Encryption Substitutes
Encryption Substitutes
Introduction
Policy experts have suggested that the rise of encrypted data is not the end of
intelligence collection because law enforcement can look to substitutesother
sources of intelligence, such as metadatathat prove to be just as valuable or more
valuable than decrypting encrypted data.1 This paper focuses on the other side of that
insight: on the substitutes available for privacy-seekers beyond encryption, such as
placing ones data in a jurisdiction that is beyond the reach of law enforcement. This
framework puts encryption in context: there are many ways to keep ones data private,
just as there are many ways that the government might get access to that data. While
encryption is typically treated as a stand-alone computer security issue, it is a piece of
alarger debate about government access to personal data.2
Law enforcement officials are, in general, agnostic about the method through which
they obtain evidencewhat matters is obtaining it. Privacy-seekers are similarly
agnostic about how they secure their privacywhat matters is having it. This means
that policymakers have a wide set of optionsnot only about whether to allow law
enforcement to access personal data, but also how to do so. This wide set of options
is not reflected in the debate over encryption, which is typically framed in all-or-
nothing terms. Some privacy advocates take a stance that seems to allow no room
for compromise (an argument that can be boiled down to its math!3) and some
government actors do the same (essentially arguing, its terrorism!4). Widening the
scope of the policy discussion to include related issueswhat I will call encryption
substitutesmay increase the chances of compromise and may generate better
policy.
In this short essay, I make a few simple assumptions that bear mentioning at the outset.
First, I assume that governments have good and legitimate reasons for getting access
to personal data. These include things like controlling crime, fighting terrorism, and
regulating territorial borders. Second, I assume that people have a right to expect
privacy in their personal data. Therefore, policymakers should seek to satisfy both
law enforcement and privacy concerns without unduly burdening one or the other.
Of course, much of the debate over government access to data is about how to respect
2
both of these assumptions. Different actors will make different trade-offs. My aim in
this short essay is merely to show that regardless of where one draws this linewhether
one is more concerned with ensuring privacy of personal information or ensuring
that the government has access to crucial evidenceit would be shortsighted and
counterproductive to draw that line with regard to one particular privacy technique
and without regard to possible substitutes.
The first part of the paper briefly characterizes the encryption debate two ways: first,
as it is typically discussed, in stark, uncompromising terms; and second, as a subset
of a broader problem. The second part summarizes several avenues available to law
enforcement and intelligence agencies seeking access to data. The third part outlines
the alternative avenues available to privacy-seekers. The availability of substitutes is
relevant to the regulators but also to the regulated. If the encryption debate is one
tool in a game of cat and mouse, the cat has other tools at his disposal to catch the
mouseand the mouse has other tools to evade the cat. The fourth part offers some
initial thoughts on implications for the privacy debate.
The debate about backdoors to encryption leaves little room for compromise.
One side characterizes the governments demands for exceptional access as math
denialism: exceptional access simply cannot be introduced into a cryptographic
system without overwhelming risk.5 The other side insists that it must be done and it
can happen, if only cryptographers and software engineers try hard enough. Former
FBI director James Comeys recent testimony on the matter is a good example
suggesting that Silicon Valley entrepreneurs simply need to apply the same grit and
determination to the encryption problem that they apply to creating new software
businesses.6 The terms of this debate are zero-sum: either it is technologically
possible to create a system that is safe but also contains a backdoor, as the FBI asserts,
or it is not.
Perhaps there is a better way to frame the debate. The government does not actually
seek exceptional access to encrypted data per se; indeed, governments did not
seekexceptional access until it became relevant to law enforcement operations.
Whatthe government is really after is crucial evidence of crimes and national
securityintelligence. Encryption is just one barrieramong manyto that evidence
and intelligence.
Consider two recent high-profile lawsuits: Apples refusal to comply with an order
to create software to unlock an iPhone7 and Microsofts refusal to comply with a
warrant compelling the production of evidence stored on overseas servers.8 These
are distinct domains as a matter of public policy, public relations, and law. The first
is about whether the All Writs Act authorizes a judge to compel Apple to write new
software that can be deployed to weaken the security of the companys phones. This
is a case about encryption and technological barriers to the states ability to access
personal data. The Microsoft case, on the other hand, is about whether the Stored
Communications Acts warrant provisions apply extraterritorially.9 This is a case
aboutjurisdictional limits on the states ability to access personal data. At a legal
doctrinal level they are different cases, and in the public eye they are different
cases.
