In a talk that a producer friend described as “Along with Rich Vogel’s talk, the two best hours of production talks ever at GDC,” Rod Fergusson, executive producer of Epic Games spoke about his methodology for producing Gears of War 2.
His most salient point was that producers should work with the team to fix a release date early, once it’s possible to have a vision for the project scope, and then stick to that date, barring extreme circumstances.
In the traditional “iron triangle” of important elements of production, there’s schedule, scope, and resources. Fergusson paraphrased Jim McCarthy, author of “The Dynamics of Software Development” in saying, “In software there are thousands of variable. Every project has risks and issues. But what if you could take just one of those variables, and fix it? Just lock it. Hold it to a certain value, and let that help you gauge your other problems.”
For Fergusson, that variable is the schedule, but of course you need a belief that the ship date is both realistic and unchangeable. “One of the great things is it creates a clear goal for the team,” he says. It proves there’s an actual light at the end of the tunnel. “They say you need constraints to have creativity and prioritization.”
To supplement this, at the beginning of Gears 2’s development, 16 project areas were asked what they thought were the five most important things to push forward in the next game, compared to the first, and they used that list to define the actual feature scope and schedule.
These main points then should be grouped to form the “pillars” of the game, according to Fergusson. This is essentially something like “engaging co-op experience,” which is supported by things like individual difficulty levels in co-op. Pillars are good talking points for the press, and also help determine what’s going on the back of the box. “If you can’t talk to the press about ‘here are the four most important things about the new game,’ you’re kind of lost,” he says.
Pillars also help when it comes time to cut or add features. “The question is does it support a pillar? If someone comes and says I really like open-world ideas, we’re going to say well that’s great, but that’s not one of our pillars,” he says. But if there’s something to add that really will support a pillar, he’s all for it.
Cut early, cut often is a mantra for many, and Fergusson believes this is necessary when managing scope. At the same, time, don’t completely throw it away. He gave an anecdote about an artist spending 45 days modeling an “Uber Reaver” for Gears 1, but they realized they didn’t have the time or resources to animate something that would only be in the game once. However, they saved the model for the second game, and the work didn’t go to waste.
Fergusson reminded attendees to always keep test burden in mind, as well. What a programmer may call a “one-line change” isn’t the only cost. “It’s also it’s four weeks of five testers doing this test plan,” reminds Fergusson.
“I am a believer that if you’re going to make a great game, and there is that caveat, I believe that crunch is necessary,” Fergusson says. “I believe it’s important because it means your ambition is greater than what you scheduled out. Going in with that idea that crunch is necessary means you can plan for it. It shouldn’t be a surprise. Crunch should be driven by the ambition of the team, and not the inaccuracy of the schedule.”
But he cautions that crunch should be managed by milestones, or some other regular method. “It’s a marathon, not a sprint,” he says, though realistically “It’s a marathon for a really long time, then at the end it is a sprint.”
He also believes in empathetic crunches. “If we’re going to crunch early for something, we made them teamwide. Everything can benefit from getting more done. If the artists were on schedule, then they crunched and they got ahead.” This sometimes led to more being added to the game, or simply more human resources later on. At the end of the project though, they keep crunches as small as possible, because having fewer hands on the project later helps polish.
He cautions though, that people have limits. “Working later than 2 am is a net loss. The productivity of the person who’s doing that to themselves ultimately ends us costing them at the end of that week,” he says. Epic has put a “go home law” in the company handbook as a result.
“Every crunch is different for every team,” he says. “If you’re not doing it because of mistakes in the schedule, but through planning, it’s much easier to go to your team and ask them how they want to crunch.” If they have that light at the end of the tunnel, in terms of a fixed launch date, crunch becomes much more manageable.