The National Academy of Sciences (NAS) released a much-anticipated report yesterday that attempts to influence the encryption debate by proposing a “framework for decisionmakers.” At best, the report is unhelpful. At worst, its framing makes the task of defending encryption harder.
The report collapses the question of whether the government should mandate “exceptional access” to the contents of encrypted communications with how the government could accomplish this mandate. We wish the report gave as much weight to the benefits of encryption and risks that exceptional access poses to everyone’s civil liberties as it does to the needs—real and professed—of law enforcement and the intelligence community.
From its outset two years ago, the NAS encryption study was not intended to reach any conclusions about the wisdom of exceptional access, but instead to “provide an authoritative analysis of options and trade-offs.” This would seem to be a fitting task for the National Academy of Sciences, which is a non-profit, non-governmental organization, chartered by Congress to provide “objective, science-based advice on critical issues affecting the nation.” The committee that authored the report included well-respected cryptographers and technologists, lawyers, members of law enforcement, and representatives from the tech industry. It also held two public meetings and solicited input from a range of outside stakeholders, EFF among them.
EFF’s Seth Schoen and Andrew Crocker presented at the committee’s meeting at Stanford University in January 2017. We described what we saw as “three truths” about the encryption debate: First, there is no substitute for “strong” encryption, i.e. encryption without any intentionally included method for any party (other than the intended recipient/device holder) to access plaintext to allow decryption on demand by the government. Second, an exceptional access mandate will help law enforcement and intelligence investigations in certain cases. Third, “strong” encryption cannot be successfully fully outlawed, given its proliferation, the fact that a large proportion of encryption systems are open-source, and the fact that U.S. law has limited reach on the global stage. We wish the report had made a concerted attempt to grapple with that first truth, instead of confining its analysis to the second and third.
We recognize that the NAS report was undertaken in good faith, but the trouble with the final product is twofold.
First, its framing is hopelessly slanted. Not only does the report studiously avoid taking a position on whether compromising encryption is a good idea, its “options and tradeoffs” are all centered around the stated government need of “ensuring access to plaintext.” To that end, the report examines four possible options: (1) taking no legislative action, (2) providing additional support for government hacking and other workarounds, (3) a legislative mandate that providers provide government access to plaintext, and (4) mandating a particular technical method for providing access to plaintext.
EFF raised concerns that encryption does not just support free expression, it is free expression.
But all of these options, including “no legislative action,” treat government agencies’ stated need to access plaintext as the only goal worth study, with everything else as a tradeoff. For example, from EFF’s perspective, the adoption of encryption by default is one of the most positive developments in technology policy in recent years because it permits regular people to keep their data confidential from eavesdroppers, thieves, abusers, criminals, and repressive regimes around the world. By contrast, because of its framing, the report discusses these developments purely in terms of criminals “who may unknowingly benefit from default settings” and thereby evade law enforcement.
By approaching the question only as one of how to deliver plaintext to law enforcement, rather than approaching the debate more holistically, the NAS does us a disservice. The question of whether encryption should or shouldn’t be compromised for “exceptional access” should not be treated as one of several in the encryption debate: it is the question.
Second, although it attempts to recognize the downsides of exceptional access, the report’s discussion of the possible risks to civil liberties is notably brief. In the span of only three pages (out of nearly a hundred), it acknowledges the importance of encryption to supporting values such as privacy and free expression. Unlike the interests of law enforcement, which are represented in every section, the report discusses the risks to civil liberties posed by exceptional access as just one more tradeoff, and addresses them as a stand-alone concern.
To emphasize the report’s focus, the civil liberties section ends with the observation that criminals and terrorists use encryption to “take actions that negatively impact the security of law-abiding individuals.” This ignores the possibility that encryption can both enhance civil liberties and preserve individual safety. That’s why, for example, experts on domestic violence argue that smartphone encryption protects victims from their abusers, and that law enforcement should not seek to compromise smartphone encryption in order to prosecute these crimes.
Furthermore, the simple act of mandating that providers break encryption in their products is itself a significant civil liberties concern, totally apart from privacy and security implications that would result. Specifically, EFF raised concerns that encryption does not just support free expression, it is free expression. Notably absent is any examination of the rights of developers of cryptographic software, particularly given the role played by free and open source software in the encryption ecosystem. It ignores the legal landscape in the United States—one that strongly protects the principle that code (including encryption) is speech, protected by the First Amendment.
The report also underplays the international implications of any U.S. government mandate for U.S.-based providers. Currently, companies resist demands for plaintext from regimes whose respect for the rule of law is dubious, but that will almost certainly change if they accede to similar demands from U.S. agencies. In a massive understatement, the report notes that this could have “global implications for human rights.” We wish that the NAS had given this crucial issue far more emphasis and delved more deeply into the question, for instance, of how Apple could plausibly say no to a Chinese demand to wiretap a Chinese user’s FaceTime conversations while providing that same capacity to the FBI.
In any tech policy debate, expert advice is valuable not only to inform how to implement a particular policy but whether to undertake that policy in the first place. The NAS might believe that as the provider of “objective, science-based advice,” it isn’t equipped to weigh in on this sort of question. We disagree.