Everything Is Overcomplicated

An internet outage exposes the gap between how we think technology might work and how it actually does.

Many Americans woke up this morning to discover that some of the most popular sites on the web were down. CNN, The New York Times, Reddit—even The Atlantic—all suffered issues. Was it a coordinated cyberattack? Something to do with Amazon Web Services? No, it was because of Fastly. As NPR explained, Fastly “provides vital but obscure behind-the-scenes cloud computing services to many of the web’s high profile sites.”

We are living in an era of massively complex technologies, of such complexity that they verge on incomprehensible. Overcomplicated, I’ve called it. But in many cases, we are shielded from the interdependencies and components of these systems that must all work seamlessly in order to operate. Until, that is, something goes wrong. Failures expose the gaps between how we thought a technology worked and how it actually did.

There are many such stories. In 2010, for example, a water main burst in the Boston suburb of Weston, depriving a large fraction of the Boston area’s population access to drinkable water for several days. The water from the Quabbin Reservoir was no longer accessible, and the backup reservoir held the equivalent of “untreated pond water.” The city of Cambridge, though, was one of the few places in the area that got its water from a separate source, and there, the water kept running. Perhaps for the first time in their life, residents of the Boston area were forced to pay attention to the complicated systems that brought water to their homes.

All too often, even experts learn that there’s a gap between how they think a system operates and how it actually does only when something goes wrong. Years ago, when IBM’s Deep Blue played Garry Kasparov in a chess match, it made a curious move in one game. Was it a good move? A bad move? A bug? According to one account, “outside analysts were divided over a mysterious move made by the program, thinking it either weak or obliquely strategic. Eventually, the programmers discovered that the move was simply the result of a bug that had caused the computer not to choose what it had actually calculated to be the best move—something that could have appeared as random play.” The error in Deep Blue’s code was revealed to the programmers solely through the mystery of its actions.

And that is how we live as well: in a world awash with incomprehensible technologies, yet oblivious to their workings until they go wrong. Engineers have labored hard to shield us from this complexity, which is far from a bad thing. We don’t want to have to open command-line terminals all the time or handle machine code. We like our friendly interfaces to our computers. But in the process, we do lose something.

When the Apple Watch was first released, The Wall Street Journal quoted a mechanical-watch enthusiast explaining his passion: “When I wear a mechanical watch, I think of the precision required to make it work on such a micro scale. That’s pretty awesome as opposed to just a chip.” A microchip is unbelievably more complex than a mechanical watch’s gears and complications. The intricacies of a microchip, however, are so small and hidden from view that we forget about their complexity.

And in the process, we can be drawn to responding to the complicated technologies that surround us as objects of either fear or inscrutable magic. We worry that machines will kill us, or we talk about the wonderful mind of Google, tended by its wise masters. But both of these responses are counterproductive, shutting down further inquiry and preventing us from understanding the systems on which we rely.

Better to approach complex technologies in a state of humility, seeking to improve our understanding although we might never fully grasp them. Because in an age when even the experts are not wholly aware of how everything is connected and what might go wrong, we need to try to peek under the hood. Otherwise, the only time we might begin to comprehend how a system works is when it fails.