Jon Bell has a good post about brainstorming this week. He outlines his "McDonald's Theory" of how to improve team brainstorming. According to Bell, he has used an effective technique whenever folks at work are trying to decide where to go to lunch. He will throw out the idea of going to McDonald's. Everyone immediately recoils at the idea, and then they begin trying to come up with better ideas. Bell argues that teams should approach brainstorming the same way. Be willing to throw out some crazy ideas in the early going. Get the creative juices flowing. If nothing else, you have given the group a notion of what they don't want to do. In some cases, the seemingly ridiculous idea will trigger some other quite interesting and useful notions.
The McDonald's Theory is quite consistent with one of the mantras espoused by leading product design firm IDEO. They argue that leaders must, "Encourage wild ideas" - particularly during the early stages of brainstorming. I would simply add one key point to Bell's thoughts. In his lunch example, people often reject the McDonald's suggestion outright. In a good team brainstorming session, people should resist rejecting those crazy early ideas. Instead, they ought to adopt a key rule from improv comedy. They ought to employ the "yes, and" approach to building on that idea. Accept the idea, as crazy as it may be, but build on it. Find some kernel in there that's useful and take the discussion from there.