Why the brash new car scene runs on hot air.

Take a hard look at the global car business. General Motors (GMGMQ) is in the midst of the largest industrial bankruptcy in history. Chrysler just emerged from Chapter 11 into an unlikely and unsure alliance with Fiat. Ford (F) had to mortgage everything in sight to stay afloat. Toyota and Honda are suffering through colossal sales slumps. So are Mercedes and BMW. And Hyundai is offering to let you give your car back if you lose your job.

So the auto industry is a mess, and this has set off a frenzied hunt for alternatives to the old way. A band of evangelists has assembled around the notion that the cure is automotive startup culture. This sounds incredibly appealing. But it's completely wrongheaded.

At the moment, there's certainly no shortage of small startups, each pledging to remake the business model melded in the mid-20^th century. Tesla Motors, Bright Automotive, Fisker, Better Place, and Aptera are among the firms vying to supplant the Big Three (reduced now to the Sort-of-Big One and Those Two Other Guys With All the Problems). There's a natural tendency to want to boost those newcomers given Detroit's plunge. But the startups have their own issues. They're uniformly undercapitalized. And often trying to build cars that are decades ahead of their time.

The auto industry is still a heavy-metal undertaking. Robots do some of the work on assembly lines, but in many plants, so does unionized labor: men and women taking pieces of things—dashboard, tires, doors—and turning them into a Chevy. It's skilled physical work, which unsettles the new economy given its embrace of brainwork over brawn. And yet the startup supporters are now aggressively applying the kind of thinking we're all familiar with from the tech sector to the auto industry. They can't stop comparing cars to computers. But it just doesn't work.

Exhibit A is Charles C. Mann's proposal in the current issue of Wired. In his estimation (and he's far from alone here), the auto industry of tomorrow should develop along the same lines as the tech industry of the late 20^th century: