10 March 1999
Source: Peter Junger
See related files: http://jya.com/pdj.htm

For full case history see Peter Junger's Web site: http://samsara.LAW.CWRU.Edu/comp_law/jvd/

__________________________

Case Number 98-4045

__________________________

In The United States Court of Appeals

For The Sixth Circuit

__________________________

Peter D. Junger,

Plaintiff-Appellant

--vs--

William Daley, United States Secretary of Commerce, et al.,
Defendants-Appellees.

____________________________________________________

Proof Brief of Appellant Peter D. Junger

____________________________________________________

Reserve Remove Page for Corporate Disclosure Statement

TABLE OF CONTENTS

Statement of Corporate Disclosure ii

Table of Contents iii

Table of Cases & Authorities vii

Jurisdictional Statement 1

Statement in Support of Oral Argument 1

Statement of Issues 1

Statement of the Case 2

Statement of Facts 3

Standard of Review 16

Summary of the Argument 17

Law & Argument 19

I. The Regulations Impermissibly Restrict the Publication of Expression Protected by the First Amendment 23

A. The Publication of Source Code Is Expression Essential to The Exposition of Many Technical and Scientific Idea 24

1. Like other statements expressed in language, writings in source code should be considered a protected form of expression within the meaning of the First Amendment 26

a. Texts expressed in source code are linguistic creations in the same sense as more traditional texts 26

b. What is special about computer programming languages 29

2. Computer Programs Written in Source Code Are an Essential Part of the Exposition of Ideas 33

B. The District Court's Analysis Is Internally Inconsistent, And Does Not Support Its Conclusion That Source Code is Unprotected Expression. 35

1. The District Court's Distinction Between Inherently Functional and Inherently Expressive Software Is Inherently Unsound 36

a. The district court misapplied Chaplinsky v. New Hampshire 37

b. The district court used an flawed analysis, mischaracterizing the expressive activity at issue 37 issue

2. The District Court Misunderstood the Communicative Aspects of Publishing Programs in Source Code Form 40

3. The District Court Drew a Distinction Between Print and Electronic Form That Makes No First Amendment Difference 42

C. The Regulations Restrict Expressive Activity on the Internet 44

II. The Regulations Create a Licensing Scheme That Violates the First Amendment 49

A. The Licensing Scheme Is An Unconstitutional Prior Restraint on Free Expression 50

1. The scheme is a prior restraint on publication 50

2. A facial challenge under Lakewood is proper because the scheme vests excessive discretion in government Officials 52

3. The scheme is unconstitutional because it fails to provide procedural safeguards and because it fails to meet the burden imposed on prior restraints under United States v. New York Times 57

B. The Licensing Scheme Is Content-Based And Cannot Survive Strict Scrutiny 59

1. The regulations, on their face, discriminate on the basis of both subject matter and content and therefore the government's purpose is not controlling 60

2. The regulations cannot survive strict scrutiny 63

Conclusion 65

TABLE OF CASES & OTHER AUTHORITIES

Cases

Abrams v. United States, 250 U.S. 616 (1919) 24

ACLU v. Reno, 929 F.Supp. 824 (E.D. Pa. 1996) 44, 45

Arkansas Writer's Project v. Raglan, 481 U.S. 221 (1987) 60

Asian-American Business Group v. City of Pomona, 716 F.Supp. 1328 (C.D. Cal. 1989) 42

Bernstein v. United States Department of State, ("Bernstein I") 922 F.Supp. 1426 (N.D.Cal. 1996) 27, 29

Bernstein v. United States Dep't of State (Bernstein III), 974 F.Supp. 1288 (N.D. Calif. 1997) 20, 34, 55

Board of Education of the Island Trees Union Free School Dist. v. Pico, 457 U.S. 853 (1982) 24, 48

Bullfrog Films, Inc. v. Wick, 847 F.2d 502 (9th Cir. 1988) 48

Carey v. Brown, 447 U.S. 455 (1980) 21

Chaplinsky v. New Hampshire, 315 U.S. 568 (1942) 36, 37

Cohen v. California, 403 U.S. 15 (1971) 31

Consolidated Edison Co. v. Public Service Comm'n of New York, 447 U.S. 530 (1980) 61