But they share many similarities. Both cases are about the authority of a US judge
to compel an American company to produce data about one of its customers in
connection with a criminal investigation into that customers activity. In both cases,
the company objects to giving the government the relevant information, alleging
that to do so would harm the privacy of the companys other customers and would
gravely harm the companys reputation.10 Both companies stand in the way of the
governments acquisition of information. Indeed, it may be the case that the increasing
use of encryption on devices located domestically is driving the Department of Justice
to seek information stored abroad, and vice versa.11 Viewed in this light, the disputes
are quite similar and perhaps even interrelated.
But even if some channels of communication have gone dark, other new sources of
intelligence are filling the void. Indeed, as Apples manager of user privacy testified in
that companys dispute with the FBI over access to an encrypted phone: There are
several other ways the government could have potentially obtained any data stored
on the subject device.15 Far from going dark, some suggest that this is in fact the
golden age of surveillance.16 Encryption is a cryptographic tool for ensuring that only
authorized users can read and understand data.17 It can be deployed on hardwareas
in an iPhone, to protect data at restand it can be deployed by services to protect
data in transit. Importantly, users might have one form of cryptographic security
but not another: someone might communicate in the clear but store communications
locally in encrypted form; others might communicate through an encrypted channel
but store their data locally on an unencrypted disk. Finally, quite apart from these
different forms of encrypting data are the use and anonymization of metadata which
can be used to identify and track a users online activities. Each of these is a potential
avenue for government evidence gathering.
Equipment Interference
This is relevant to a number of ongoing debates. For example, the Senate recently
adopted a recommended change to Rule 41 of the Federal Rules of Criminal Procedure,
which allows courts to compel device-hacking on a broad scale.20 In the United
Kingdom as well, similar rules have been adopted. The Investigatory Powers Act,
a comprehensive bill that reformed the governments ability to access user data,
authorizes equipment interference, whereby the government can seek a way around
a devices cryptographic security.21 The motivation for these bills is simple: without
a way around the disk-level encryption that protects the devices used by criminal
suspects for local data storage, law enforcement officials argue, they cannot access
critical evidence.
Metadata
Perhaps the most fundamental challenge to the idea that law enforcement is
going darkor that going dark is actually a problemis the ready availability
of an increasingly powerful source of intelligence: metadata. Metadata or non-
content dataoutside the envelope information, such as sender and receiver
identification, IP address, basic subscriber information, date, time, and location
datacan be surprisingly revealing.22 This information is often as valuable or more
so for law enforcement than content data.23 Search-and-seizure lawin the form
of the Fourth Amendment doctrine, the Omnibus Crime Control and Safe Streets
Act, and the Electronic Communications Privacy Actdraws a sharp distinction
between content and non-content data, typically providing fewer legal barriers to law
enforcement attempts to access metadata.24 With enough of this non-content data,
law enforcement can gather and infer enormously useful information, such as whom
a subject was communicating with, about what, where, and whenmuch of the
most important information for conducting criminal investigations.25 Since this non-
content information is not typically encrypted, cries of going dark may ring hollow.
Law enforcement has access to an enormous new trove of non-content data in the
form of e-mail logs, GPS location data, and more.
Market-Driven Data
Perhaps the best counter to the going dark worries was made in a report by Harvards
Berkman Klein Center for Internet & Society that argues that law enforcement has
accessor will soon have accessto ready substitute avenues of intelligence if and
when current channels go dark.26 The report finds that there are structural reasons why
many Internet communication channels will never be fully encrypted. For example,
while Apple can afford to take a strong pro-encryption stance because it derives
most of its revenue from hardware sales, Google and Facebook make their money on
advertisements, which often require the ability to scan through user data, a task that is
currently not possible if the data is encrypted.27 Even if Google and Facebook roll out
services that are encrypted, such as WhatsApp (a Facebook product), other services will
remain in the clear. The market for technology services is diverse, and this means that
even if some services end up encrypted, others will remain unencryptedtypically
so that the company offering the service can monetize user data (primarily through
advertisements).
