Software Truly Does Not Answer a Trump Election (a Case for Politics)

CC: Susan Melkisethian

In the immediate days after the United States presidential election, several media outlets offered articles instructing readers on how to “secure your smartphone before Trump takes power.” The appeal of these articles was obvious — they provided audiences with a welcome belief that one could reassert agency in an otherwise twisted and unpredictable situation. However, while digital security can mitigate some forms of surveillance, technology alone cannot protect against the nightmare scenario of an illiberal administration that actively undermines fundamental human rights in the United States, nor is privacy the ends in itself.

This article’s intent is to briefly reaffirm two points made elsewhere, that:

  • Media outlets and civil society need to responsibly communicate the limitations of any technology in mitigating potential abuses of an illiberal presidency; and,
  • Software developers need to understand that technology is not an island, and that it must either directly participate in the political process or substantially support those who do so on their behalf.

There are clear reasons for privacy advocates and human rights defenders to be legitimately terrified about the prospects of a Trump Administration. Trump has promised to at a minimum roll back hard fought checks on surveillance. Enough has been said about this topic, and an excellent illustration can be found in the ACLU’s “The Trump Memos,” which includes analysis on his positions on civil liberties as a candidate. Privacy technologies remain dependent on a certain level of “benevolence” on the part of the U.S. Government that will present political and legal challenges to software developers and platforms.

Uncertain Context and Uncertain Roles

Unfortunately, as is all too common, the hot-take digital security advice offered in the press ranged from terrible to inappropriate. Quite simply, complex passwords and two-factor authentication does not protect against a subpoena to Google (but, people should still enable it). The recommendations do provide a decent level of protection against certain tactics in passive bulk network surveillance and local law enforcement agencies (which Micah Lee notes). However, these recommendations were better suited against third-tier state actors, criminals, and domestic violence situations. In response, respected privacy advocates stepped in to offer better solutions.

In response to the flurry of recommendations, the cryptographer and privacy advocate Nadim Kobeissi offered a provocation “Software Does Not Answer a Trump Election,” which challenged the appropriateness of the tools cited in various articles and the state of the ecosystem. While lost amongst condescension against the privacy community, Kobeissi rightly notes that when one’s threat model includes the National Security Agency, or even the Federal Bureau of Investigation, then certain assumptions about the trustworthiness and reliability of systems breaks down. Put another way, it would be hard to credibly argue that Tor — the poster child of the privacy community — is completely safe against the N.S.A., especially if Congress restricts the distribution and development of privacy tools. Or as Kobeissi writes,

The problem here is that this advice is being peddled as if it is directly relevant to protecting oneself against a Trump election, and without any notion of a threat model or concrete security goals.

Kobeissi encourages the reader to focus on the 1.) sociopolitical environment that has promoted a potential authoritarian into office, and 2.) to understand the legal repercussions on software vendors and limitations of certain tools. Indeed, tools alone will not address the situation, and defending against a truly malicious N.S.A. will be an improbable task (which was not likely the argument of those subjected to the criticism).

(It is worth also mentioning that attacking respected lawyers with longstanding histories on digital rights issues is a poor tactic toward common goals.)

No App is an Island

The broader argument has merit, and warranted much more attention. Much of the information security and privacy community has grown up with the individualistic myth of the maverick hacker unilaterally resisting the state (e.g. War Games, Hackers, Sneakers, etc.). As a result, a common sentiment amongst software developers is political disengagement in favor of designing new systems. This mindset fails in merit, and must be abandoned toward an understanding of the proportional role of technology in social change.

In order to account for the liabilities of any privacy recommendation and the coming policy fights, consider the encrypted messaging application Signal running on a iPhone — a common and reasonable recommendation. Signal, maintained Open Whisper Systems, is financially supported by the Open Technology Fund (OTF), a part of Radio Free Asia (RFA). RFA is primarily funded through Congressional appropriations directed to the Broadcasting Board of Governors, an independent federal agency with an international mandate. This means that Signal is largely dependent on the largesse of the United States government–thus far to the tune of at least $2,255,000 (infrastructure costs meaningful sums of money). Other well-known names that appear in the post-Trump articles are similarly dependent on Internet freedom funds, all intended to promote democracy abroad. One decision by the House and Senate Committees on Appropriations could kneecap such applications.

