Project Horseshoe doesn't have keynotes. The annual invite-only think tank for game development's best and brightest has a focus on brainstorming and breakout sessions, but there are still a handful of talks on the first morning.

In his introductory remarks, host George Sanger said, "I've asked Mike to let it all hang out - if you can't do it here, it can't be done." And in the next 25 minutes Mike Verdu would give an unvarnished look at the successful production of The Battle for Middle-earth and Command & Conquer 3 games from EA LA and their team structures.

"It's a topic I think of as meta-design," began Verdu, who was appointed new General Manager of Electronic Art's Los Angeles studio four days before Project Horseshoe. "It really comes down to the interaction of large numbers of people and the creative process."

"Team size affects creative output," he said. "You can do things that are in a traditional, hierarchical way, and get a process that is somewhat inefficient and might make people unhappy. Or you can actually unlock the creative firepower of a big team, and the result is spectacular."

Verdu started in the game industry in 1989, partnering with Bob Bates to start Legend. Their peak team size was 10, half that for most projects. Ten years later, Verdu found himself working on a 3-D fantasy action game called The Wheel of Time, powered by the Unreal engine. That game had a 25-person team.


When Verdu joined EA in 2002, the teams started getting bigger, between 40 and 50 people. Then there was a big surge in team size to over 80 people. That's the point where it's hard to know the names of each team member and what he's doing.

Verdu told the audience that between 1989 and 2004, he was used to working with a traditional development structure. There were leads and three groups of people: the artists, the designers and the engineers. This was fine when the teams were small.

When Verdu speaks of small teams, he uses examples of improvisational jazz, the Power Rangers or flight-deck crews on aircraft carriers. Each small team communicates well and completes very complex tasks.

"Communication and investment make for high performance," Verdu said. But this isn't a new discovery. "People have been talking about high-performance teams for years and years and years."

The magic size, it seems, is between seven and 12 people. Above that threshold, it's hard to get the same sense of family or investment in the work.

The inefficiencies introduced by larger teams could be managed until they grew beyond 60-80 people. "What happened then was we had our group of leads and what were essentially bottlenecks between the leads and these large organizations that were working on different parts of the game."

Leads could only process a little information at a time, and the same was true for their communication output. The development process became inefficient, and a by-product of this organizational structure was a lack of investment and ownership on the part of the team members. "No one felt like it was their game," Verdu said, explaining that the organizational structure sometimes caused feelings of detachment.

Comments on