Moreover, even if some communication services go dark, the wide adoption of sensors
in everyday productsthe so-called Internet of Thingswill mean that there are
many, many sources of data available to law enforcement beyond phone calls and
e-mails.28 Security cameras, thermostats, Internet-connected refrigerators, voice-
enabled assistants like Amazons Echo and Googles Homethese are a few of the
many devices that now collect data about their users, data that can be scooped up
by law enforcement agents. Two recent examples are illustrative. In December2016,
law enforcement agents requested data from an Amazon Echo device installed in the
home where a crime allegedly occurred.29 In another example, a Connecticut man
is being charged with murder in part based on the data taken from his wifes Fitbit
fitness device, which suggested he was lying about his activities on the day of the
murder.30
Privacy Substitutes
Jurisdictional Substitutes
Blocking Statutes Suppose that the French government investigates a murder in Paris.
If the suspect uses an American Internet service like Gmail, which claims to store the
users data in the United States, then the French authorities would not be able to access
the suspects e-mail using French legal process. Instead, the French authorities would
need to ask the United States for mutual legal assistance. This process takes upward
of a year and effectively means that the content is unavailable for law enforcement
purposes. This is all the result of the fact that the data resides in America, and the US
Electronic Communications Privacy Act functions as a blocking statute, preventing
Google from complying with a legitimate French legal process. The presence of a
blocking statute is as much a barrier between law enforcement and the evidence
itseeks as any technological barrier.
Technological Substitutes
Just as there are jurisdictional substitutes to device encryption, there are technological
substitutes as well.
Anonymization Tools Perhaps the most useful way to ensure a measure of privacy
online is to operate anonymously. Often, law enforcement will need some amount
of identifying metadata before it can search or seize a suspects digital content
data, like e-mails and photos. Suppose that the police receive a tip (or intercept a
message) that suggests that a criminal is communicating using the e-mail account
[email protected]. Without having some way to connect a particular suspect
to this account number, it may be difficult for law enforcement to gather enough
evidence to ask for a warrant to get access to the accounts contents. This is
why privacy-seekers use anonymization tools like Tor, which masks their online
activities. 35 If hiding the contents of your communications is good, not having
anyone know they are yours is even better. If a user cannot be identified, it does
not matter what his messages say (or whether the messages are encrypted, either
locally or in transit).
What conclusions can we draw from the fact that encryption is neither the only tool
available for privacy-seekers nor the only barrier to law enforcement seeking access to
digital evidence? I think at least five conclusions follow.
It seems likely that social preferences will be maximized by picking one domain
to delimit government access to data over another. Each domain presents a different
set of privacy trade-offs and social preferences will be maximized by some domains
more than others. Consider the following privacy concerns, which any particular
government action might trigger: How widespread is the privacy harm (how many
peoples privacy interests are at stake)? How total is the privacy harm (how much
stuffand what percentagegets revealed to the government)? How long (temporally)
is the harm (finite or otherwise)? And so on. Reducing jurisdictional barriers to law
enforcement access to data may, on balance, be preferable to creating exceptional
access to encrypted services. That is, building a backdoor to Gmails servers may
raise more of these concerns than the contemplated US-UK agreement regarding law
enforcement access to data. Even among technological domains, one approach to
delimiting government access to data may more closely track social preferences than
another. Forexample, allowing the police to lawfully hack into individual suspects
devices islikely less privacy-invasive than forcing providers to create backdoors to
theirservices.
Suppose that the French government is investigating an attack in Paris. The suspect is
thought to have used an encrypted device to communicate with his conspirators using
an unencrypted American service. The government has two avenues for accessing
relevant evidence: decrypt the phone or obtain the suspects e-mail from the American
service. If jurisdictional barriers prevent the French government from obtaining the
e-mails from the American service, there will be significant pressure to decrypt the
phone. If the e-mails can be obtained in a timely manner by asking the US service for
them, the pressure to decrypt the phone may wither.
Suppose that the US government is contemplating two laws. One would prohibit
American providers from encrypting users communications while the other would
allow the government to obtain warrants to hack into suspects devices. Which
of these laws is more easily evaded by criminals? Undoubtedly, it is the former. If
encryption services in the United States were to be weakened or eliminated, they
would be replaced tomorrow by hundreds of others offered overseas. Lawbreaking
behavior could easily be displaced from one service based in the United States to
another based overseas. But under the latter regime, there is no obvious displacement.