More broadly, the long-term viability of any recommendation is vulnerable to legal mandates that could require exceptional access and weakened encryption. Restrictions on the distribution of end-to-end encryption could make it more difficult to a.) develop Signal inside of the United States, b.) include the protocol in commercial applications, and, c.) distribute the application through gatekeeper markets (AppStore, Play Store, etc). These challenges are all the more apparent when it comes to private implementations of the Signal Protocol (e.g. WhatsApp, Facebook Messenger, and Google Allo), which collect more data for commercial purposes and are less inclined to be outright hostile to the United States government than activist developers.

One’s threat model then must include not only technical design measure but the legal obligations of vendors.

The potential effects of bad policy can proliferate down to local law enforcement and across borders. For example, mandating that hardware vendors provide backdoor access to locked devices would enable local police departments access to encrypted content stored on devices. Moreover, Signal and iOS will not defend against aggressive government hacking empowered by the resources of the United States intelligence community (threat models). This is a political battle that arose under the Obama administration, and will be more difficult with his successor.

History has also demonstrated that the counterproposal to this problem is not solely “decentralization and open source”— building usable systems that scale and functionally compete with commercial products requires significant resources. While Signal’s developers have expressed optimism about their sustainability outside of U.S. funding based on community development, there’s reason to be skeptical. Other experiences suggest that rather than open collaboration, privacy tools often turn into a tragedy of the commons. Moreover, the boom and bust cycle of secure messaging applications after the Snowden disclosures vividly describes the optimism of start-ups and privacy advocates facing the crushing burden of funding and maintenance.

Forward

The structural reform, litigation, socialization, and community organizing required to change the political discourse of the United States will be a maddeningly complex task. Kobeissi takes the easier route in focusing on critiques rather offering specific actions — instead inviting the reader to “focus on the emerging problems” including the “long-term social and institutional decay in the United States.” Personally, it was a disempowering, out-of-college lesson that democratization and civil participation is about education and culture, which is a grueling and slow process that has no sole causal factor nor standard blueprint.

The primary responsibility of the privacy community will be support change agents within society, rather than to act as the central figure in reform. Secure communications tools can only provide the public the means to affect change in society, and cannot be seen as the vision or end goal itself. In the interest of moving the conversation forward, I offer a few general principles in closing, all of which will be both unsurprising and completely unsatisfactory:

  • American technology companies — not solely Fortune 500 companies, but even startups — must be made to understand that international perceptions of their integrity and trustworthiness depends on their successful defense against encroachments on user privacy (Microsoft Corporation v. United States of America, In re Order Requiring Apple Inc. to Assist in the Execution of a Search Warrant Issued by This Court, FISA Court № 08–01).
  • Software engineers must account for the fact that the design of systems and the collection of data predefines the opportunities for law enforcement agencies, and subsequently their ability to comply with onerous demands (see IRTF’s Human Rights Protocol Considerations Research Group and Signal’s inability to comply with user data requests, this does matter even if its shapable by law).
  • Interested members of the public should financially contribute to the civil liberties and digital rights organizations that will be the first line of defense against attempts to rollback recent reforms (see Katherine Maher’s list of organizations, and Bernstein v. United States for why this matters for cryptographers).
  • Civil society must vigorously protect U.S. government funding allocated for Internet Freedom, and encourage private foundations and European governments to play a more active role in financially supporting tools that enable secure communications over the Internet.
  • Experts in cybersecurity, critical infrastructure, and other topical technical subjects should pursue public service, including Fellowships such as TechCongress’s Congressional Innovation Fellowship.
  • Privacy advocates should understand that Trump != the United States Government; opportunities will still exist within federal agencies (e.g. certain Commissioners on the FCC and FTC), the courts, and in Congress to pursue well-informed technology policy; moreover, on some issues privacy advocates will find common cause with the majority party.
  • Researchers and developers should focus on solving the real world problems that manifested themselves uniquely online during this election, such as the trolling and the promotion of fake news that has intimidated critical voices and exacerbates partisanship.

If there’s interest and feedback, I would be happy to amend this list with more recommendations.