Maxis didn’t want to make professional simulation games

Wednesday, June 3rd, 2020

SimCity wasn’t meant to be taken seriously:

The game was inspired by research on real-world urban planning concepts, and although it was created as a way for players to experiment running a city, the goal was to be fun rather than accurate. “I realized early on, because of chaos theory and a lot of other things,” said designer Will Wright, “that it’s kind of hopeless to approach simulations like that, as predictive endeavors. But we’ve kind of caricatured our systems. SimCity was always meant to be a caricature of the way a city works, not a realistic model of the way a city works.”

“I think if we tried to make it realistic, we would be doing something that we wouldn’t want to do,” Wright said in an interview in 1999. But that didn’t stop companies from believing Maxis could design realistic simulations. Will Wright didn’t believe that was even possible. “Many people come to us and say, ‘You should do the professional version,’” he continued. “That really scares me because I know how pathetic the simulations are, really, compared to reality. The last thing I want people to come away with is that we’re on the verge of being able to simulate the way that a city really develops, because we’re not.”

Maxis didn’t want to make professional simulation games. But for two brief, strange years, they did.

From 1992 to 1994, a division called Maxis Business Simulations was responsible for making serious professional simulations that looked and played like Maxis games. After Maxis cut the division loose, the company continued to operate independently, taking the simulation game genre in their own direction. Their games found their way into in corporate training rooms and even went as far as the White House.

Almost nothing they developed was ever released to the public.


For Wright, games were a way of helping people create “mental models” for understanding parts of the world. The team at Maxis would research a topic like urban dynamics — or something like ant colony behavior, in the case of another game they made called SimAnt — and create a game where players could experiment with those ideas. The goal wasn’t to teach anything directly, but rather to help the player get the model of SimCity in their head, so that playing this game could help them understand how the different systems within a city interact.

For many people though, that nuance was lost, and instead they treated it like Maxis could build accurate simulations of the real world. And they wouldn’t stop asking about it. “In the first couple months after SimCity appeared,” Wright told Wired, “we were approached by a number of companies saying, ‘Hey that’s great! If you can do a city like that, we want you to do SimPizzaHut, or SimWhatever.’ We thought these things were so weird that we said no, but they kept coming in.”

“So at some point, as we got big enough, we decided to give it a go.”

John Hiles knew about SimCity. He also believed in the power of building mental models, and he saw something in SimCity that was missing from the simulation modeling work happening at Delta Logic: it was fun. It had an intuitive interface and friendly graphics. That was the missing ingredient. Hiles believed that if they teamed up — Maxis’s style with Delta Logic’s systems — they could create simulations that were fun and powerful. Maxis had been looking for new partners for software development, so Hiles used that as an opportunity to get in their orbit. He approached Jeff Braun, and in 1991, his company became a contractor for Maxis.


As part of the company’s restructuring in the wake of SimCity, in the summer of 1992, Maxis accepted a $10 million investment from Warburg Pincus Ventures, who received a 30% stake in the company and a seat at the board. According to Braun, Warburg Pincus wanted Maxis to start doing business simulation games more seriously.

With their new directive, Maxis decided to jump in all the way. That July, they purchased Delta Logic, turning them into a new division of the company — Maxis Business Simulations. John Hiles was named VP and general manager.

Their first project? Chevron wanted them to make a game about an oil refinery.

Oil refineries are really, really complicated. That’s why Chevron wanted Maxis to make them a game like SimCity, to teach the employees at their oil refinery in Richmond, California how it all worked.

To be clear, they didn’t want a game that was supposed to accurately train people how to run an oil refinery or replace an education in chemical engineering. That would’ve been incredibly dangerous. What they wanted instead was something that showed you how the dynamics of the refinery worked, how all the different pieces invisibly fit together, like SimCity did for cities.

