James Surowiecki writing in the New Yorker:
In the past, the F.A.A. was remarkably hesitant to take planes out of service. The problems with the DC-10 were well known to regulators for years before a 1979 crash forced them to ground the plane. But, again, those standards no longer apply. In the nineteen-seventies, after all, airplane crashes occurred with disturbing regularity. Today, they are extraordinarily rare; there hasn’t been a fatal airliner crash in the United States in almost four years. The safer we get, the safer we expect to be, so the performance bar keeps rising. And this, ultimately, is why the decision to give other companies responsibility for the Dreamliner now looks misguided. Boeing is in a business where the margin of error is small. It shouldn’t have chosen a business model where the chance of making a serious mistake was so large.
Some incomplete thoughts:
But, of course, there is a ‘cost’ to being modular too.
“They still think they are in charge,” says Christensen commenting in general on outsourcing, “but they aren’t. They have outsourced their brains without realizing it.”
Christensen believes outsourcing is driven by ratios and returns.
Americans measure profitability by a ratio. There’s a problem with that. No banks accept deposits denominated in ratios. The way we measure profitability is in ‘tons of money’. You use the return on assets ratio if cash is scarce. But if there is actually a lot of cash, then that is causing you to economize on something that is abundant. …Modular disruptors should carry their low-cost business models up-market as fast as possible, to keep competing at the margin against higher-cost markers of proprietary products
This, however, can lead to disaster.
If you study the root causes of business disasters and management missteps, you’ll often find a predisposition toward endeavors that offer immediate gratification. Many companies’ decision-making systems are designed to steer investments to initiatives that offer the most tangible returns, so companies often favor these and short-change investments in initiatives that are crucial to their long-term strategies.
Will anyone be able to “snap” together a plane in the future?
…if it’s becoming commoditized and modular, you cannot make money at that level in the stack. But the whole industry doesn’t become unprofitable, rather its activities above and below that original [product or service], that’s where the money is made. And that has to be happening in the pharmaceutical industry, but I can’t see what it is yet. By example, the auto industry is becoming commoditized; cars are being assembled by sub-assemblies from tier-one suppliers. Anybody can get these modules and snap together a car. So it’s really hard to differentiate your car from anybody else’s car, so where the money is being made is in the subsystems that define the performance of the car, and by activities that sit on top of that, like OnStar. That’s where the money is made.
And these thoughts:
Before long, modularity rules, and commoditization sets in. When the relevant dimensions of your product’s performance are determined not by you but by the subsystems, it becomes difficult to earn anything more than substinence returns. When your world becomes modular, you’ll need to look elsewhere in the value chain to make any serious money.
I'll end with some more thoughts from Christensen
But even in a modular architecture, successful companies still are integrated—just in a different place. Consider the computer industry in the 1990s. The computer's basic performance was more than good enough. What did customers want instead? They wanted lower prices and a computer customized for their needs. Because the product's functionality was more than good enough, companies like Dell could outsource the subsystems from which its machines were assembled. What was not good enough? The interface with the customer. By directly interacting with customers, Dell could ensure it delivered what customers wanted—convenience and customization. Value flowed to Dell and to the manufacturers of important subsystems that themselves were not good enough, like Microsoft and Intel.
In short, companies must be integrated across whatever interface drives performance along the dimension that customers value. In an industry's early days, integration typically needs to occur across interfaces that drive raw performance—for example, design and assembly. Once a product's basic performance is more than good enough, competition forces firms to compete on convenience or customization. In these situations, specialist firms emerge and the necessary locus of integration typically shifts to the interface with the customer.
The real take-away is knowing what industry you're in, the complexity of your products, and the second and third order effects of outsourcing.
If you do outsource, pay attention to what you're giving up in terms of information and complexity — what starts as “raw labor” easily moves up the value chain. Pay attention to the reasons for your outsourcing: are they all driven by financial ratios? If so, that's a red flag. Try to think a decade ahead. I know this all sounds very ambiguous and difficult but if it were easy everyone would know the answer.