SOFTWARE AS "SPEECH": NOTES AND QUESTIONS
1. In 1978, it was unclear whether "software" or computer programs
would qualify for copyright protection. In that year the Commission
on New Technological Uses of Copyrighted Works, called the "CONTU Commission"
for short, issued its Final Report, recommending copyright protection
for computer programs.
In a famous dissent, Commissioner John Hersey disagreed with the Commission's
recommendation. He argued that computer programs were just parts
of machines and should not be given equal dignity with works of literature
and music. As parts of machines, he argued, computer programs are
not designed to communicate with people and therefore are not proper objects
of copyright protection.
The majority of the CONTU Commissioners disagreed with Commissioner Hersey
and supported the Commission's recommendation. In 1980, Congress
implemented that recommendation by adopting the Software Copyright Amendments
of 1980, Pub. L. No. 96-517, § 10, 94 Stat. 3015, 3028-3029 (Dec.
12, 1980), codified at 17 U.S.C. 101, 117. The amendments made clear
that computer programs are proper subjects of copyright protection by,
among other things, adding a definition of "computer program" to Section
101 of the Copyright Act of 1976, 17 U.S.C. § 101. ("A
‘computer program' is a set of statements or instructions to be used directly
or indirectly in a computer in order to bring about a certain result.")
2. Even before its amendment, the Copyright Act of 1976 suggested
that direct communication to people is not a requirement of copyright
protection. The statute defined and still defines "copies" as "material
objects . . . in which a work is fixed by any method now known or later
developed, and from which the work can be perceived, reproduced, or otherwise
communicated, either directly or with the aid of a machine or device."
17 U.S.C. 101 (definition of "copies").
The statute's definition of "literary works" also seemed to accommodate
software. It read and still reads as follows:
"‘Literary works' are works, other than audiovisual works, expressed in
words, numbers, or other verbal or numerical symbols or indicia, regardless
of the nature of the material objects, such as books, periodicals manuscripts,
phonorecords, film, tapes, disks, or cards, in which they are embodied."
17 U.S.C. 101 (definition of "literary works").
Does either of these definitions suggest that copyrighted works must be
immediately intelligible to or legible by people? Do any words in
these definitions suggest the contrary? Does either definition appear
to exclude computer programs?
3. As a result of these and other statutory provisions, plus the
1980 amendments, virtually every court that has addressed the issue has
rejected Commissioner Hersey's argument and has held that computer programs
are proper objects of copyright protection whether or not they are directly
intelligible. In other words, software need not "speak" to people
in order to merit copyright protection.
The Corley decision, however, raises a related but very different
question: if software is not just a part of a machine as Commissioner
Hersey argued, is it then "speech" eligible for First Amendment protection?
Does the history of copyright protection for software necessarily
imply that software is "speech"?
4. A general answer may be elusive, in part because the word "software"
is so expansive. Some use the term to refer to any digitized information,
including digitized text, photos, film, movies and music. Surely
a politician's oration is none the less "speech" although it is converted
into and stored in digital form.
Others, however, use the term "software" more restrictively, as meaning
"computer programs" and related utilitarian things. Yet even in this more
restrictive meaning software can take several forms. "Source code"
is a form of computer programs written in higher-level languages (such
as FORTRAN, COBOL, BASIC, C or C++). It is directly intelligible
to trained programmers and indeed is intended to be so. Yet it can
also be "read" by machines using other computer programs, called "compilers"
or "interpreters," which convert it into binary form for execution by
machines. Thus "source code" is both human-readable and machine-readable.
A very simple example of source code is the following program, in
the BASIC language, which adds two and two and prints out the result:
A = 2
B = 2
C = A + B
5. In contrast to source code, "binary" or "object" code (also known
as "machine language") is software as machines produce and use it. It
consists of large, undifferentiated blocks of binary digits, i.e., ones
and zeroes, without any demarcation of where one line or command ends and
the next one begins. For example, a block of binary code (not the
compiled version of the BASIC program above!) might look like this:
Although trained programmers can, in theory, read object
or binary code, doing so is a bit like deciphering the Rosetta stone.
(Unless one knowsor until one determines by trial and errorhow
long the binary "words" or instructions are, one does not even
know where to divide the block into separate instructions to begin the
translation.) In addition, there is no one-to-one correspondence
between the instructions in binary code and the instructions in source
code, because many of the instructions in source code must be broken down
into smaller "baby steps" before being translated into machine
language. No one offered a more efficient alternative would ever
voluntraily attempt to read binary code by eye.
As the foregoing examples show, source code and binary code look vastly
different when printed out. Source code is immediately intelligible
to trained programmers; binary code can be deciphered only with great
difficulty and labor, or by the use of machines. Yet both contain
virtually the same information and expression.
Just as Tolstoy's War and Peace may be unintelligible to an American
reader in its native Russian, the Russian version contains the same "pattern"
of expression as the English translation. So undecipherable binary
code compiled from source code may seem just another side of the same
coin. In other words, if a politician's address is still "speech"
when converted into binary form, why not also the source code of a computer
program when converted into binary or object code?