If the government can obtain and execute warrants to hack into devices, it will be
harder for criminals to sidestep the law.38
There is a substantial scholarly literature that suggests that negotiators are more likely
to achieve mutually beneficial agreements by linking issues that allow each party to
compromise on an issue of low priority in exchange for an issue of high priority.39
This suggests that jurisdictional and technological barriers to government access to
data ought to be considered together, not separately. Yet this is not the considered
view of either civil societywhich treats encryption as a third railor government
actors, who have largely negotiated these issues discretely. Both sets of actors seem
to have calculated that they are better off negotiating about encryption, cross-
border data access, and lawful hacking in isolation. Presumably, both sides imagine
that they can win in each domain. But if the government cares only about getting
relevant informationand weve seen that there are several largely interchangeable
avenues for obtaining that informationthen to grant government access in none of
these domains will be suboptimal, just as granting government access in all of these
domains will be suboptimal. The key is picking the right domainsomething that can
only happen if theyre considered side by side.
Conclusion
Notes
1Dont Panic: Making Progress on the Going Dark Debate, Berkman Klein Center for Internet & Society,
Harvard University, February1, 2016, https://cyber.harvard.edu/pubrelease/dont-panic/Dont_Panic
_Making _Progress _on_Going _Dark _Debate.pdf.
2Generally, the debate over encryption occurs at a level of abstraction that suggests a lack of
sophistication about the technical aspects of encryption, so I hesitate to suggest that the encryption
debate ought to be even broader than it already is. But for reasons that I hope will become clear, I think
compromise and sensible policy are most likely when we consider encryption in the context of other,
similar domains.
3See, e.g., Rainey Reitman, An Open Letter to President Obama: This is About Math, Not Politics, Medium,
March18, 2016, http://bit.ly/2drhyR9.
4See Encryption Tightrope: Balancing Americans Security and Privacy, Hearings Before the Senate
Judiciary Committee, December1, 2015 (testimony of JamesB. Comey), http://bit.ly/2d3fgeM.
5Cory Doctorow, Obama: Cryptographers Who Dont Believe in Magic Ponies Are Fetishists, Absolutists,
BoingBoing, March12, 2016, http://boingboing.net/2016/03/12/obama- cryptographers-who- don.html.
7Eric Lichtblau and Katie Benner, Apple Fights Order to Unlock San Bernardino Gunmans iPhone,
NewYork Times, February18, 2016.
8In re Warrant to Search a Certain EMail Account Controlled and Maintained by Microsoft Corp., 829 F.3d
197 (2d Cir. 2016).
9Electronic Communications Privacy Act of 1986, Pub. L. No.99-508, 100 Stat. 1848 (codified as 18 U.S.C.
1367, 2521, 27012711, 3117, 31213127 (2013)).
10See Microsofts Objections to the Magistrates Order Denying Microsofts Motion to Vacate in Part a
Search Warrant Seeking Customer Information Located Outside the United States, 8, Microsoft Email
Search Warrant Case, 15 F. Supp. 3d 466 (S.D.N.Y. 2014) (No.13-Mag-2814) (arguing, implicitly, that corporate
reputation depends on the firms ability to reject law enforcement demands); Notice of Objections to
February16, 2016 Order Compelling Apple Inc. to Assist Agents in Search, In the Matter of the Search of an
Apple IPhone Seized During the Execution of a Search warrant on a Black Lexus IS300, California License
Plate 35KGD203, March22, 2016, CD No. CM 16-10 (SP).
11Swire and Hemmings have argued, essentially, that increasing use of encryption causes more
pressure on the cross-border data request regime. See Peter Swire and JustinD. Hemmings, Mutual
Legal Assistance in an Era of Globalized Communications: The Analogy to the Visa Waiver Program, New
York University Annual Survey of American Law 71, no.687 (2017), http://papers.ssrn.com/sol3/papers.cfm
?abstract_id=2728478.
12JamesB. Comey, Going Dark: Encryption, Technology, and the Balances between Public Safety and
Privacy, Joint Statement with Deputy Attorney General Sally Quillian Yates Before the Senate Judiciary
Committee, July8, 2015, https://w ww.fbi.gov/news/testimony/going- dark- encryption-technology-and-the
-balances-between-public-safety-and-privacy.
13Ibid.
14For a long list of articles addressing the going dark problem going back many years, see Going Dark,
Lawfare (blog), https://w ww.lawfareblog.com/topic/going-dark.
15Declaration of Erik Neuenschwander in Support of Apple Inc.s Motion to Vacate Order Compelling Apple
Inc. to Assist Agents in Search, and Opposition to Governments Motion to Compel Search, paragraph 54,
In the Matter of the Search of an Apple IPhone Seized During the Execution of a Search warrant on a Black
Lexus IS300, California License Plate 35KGD203, March22, 2016.
