Why ‘Back Doors’ Are a Bad Idea

Winter 2016

Why Back Doors Are a Bad Idea… and why continued debate leaves us vulnerable.

For the first time in history, the technology industry has achieved a longtime dream of restoring user privacy through widespread access to encryption technology. But the reaction from governments has been a near panic. FBI director James Comey recently took his case to Congress, arguing that this technology risks assisting terrorist groups like ISIS. Law enforcement agencies in the U.S. and abroad have publicly mooted the addition of some sort of “back door” that would give them access to Silicon Valley’s encryption.

Indeed, law enforcement has been so vocal in its concerns that it would be reasonable for a nontechnical reader to conclude encryption is some sort of radical idea. But this is hardly the case. The reason for deploying end-to-end encryption has nothing to do with ISIS; it’s fundamentally about securing our online systems—systems that, if the news is any indication, are more insecure than they ever have been.

From the beginning of electronic communications until the 1990s, most data was transmitted unencrypted and over relatively insecure communication lines.

But promises of good behavior that work well in a network with several dozen machines are less than worthless on a global network with billions of connected machines—many of which are actively controlled by organized criminals. To work in this modern world, security engineers have adopted a single maxim: Treat the network as hostile. We assume that every machine and network connection we don’t actively trust is trying to harm us.

Fundamentally, encryption allows us to “extend trust” from one small trusted location to another, regardless of what lies between.

Law enforcement agencies are not against strong encryption. What they ask is a bit subtler: They’d like a new form of encryption that’s strong against everybody but law enforcement. Proponents of this approach would see all encryption technology equipped with a “back door” that could be restricted only to lawful requests. If this encryption was deployed, they argue, law enforcement and digital security could peacefully coexist.

It’s a beautiful dream. It just seems extraordinarily difficult to realize, at least at the scale required. The technical reasons for this are many and varied, but rather than re-hash this debate, I would make a different point: Until the issue is resolved with concrete technical proposals, even the “debate” itself is doing damage.

From the point of view of U.S. companies, there already exists a readily available technique for building “back doors” that is inexpensive and easy to deploy today. And that technique is to avoid deploying end-to-end encryption in the first place. The uncertainty created by the current Washington debate, combined with the lack of any concrete technical proposal, is, in fact, already motivating companies to avoid deploying such encryption.

And this may seem just fine by you if your concern is limited to simplifying the task of law enforcement. But if you think more about the security of U.S. information systems, it’s an absolute tragedy. The primary lesson we’ve learned over the past several years is that our current security measures are fundamentally not up to the task of keeping corporate and government secrets out of the hands of motivated attackers. The losses we’ve sustained so far are likely only the beginning.

By prolonging this debate any longer and failing to offer up concrete technical solutions, our own governments may be dooming us to another decade of poor security. This is a price we can’t afford to pay.