This essay appeared as a response to Edge’s annual question, “what scientific term or concept ought to be more widely known?”
There’s a concept from computer security known as a class break. It’s a particular security vulnerability that breaks not just one system, but an entire class of systems. Examples might be a vulnerability in a particular operating system that allows an attacker to take remote control of every computer that runs on that system’s software. Or a vulnerability in Internet-enabled digital video recorders and webcams that allow an attacker to recruit those devices into a massive botnet.
It’s a particular way computer systems can fail, exacerbated by the characteristics of computers and software. It only takes one smart person to figure out how to attack the system. Once he does that, he can write software that automates his attack. He can do it over the Internet, so he doesn’t have to be near his victim. He can automate his attack so it works while he sleeps. And then he can pass the ability to someone—or to lots of people—without the skill. This changes the nature of security failures, and completely upends how we need to defend against them.
An example: Picking a mechanical door lock requires both skill and time. Each lock is a new job, and success at one lock doesn’t guarantee success with another of the same design. Electronic door locks, like the ones you now find in hotel rooms, have different vulnerabilities. An attacker can find a flaw in the design that allows him to create a key card that opens every door. If he publishes his attack software, not just the attacker, but anyone can now open every lock. And if those locks are connected to the Internet, attackers could potentially open door locks remotely—they could open every door lock remotely at the same time. That’s a class break.
It’s how computer systems fail, but it’s not how we think about failures. We still think about automobile security in terms of individual car thieves manually stealing cars. We don’t think of hackers remotely taking control of cars over the Internet. Or, remotely disabling every car over the Internet. We think about voting fraud as unauthorized individuals trying to vote. We don’t think about a single person or organization remotely manipulating thousands of Internet-connected voting machines.
In a sense, class breaks are not a new concept in risk management. It’s the difference between home burglaries and fires, which happen occasionally to different houses in a neighborhood over the course of the year, and floods and earthquakes, which either happen to everyone in the neighborhood or no one. Insurance companies can handle both types of risk, but they are inherently different. The increasing computerization of everything is moving us from a burglary/fire risk model to a flood/earthquake model, which a given threat either affects everyone in town or doesn’t happen at all.
But there’s a key difference between floods/earthquakes and class breaks in computer systems: the former are random natural phenomena, while the latter is human-directed. Floods don’t change their behavior to maximize their damage based on the types of defenses we build. Attackers do that to computer systems. Attackers examine our systems, looking for class breaks. And once one of them finds one, they’ll exploit it again and again until the vulnerability is fixed.
As we move into the world of the Internet of Things, where computers permeate our lives at every level, class breaks will become increasingly important. The combination of automation and action at a distance will give attackers more power and leverage than they have ever had before. Security notions like the precautionary principle—where the potential of harm is so great that we err on the side of not deploying a new technology without proofs of security—will become more important in a world where an attacker can open all of the door locks or hack all of the power plants. It’s not an inherently less secure world, but it’s a differently secure world. It’s a world where driverless cars are much safer than people-driven cars, until suddenly they’re not. We need to build systems that assume the possibility of class breaks—and maintain security despite them.