6. There are many difficulties with the conclusion that binary program
code is equivalent to translated or digitized text. Among other
things, digitized text is a direct character-by-character translation
of individual letters and punctuation into their numerical equivalents,
which translation is easily reversible. (For example, the capital
letter "B" is represented in computers as the decimal number
"66," the equivalent hexadecimal number "42," or the
equivalent binary number "1000010.") In constrast, translating
source code into binary code requires a complex process of: (1) distilling
the conceptual mathematical or storage operations represented symbolically
by source code into "baby steps" that can be expressed in lower-level
languages; (2) writing the commands to execute those "baby steps"
in the lower-level languages; and (3) translating the resulting lower-level
commands into their binary equivalents. The process is far more
complex than character-by-character translation. Furthermore, it
is prone to error and generally irreversible (because attempts to reverse
the process do not necessary produce the same source code, but rather
different source code with equivalent mathematical effect). For
a more detailed description of the process of converting source to binary
program code and its practical effects and consequences, see Jay Dratler,
Jr., Cyberlaw: Intellectual Property in the Digital Millennium
§ 2.10[b][i] (Law Journal Press 2000, updated semiannually).
Another difficulty with the conclusion that binary program code is just
like digitized speech is that, unlike digitized speech, binary program
code is virtually never intended to communicate with humans. Rather,
it is intended to operate machines directly. Most computer software
vendors market computer programs only in the form of binary executable
code precisely because they are difficult to understand and modify in
that form. Several courts have suggested that the difficulty of
"reverse engineering" binary executable code is sufficient to
provide a measure of trade-secret protection for algorithms and programming
techniques buried in the code. Moreover, if hackers like Corley
use binary code to distribute their decryption tools, they probably do
so partly in order to make it more difficult for the authorities and the
vendors of encryption devices to decipher the operation of their computer
programs and thereby defeat them or trace their origins. In other
words, people most often distribute binary program code, as distinguished
from source code, precisely because they do not wish to communicate.
How should these considerations affect the First-Amendment protection
given binary executable code?
7. For copyright purposes, focusing on the medium or form
of recording expression is perilous, because Congress wrote the Copyright
Act of 1976 to make copyright protection independent of the media and
technologies used. This "technology-independent" approach was
one of the chief advantages of the 1976 Act over previous versions of
copyright law. The statute implements this approach by referring
repeatedly to media and recording techniques "now known or later developed"
in many of the statutory definitions. See, e.g., 17 U.S.C. §
101 (definitions of "copies" and "phonorecords"),
17 U.S.C. § 102(a) (fundamental provision for copyright protection).
Congress did not want to have to amend the copyright statute every time
a new recording technology emerged. If it had, over nine new copyright
statutes would have been necessary during the twentieth century for music
alone, for there were at least nine generations of music recording technology
(wax cylinders, plastic disks, wire, tape reels, tape cassettes, magnetic
disks, CDs, RW-CD-ROMs and MP3 files).
Thus, copyright protection does not depend upon the mode or medium or
recording, i.e., upon whether the "software" is recorded as source or
8. But does it follow that all recorded expression, including all
"software," merits equal protection as "speech" under the First Amendment?
All of the following might be categorized as "software," but do
all deserve the same level of First Amendment protection?
a. A digitized database contained the freezing
and boiling points, in degrees centigrade, of all known compounds of
the element calcium.
b. A digitized audiovisual clip of The Reverend Martin Luther King's
"I Have a Dream" speech.
c. The "public key" for an encryption code, i.e., a long binary
number, published on the Web, that can be used with a corresponding "private
key" to decrypt encrypted matter using a particular computer program.
d. The "private key" corresponding to the "public key" in
point (c) above.
e. A computer operating system, i.e., a basic computer program
that moves data between a computer's central processing unit (usually
a microprocessor) and "peripheral" devices like the keyboard, disk drive,
display screen, loudspeakers, etc.
f. A computer program that seeks to "break" encryption codes
without using the published or private keys.
g. A computer program that analyzes explosions of nuclear
weapons in order to determine how to design and maintain them.
9. In Corley, the Second Circuit notes the case of Junger
v. Daley, in which the Sixth Circuit upheld federal regulations prohibiting
the exportation of certain decryption software against a First Amendment
challenge. Was that case correctly decided? Can you distinguish
it from United States v .Cohen (the "F*** the Draft" case) discussed
in Reno? On which side of the speech/conduct line does each
10. Can the idea of software as "speech" be stretched too far? What
about a computer program used to design and maintain nuclear weapons (see
point 8(g) above). A government scientist was suspected of, but never
charged with, supplying such programs, developed at great expense in government
laboratories, to agents of a foreign power. If he had been charged
with doing so, could he have defended on the ground that the software is
"speech" and that any prohibition on its export or dissemination is unconstitutional?
If United States government agents "caught" the software in the Web
on its way to foreign agents, would that be a prohibited "prior restraint"
on the dissemination of speech?
11. If you see a distinction between software for encryption and that
for designing nuclear weapons, on the one hand, and digitized music or more
pedestrian computer programs on the other, where and how do you draw the
line? On which side of the line does DeCSS fall? Can you articulate
a principled basis for your distinctions? Does it require that some
kinds of "speech," or at least some kinds of software, be given less protection
Should regulation of software be subject to the same kind of "strict
scrutiny" that applies to political speech? Should the answer depend
upon the form of the software, i.e., whether it constitutes readable source
code or undecipherable (except with great difficulty) binary program code
intended to operate machines? Or should some regulation of software
be treated as a "time, place and manner" restriction that does not impact
expressive content? Should software be viewed as just another variety
of "commercial speech," which some believe deserves less protection than
political speech, or should it be viewed as sui generis?
Back to Top
12. Are computer programs, properly defined, technology or speech?
Does that depend upon their form, i.e., as source or binary code?
Was the Second Circuit too solicitous of the alleged expressive
aspects of computer programs, even in the form of binary program code
intended for no purpose other than to operate machines? Or
is the court's fastidious analysis necessary to protect our fundamental
rights in the digital age? Is object code or binary program code
a communications medium or a functional part of machines?