OPINION:IT HAS become conventional wisdom that the reeling US car industry desperately needs to innovate. The hard part for Detroit is working out how. There is hope, even for an entrenched, sprawling company such as General Motors.

We don't need Michael Moore to imagine the misery that would be unleashed if GM were to go under. But before we spend more taxpayer money bailing anyone out, let's recognise that Detroit's worst failure is a recent one, not the long-gone blunder of betting that SUVs were the future.

The real problem is failing to stay competitive with global rivals in the realm of advanced principles of design and manufacturing - principles that exploit global, peer-to-peer information platforms to increase the variety of smash hits a firm might produce. While Volkswagen, Toyota and Renault-Nissan have used these platforms to build flexibility into their product lines, positioning themselves for long-term profitability, GM and Ford have played catch-up on manufacturing "quality" - in effect using the proceeds from SUV sales to fight the last war.

What's the next one? The key to making any manufactured product profitable these days is lowering the transactional costs of designing it. Look at Sony or Samsung or Apple or Honda. What these companies have cultivated is the capacity to experiment. Product teams within each group design prototypes that will appeal to their niche customers. Company leaders then dump the likely losers, batting for singles and the odd home run. (Apple, remember, had no idea that the iPod would be a grand slam.)

The point is, you don't want that much riding on each try. You want (if you'll pardon more sports metaphors) to transform your design-to-manufacturing paradigm from football to basketball - that is, to set yourself up to rush the basket many times per game, not painfully drive your way toward the end zone a few times.

Pulling this off in motor groups such as GM does not mean (except in special cases such as Chevrolet's much-hyped plug-in car, the Volt) a design programme driven from the top. It means having product teams within every brand unit share data about customers, technical specifications on components, relationships with suppliers and so forth with their peers in every other unit. You want design and production managers in each unit to be able to gain access quickly to cutting-edge components developed throughout the group - and find new ways to integrate them into new products.

Take Skoda, the Czech car manufacturer, which in 1991 became a part of the Volkswagen group, the largest car manufacturer in Europe. Skoda is thriving today because its elegance-minded Bohemian designers have learned to exploit the access VW management has given them to virtually the entire spectrum of the conglomerate's components.

At first, Skoda also simply put a Czech skin around the German-engineered Golf. But today, the company creates original cars for low-end, low-tech markets whose boundaries are carefully negotiated with other VW group members (it reportedly exports 80 per cent of its vehicles to 92 countries). Skoda's former CEO, Detlef Wittig, told me that his firm's recent model, the Roomster, would cover its costs after selling only 60,000 units a year.

Skoda is a stirring example. But all global automotive groups need to decentralise their vehicle design this way: empowering interconnected product teams to integrate components, information, code and so forth. The goal should be to introduce new models in ever shorter development cycles (to be competitive these days, car makers have to roll them out in less than 36 months) and to break even on ever smaller production runs - say, 50,000 vehicles, which is less than a third of the US standard on genuinely new models. Sharing components means that suppliers will have to find new ways to achieve economies of scale and plant managers will have to learn assembly operations from one another.

Innovation is mainly a matter of integration. These "modular" principles have been true for consumer electronics for years; car companies are simply getting up to speed. In fact, this set of principles applies to pretty much every high-tech product and every high-tech manufacturing process that produces a low-tech product. Peer-to-peer networks have changed the rules. Nobody is as smart as everybody.

Washington cannot save GM and Ford by handing them money and waiting for them to produce an advanced hybrid. You can't order up innovation; you have to empower entrepreneurial teams to assemble delight, piece-by-piece, for specific customers. The Volt will need to be a part of a family of cars for GM to succeed. The Volt's key power-train components should quickly be absorbed into an environment-friendly Saab, for instance, or into some California-style sports car built around an iPhone. Even the coolest of these models will face stiff competition from global rivals, so they had better be made right.

Governments can help only if they grasp the way manufacturing companies work. The shakiest firms will need a tariff regime that permits an auto group to import components from the country where they are designed or most competitively produced. (The EU's trade rules were a huge help in making it possible for Skoda to acquire components from VW group companies.)

Federal and state governments should help jump-start a grid for electric cars, as Israel is doing. Most important, perhaps, Washington should move to stimulate innovation in entrepreneurial companies along the whole supply chain - companies aspiring to provide new generations of components.

In particular, Washington should make patent protection harder to come by, lest big companies become complacent and would-be competitors get stifled. In much the same way that the US government is now taking stakes in banks that it's bailing out, it should also take stakes in car companies it invests in.

Washington should also work with Detroit's management to make critical RD programmes more transparent and less vulnerable to battles over intellectual property - which would let university labs and freelance inventors join in the quest to innovate. Washington should encourage supplier companies to find new ways to share intellectual property, such as global ideas exchanges that apply something similar to copyright rules to unpatented technological innovations.

None of these actions will guarantee that US car companies will use funds from a bailout to create cars that customers will actually want to buy. If, after several years, US car companies don't improve, some of their key assets and manufacturing capacity could be sold to global firms such as Toyota. Such sales might reinvigorate these plants and suppliers, and they would keep their workers afloat.

The question is not whether the US motor industry faces extinction or whether GM is too big to fail. The question is how the healthy parts of Detroit's behemoths might survive into a new generation and in a changing business landscape.

The time it takes to generate a new car, like the lifespan of a company, is only going to get shorter. During the 1980s and 1990s, as Arie Lewin of Duke University discovered, it took about 13 years for a third of the Fortune 500 to be "selected out" - to fail or be acquired by other firms. Today, it takes about four years.

The coming contraction need not spell disaster for Detroit. But we must understand that financial capital isn't the only kind that flows around the world or is managed and regulated. The same is true of intellectual capital - the sheer capacity to turn learning into stuff. The good news is that the US still has the world's largest proven reserves of intellectual capital.

The bad news is that, unlike oil, the fact that we have it doesn't mean that others do not. - ( LA Times-Washington Postservice)

• Bernard Avishai is a former technology editor of the Harvard Business Review and a former international director at KPMG.