Is data speech?

• Author: Bambauer, Jane
• Position: Introduction into III. The Right to Create Knowledge, p. 57-87

INTRODUCTION I. WHAT Is "DATA," AND WHAT IS "SPEECH"? A. Data B. Speech II. DATA IN FIRST AMENDMENT PRECEDENT A. Existing Data 1. Gleanings from privacy law 2. Gleanings from commercial speech regulations 3. Gleanings from copyright law B. Data Creation as Nonexpressive Conduct C. Data Creation as Expressive Conduct III. THE RIGHT TO CREATE KNOWLEDGE A. The Negative Right to Create Knowledge B. Data and First Amendment Objectives 1. Marketplace of ideas and public good theories 2. Deliberative democracy and a check on state power 3. Self-Determination C. The Level of Scrutiny IV. COPING WITH SCRUTINY A. Concern 1: Scrutiny Will Kill Privacy and Other Good Things B. Concern 2: If Data Is Speech, Regulators Are Hamstrung C. Concern 3: Scrutiny Is Subjective at Best, Political at Worst CONCLUSION [I]f there is any principle of the Constitution that more imperatively calls for attachment than any other it is the principle of free thought....

--Justice Oliver Wendell Holmes, Jr. (1)

INTRODUCTION

When does factual information become speech? The appealing, simple answers reside at the extremes.

Perhaps information is always speech. After all, information communicates, educates, and persuades. A single plain fact can do more to change minds and alter debates than a thousand opinions. But this rule goes too far. There are many times that an event will leave a mark that has the potential to retell its story. A car may careen into a barrier and leave a streak of paint. Long after the car is towed, the streak states, in a way, when and where the crash occurred, how fast the car was traveling during impact, and what color the car was. The streak of paint can be received and interpreted by a human to create knowledge. (2) But if a city repainted the barrier, we would not interpret this as a decision related to speech. Likewise a crack in the sidewalk might tell the story of a frost, but a municipal ordinance requiring property owners to maintain their sidewalks would not be an act of censorship. Every cell contains DNA, the body's ultimate archive of information, and yet the proper disposal of used syringes does not, and should not, implicate First Amendment scrutiny.

Since data is expressed in alphanumeric symbols, it certainly looks a lot more like traditional speech than a crack in the sidewalk. (3) However, conceptually it is sometimes not so far off. When data is the byproduct of other events and services--transactions between a home computer and a website's server, or between a cell phone and a cell tower--these records are no different from other unanticipated marks created by the bustle of life. They have no intended author, and no intended audience. They are mere footprints.

If information isn't always speech, then perhaps it is never speech. After all, the quintessential First Amendment litigant expresses some unpopular idea or opinion. Maybe the domain of the First Amendment is opinion, and other products of human subjectivity. This rule runs into unsalvageable problems even more quickly than the last. Some of the most important modern speech cases concerned the unadorned reporting of raw information. The Supreme Court protected the New York Times's distribution of the Pentagon Papers from prior restraint by the federal government on the basis of its First Amendment rights to do so. (4) The New York Times had received a leak of the Pentagon Papers--a report created by the Department of Defense to document the U.S. involvement in the Vietnam War. Many details in the report conflicted with the information that the Johnson Administration had told the public, so the Pentagon Papers were significant not only for the details within them, but also for the inevitable inference that President Johnson had deceived Americans. Justices Black and Douglas had no trouble recognizing the speech interests in raw facts--the fact of the Pentagon Papers' existence and the facts that were reported in them. (5) It was not necessary for the New York Times to editorialize in order to access the First Amendment's protection. (6)

A distinction can be made between statements of fact that are observed and written by a human and those that are collected mechanically, and it might be tempting to draw the First Amendment line between the two. But the distinction is untenable. Suppose Congress were to pass a law mandating the destruction of mechanically captured climate science data and the discontinuation of the collection of core samples. It is implausible to think that a court would not employ some form of First Amendment scrutiny. (7)

The two absolute positions, that data always is speech or never is, simply fail to describe an understanding of the First Amendment that society could tolerate. If neither of the poles can be correct, the truth is somewhere in the messy middle.

