trying to give the people who walk around the world seemingly rational behavior.
Dollhouse takes the third person view (looking down on your character), instead of the first person view (looking out of the eyes of your characters). You view your character from above, with a 45 degree orthographic view like SimCity 2000 uses to display buildings. Will has found that it works quite well, since you can see yourself as others see you, as well as seeing other people around you. If there are a bunch of people gathering around some interesting person or place, it's easy to tell what's going on, and navigation is simple and direct. It doesn't suffer from the disorienting navigational problems that a first-person view like Doom imposes. Being able to see yourself as others see you seems to make interpersonal interactions involving body gestures much easier.