The Lesson of the Delta Debacle: The Future Isn't That Fast

You probably saw the reports about a power outage that led to a massive failure in the Delta Air Lines online reservation system. Hundreds of flights were cancelled, and they're still not done sorting it out. This followed another failure earlier for Southwest Airlines.

Note that part of the problem was "antiquated technology," as mergers in the airline industry piled legacy system on top of legacy system and left them "reliant on IT systems that date from the 1990s." Certainly, the airlines might be able to create a more reliable system by building the whole thing up from scratch with new technology. But that's actually quite expensive and difficult, requiring new equipment and extensive training. And the airlines still have to connect with other organizations (like airports) that also have legacy systems. So they limp along with the old technology, hoping to slowly turn it into something newer, patch by patch.

From this little debacle in one industry, we can draw some important lessons for the future of technological change in every industry. The main lesson is that the future rarely arrives as fast as it seems like it should.

That's worth pointing out because sometimes both the boosters and the alarmists assume that new technology will change the world much more rapidly and extensively than is really likely. Take a recent Vox analysis worrying about the sudden disappearance of relatively well-paid blue-collar trucking jobs as self-driving trucks take over. The key line is: "Maybe it’s two years, maybe five, maybe 10, but either way, the trajectory is toward drivers being put out of business." Now, let's stipulate that any estimate that vague should probably just be shortened to: "I don't know." But given that autonomous vehicles are still very much in the testing phase, it's definitely not going to be two years, and it probably won't even be ten years. Judging from history, it will probably take closer to 30 years before self-driving trucks are the norm.

There is a whole host of reasons. The first is that self-driving trucks are going to be expensive. The new computers and software and sensors will require a significant layout of capital that will be used only for the most high-value applications first. The nation's entire long-haul trucking fleet costs billions of dollars to replace, and that will be done over a long period of time. A typical semi-trailer used for long-haul applications can last for a million miles over ten years. So it will take longer than that to phase them out in favor of new technology.

Plus, the first iteration of this technology, and probably the first several iterations, is going to have a lot of limitations. For now, it's projected to only work well in the relatively simple and controlled environment of major interstate highways. But it won't work yet on smaller side roads, much less negotiating the loading dock at the Piggly Wiggly. And part of a truck driver's job is to supervise the loading and unloading of his cargo, which the truck won't be able to do for him. So to begin with, self-driving trucks won't take away the jobs of truckers. They will make their job easier, sparing them the tedium and fatigue of long hours of highway driving.

So a combination of the expense of replacing older vehicles and the continuing limitations of new vehicles mean that we're likely to see a very long phase-in for self-driving technology and a very long phase-out for human drivers, some of which will simply be a shift to new jobs that open up in the transportation industry. (For example, new and even better-paying jobs for highly skilled mechanics to service the fleets of self-driving trucks.)

But I guarantee you that forty years from now, there will still be a guy who earns a living making deliveries in a beat up old box truck. After all, in big cities there are still people who get paid to hand-deliver single packages by bicycle.

Or consider this. The very last manufacturer of VCRs, a technology from the early 1980s, just stopped producing them last month. Some products that are even older stubbornly refuse to die: fax machines, floppy disks, beepers—even, gasp, Windows XP. The last telegram, a technology,not from the last century but from the century before, was delivered in 2013.

This is why the alarmists are wrong. The history of technology and of the economy is a history of constant disruption, to which we are constantly adapting. But it is also a history of non-disruption, of old ways that persist because they still work, and because for some uses, it's just too much cost and trouble to replace them.

The future, like your Delta flight, is just not going to take off as quickly as you thought it was.