From the piece:
G&T say that you need to start by recognising that innovation is unnatural. Established businesses are built for efficiency, which depends on predictability and repeatability—on breaking tasks down into their component parts and holding employees accountable for hitting their targets. But innovation is by definition unpredictable and uncertain. Bosses may sing a pretty song about innovation being the future. But in practice the heads of operational units will favour the known over the unknown.
G&T argue that companies need to build dedicated innovation machines. These machines need to be free to recruit people from outside (since big companies tend to attract company men rather than rule-breakers). They also need to be free from some of the measures that prevail in the rest of the company. But they must avoid becoming skunk works. They need to be integrated with the rest of the company—they must share some staff, for example, and they must tap into the wider company’s resources as they turn ideas into products. And they must be tightly managed according to customised rather than generic rules. For example, they should be held accountable for their ability to learn from mistakes rather than for their ability to hit their budgets.
Sounds good, but it's easy to give recipes. Still, it's a good quote