16Peter Swire and Kenesa Ahmad, Going Dark Versus a Golden Age for Surveillance, Center for
Democracy and Technology, November28, 2011.
18Hayley Tsukayama, The Two Sides of the Apple Debate, Washington Post, February19, 2016, http://
wapo.st/2drHHzn.
20Swati Khandelwal, Rule 41FBI Gets Expanded Power to Hack Any Computer in the World, Hacker
News, November30, 2016, http://thehackernews.com/2016/11/fbi-rule- 41-hacking.html#sthash.GsAGAKsQ
.dpuf.
21Matt Burgess, What Is the IP Act and How Will It Affect You? Wired, May8, 2017, www.w ired.co
.uk/a rticle/ip -b ill-law- d etails-passed; Daniel Severson, Taking Stock of the Snoopers Charter:
TheU.K.s Investigatory Powers Bill, Lawfare (blog), March14, 2016, https://w ww.lawfareblog.com
/n ode/10447.
22See MatthewJ. Tokson, The Content/Envelope Distinction in Internet Law, William & Mary Law Review
50, no.6 (2009): 212425.
23See generally StevenM. Bellovin, Matt Blaze, Susan Landau, and StephanieK. Pell, Its Too
Complicated: How the Internet Upends Katz, Smith, and Electronic Surveillance Law, Harvard Journal of
Law & Technology 30, no.1 (2016) (describing several ways in which the content/non-content distinction is
becoming blurred).
24Ibid., 34.
25Ibid., 73.
27Ibid., 1012.
28Ibid., 1215.
29Alina Selyukh, As We Leave More Digital Tracks, Amazon Echo Factors in Murder Investigation,
Morning Edition, National Public Radio, December28, 2016, www.npr.org /sections/alltechconsidered
/2016/12/28/507230487/as-we-leave-more -digital-tracks-amazon-echo-factors-in-murder-investigation.
Amazon initially resisted the police order, and the case is still pending. But the point remains: encryption
is no technological barrier to law enforcement access to the devices data. Amazon has since given up
thedata.
30Amanda Watts, Cops Use Murdered Womans Fitbit to Charge Her Husband, CNN, April26, 2017, www
.cnn.com/2017/0 4/25/us/f itbit-womans-death-investigation-trnd/index.html.
32See Bruce Schneier, Kathleen Seidel, and Saranya Vijayakumar, A Worldwide Survey of Encryption
Products, Berkman Center Research Publication No.2016-2, February11, 2016, http://papers.ssrn.com
/sol3/papers.cfm?abstract _id=2731160.
33Tor is the most well-known anonymity software. It claims to prevent...people from learning your
location or browsing habits. See Tor Project, https://w ww.torproject.org.
34See Brief for the United States, 48, In re Warrant to Search a Certain EMail Account Controlled and
Maintained by Microsoft Corp.
35See Thorin Klosowski, What Is Tor and Should I Use It? Lifehacker, February21, 2014, https://lifehacker
.com/what-is-tor-and-should-i-use-it-1527891029.
36Cade Metz, Forget Apple vs. the FBI: WhatsApp Just Switched on Encryption for a Billion People,
Wired, April5, 2016, http://bit.ly/2drJYut.
37David Meyer, Brazil Arrests Senior Facebook Exec over WhatsApp Aid in Drug Case, Fortune, March1,
2016, http://for.tn/216L67V.
38The Apple vs. FBI dispute seems like good evidence of just that, since the government was able, with the
help of a private firm, to get into the device without Apples help.
39See JohnS. Odell and Dustin Tingley, Negotiating Agreements in International Relations, in
Negotiating Agreements in Politics, task force report of the American Political Science Association, 2013, 161,
summarizing the literature on issue linkage.
The publisher has made this work available under a Creative Commons Attribution-NoDerivs license 3.0. To view a copy
of this license, visit http://creativecommons.org/licenses/by-nd/3.0.
Hoover Institution Press assumes no responsibility for the persistence or accuracy of URLs for external or third-party
Internet websites referred to in this publication, and does not guarantee that any content on such websites is, or will
remain, accurate or appropriate.
Copyright 2017 by the Board of Trustees of the Leland Stanford Junior University
Preferred citation for this publication is Andrew Keane Woods, Encryption Substitutes, Hoover Working Group on
National Security, Technology, and Law, Aegis Series Paper No. 1705 (July 17, 2017), available at https://lawfareblog.com
/encryption-substitutes.