The operators at the refinery sometimes had trouble getting a big picture for what was happening at the plant beyond their particular area of focus. “The whole goal if this was to teach operators that they are part of a bigger system,” Skidmore said. “Their concern at the time was that operators tended to be very focused on their one plant, and their one thing they do, and so [they] weren’t keeping in mind that what they do affected other parts of the plant. So they wanted a training tool that allowed operators to manipulate inputs and outputs of the various pieces of the refinery process to see how they impact.”

The non-technical staff at the Richmond refinery needed to know how it worked too. The people in human resources and accounting weren’t chemical engineers, but it would help their work to see how the different areas of the plant were networked together, how one department affected another department.

Chevron paid Maxis $75,000 for a prototype of a refinery simulator. The project began even before Maxis bought Delta Logic, back when they were still just contractors.

How do you get started on a project like this? They did it the same way Maxis developed their own games: they did research.

John Hiles and the Bruces took a visit to the Chevron Richmond Refinery, where they met with a specialist who took them on a tour of the plant and explained how it worked. It was a collaborative relationship with Chevron throughout the development process; Chevron sent them the raw formulas they used at the refinery, and as Maxis Business Simulations turned that into a game, Chevron would double-check their work.


John Hiles said that most of the trainers at Chevron wanted to use it as a conventional training tool, “but some of the more astute teachers said, ‘Let’s just get you started here by seeing if you can wreck the oil refinery, if you can abuse the inputs and the settings and essentially get fired,’” he remembered.

That was a legitimate way to learn how a refinery worked: if you start breaking the refinery, you can see how ruining one part of the plant will affect the other parts of the plant. “The tool — the game — was agnostic,” Hiles explained, correcting himself. “It would work for someone trying to ruin an oil refinery just as well as somebody trying to run it efficiently.”

SimRefinery was finished in fall 1992, earlier than the 1993 date that’s usually reported online. The trademark registration for SimRefinery suggests that the game was officially handed over to Chevron on Monday, October 26, 1992. (It’s unusual to have a specific release date for a corporate training product, but that’s a result of Maxis trademarking the SimRefinery name almost a year after it was completed.)

Chevron liked it. They started testing the game with their staff in September, and Chevron reported that communication from marketing and finance staff “improved dramatically.” Speaking to The Plain Dealer in Cleveland, Chevron training specialist Susan Gustin praised the game’s effectiveness. “Just dumping information on people isn’t effective,” she said. “People only remember what they use.” She told Computerworld, “Some of these relationships aren’t at all obvious until you play the game a bit.”

It seems to have even won over one of its critics, Will Wright. “He was initially skeptical,” Skidmore said. “I think when we eventually finished SimRefinery, I think he approved of it.”


Whatever the long-term interest in SimRefinery, it wasn’t adopted at Chevron out of the gate, and that was the start of a pattern for the games by Maxis Business Simulations — a skepticism towards the idea that a simulation game could teach you something. Or should teach you something.


  1. Harry Jones says:

    What’s the point of a mental model if you don’t at least hope to arrive at an accurate simulation at some point?

    If you’re not trying to get a grip on the real world, then all thought is mental masturbation.

  2. Gavin Longmuir says:

    Harry J. “accurate simulation”

    Can there ever be such a thing as an “accurate simulation”, i.e. one that reliably predicts the future?

    The more realistic approach is contained in George Box’s famous aphorism: “All models are wrong. Some models are useful”.

    I once read an account of mountaineers preparing for an assault on a difficult Himalayan peak. They built a model of the climb — rate of climb on different stages, Sherpas carrying oxygen and supplies, alternate routes, etc.

    The model could not be used for prediction, because there would be too many unknowns in the actual climb — changing weather conditions, falls, sickness in team members, etc.

    However, the model was very useful for making the climbers aware of interactions — the chances of reaching the summit were very dependent on what happened earlier at lower altitudes in the supply chain. The climbers found that this awareness helped them make better decisions during the climb, as real-world exigencies forced departures from the original plan.

    Perhaps a simulation does not have to be accurate to be useful?

Leave a Reply