Delineating exactly when information receives speech protection, and why, would not have been worth the energy before the era of mass computing, but the building momentum of data analytics and the competing interests of personal privacy have turned this once-arcane question into a matter of great significance and urgency.

Privacy advocates naturally turn to lawmakers to combat the effects of increased data collection and cheap data storage. They are not alone. As smart phones and other recording devices become ubiquitous, corporations have come to the well, too, pressing legislators to create or strengthen laws that protect their interests in secrecy. Of course, these political forces have little overlap in support. Consumer groups press for privacy laws while corporations lobby for trade secret laws and various industry-specific protections (for example, so-called "ag-gag" laws for animal agriculture companies (8) or criminal statutes outlawing card-counting devices for casinos (9)), but the motivation for these disparate efforts is the same: to preserve the information status quo.

These efforts clash with the First Amendment.

This Article contends that the freedom of speech carries an implicit right to create knowledge. When the government deliberately interferes with an individual's effort to learn something new, that suppression of disfavored knowledge is presumptively illegitimate and must withstand judicial scrutiny.

The application of this rule to information law is straightforward. Data is not automatically speech in every context; as the opening hypotheticals illustrate, data can be generated without any expectation to be reviewed and interpreted. But asking whether all data should be treated as speech misses the point: any time the state regulates information precisely because it informs people, the regulation rouses the First Amendment.

The right to create knowledge reinforces American commitments to autonomy and intellectual curiosity. It at once transcends and supports "speaker" and "listener" rights by protecting observation and thought--the very things that make speaking and listening so valuable. It deserves to be celebrated. Nevertheless, I will proceed with caution because the implications of the right are quite far reaching and profound.

For one thing, as I describe in Part II, the right to create knowledge explodes the classic distinction between information and information gathering, which has long been used by courts to define a boundary between protected speech and unprotected conduct. Mechanically captured recordings of conversations and images are often treated as information-gathering conduct, and have avoided First Amendment scrutiny on that basis. A large and growing body of wiretap statutes and information security laws rely heavily on the distinction. But a rigorous review of the cases distinguishing information from information gathering reveals that the reasoning is flawed. The cases treat a journalist who observes an event differently depending on whether she records the incident by taking meticulous notes using pad and paper (protected) or by using the video function on her iPhone (unprotected). The line arbitrarily favors old technologies (memory, pens, even word processors) over new ones.

Recent, better-reasoned case law has already begun to correct the inconsistency. (10) If the dissemination of mechanical recordings receives First Amendment protection (which it does), (11) then the creation of those same recordings must have First Amendment significance, too.

Many readers will feel uneasy about the breadth of a First Amendment that interferes with the modern attempts to limit the collection of personal data. The proposals in this Article can be used to challenge the constitutionality of many popular privacy statutes (for example, the Health Insurance Portability and Accountability Act of 1996 (12)). They throw into doubt the Federal Trade Commission's varied privacy initiatives and frustrate the White House's proposed "Consumer Privacy Bill of Rights." (13) All-party-consent wiretap statutes would have to be justified by important or compelling state interests, and the various "Do Not Track" bills, as well as the pending and future attempts to limit the collection and disclosure of location-tracking data, would be subjected to constitutional scrutiny, too. It would be quite natural to have reservations about a theory of constitutional law that has such sweeping effects.

Anticipating the disruption that constitutional scrutiny would cause, privacy scholars have built a consensus conveniently placing data outside the scope of the First Amendment's protection. Neil Richards has argued that personal data is not speech because it is more commodity than expression. (14) Ashutosh Bhagwat recognizes that there is some basis for treating data as speech, but because the consequences are "dramatic and troubling," he advises jurists to evade scrutiny by characterizing data as less valuable, purely private speech. (15) And Tim Wu concludes that data cannot be speech because computers do not have constitutional rights. (16)

These...