Dambrot v. Central Michigan University, 55 F.3d 1177 (6th Cir. 1995) 24, 28, 41

East Brooks Books, Inc. v. City of Memphis, 48 F.3d 220 (6th Cir. 1995) 57, 58

Freedman v. Maryland, 380 U.S. 51 (1965) 21, 56, 57, 58

Forsyth County v. The Nationalist Movement, 505 U.S. 123 (1992) 54

FW/PBS, Inc. v. City of Dallas, 493 U.S. 215 (1990) 55, 57, 58

Hurley v. Irish-American Gay, Lesbian & Bisexual Group, 515 U.S. 557 (1995) 24

Junger v. Daley, 8 F.Supp.2d 708 (N.D.Ohio 1998) passim

Lakewood v. Plain Dealer Publish Co., 486 U.S. 750 (1988) 39, 40, 52, 53, 54

Montgomery v. Carr, 101 F.3d 1117 (6th Cir. 1996) 16

Meyer v. Nebraska, 262 U.S. 390 (1923) 42

New York Times Co. v. United States, 403 U.S. 713 (1971) 56, 57, 59

Nebraska Press Assoc. v. Stuart, 427 U.S. 539 (1976) 57

Pacific Gas & Electric Co. v. Pacific Utilities Comm'n of California, 475 U.S. 1 (1986) 48

Procter & Gamble Co. v. Bankers Trust Co., 78 F.3d 219 (6th Cir. 1996) 16, 57

R.A.V. v. St. Paul, 505 U.S. 377 (1992) 63

Reid v. Covert, 354 U.S. 1 (1957) 48

Reno v. ACLU, 117 S.Ct. 2329 (1997) 20, 45, 62, 63

Roulette v. City of Seattle, 850 F.Supp. 1442 (W.D. Wash. 1994) 52

Roulette v. City of Seattle, 97 F.3d 300 (9th Cir. 1996) 51, 52

Sable Communications of California, Inc. v. FCC, 492 U.S. 115 (1989) 64

Sandul v. Larion, 119 F.3d 1250 (6th Cir.) 25

Simon & Schuster Inc. v. New York State Crime Victims Board, 502 U.S. 105 (1991) 21

Southeastern Promotions, Ltd. v. Conrad, 420 U.S. 546 (1975) 50, 56

Spence v. State of Washington, 418 U.S. 405 (1974) 25

Teitel Film Corp. v. Cusak, 390 U.S. 139 (1968) 58

Texas v. Johnson, 491 U.S. 397 (1989) 23

Turner Broadcasting System, Inc. v. Federal Communications Comm'n, 114 S. Ct. 2445 (1994) 60, 61

Vance v. Universal Amusement Co., 445 U.S. 308 (1980) 55, 56

Ward v. Rock Against Racism, 491 U.S. 781(1989) 50, 60

Yniguez v. Arizonans for Official English, 69 F.3d 920 (9th Cir. 1995) 25

Yu Cong Eng v. Trinidad, 271 U.S. 500 (1926) 42

Statutes

28 U.S.C. § 1291 1

50 U.S.C. § 1701 et seq 3

50 U.S.C.A. § 2401 et seq 6

50 U.S.C. App. § 2415(4) 7

Regulations: Code of Federal Regulations

15 C.F.R. § 732.3(d) 46

15 C.F.R. § 732.3(d) 6

15 C.F.R. § 732.3(d)(5) 6

15 C.F.R. § 734.2(b)(9)(ii) 9, 20, 27, 46

15 C.F.R. § 734.2(b)(1) 9

15 C.F.R. § 734.3(b) 8

15 C.F.R. § 734.3(b)(2) 8, 9

15 C.F.R. § 734.3(b)(3) 8, 9, 46, 61

15 C.F.R. § 734.3(b)(3)(ii) 8

15 C.F.R. § 734.3(b)(3)(iii) 8

15 C.F.R. § 734.4 8

15 C.F.R. § 734.7 46

15 C.F.R. § 734.7(c) 8, 46

15 C.F.R. § 734.8 8, 46, 61

15 C.F.R. § 734.8(a) 8, 46

15 C.F.R. § 734.9 8, 46, 61

15 C.F.R. § 738 6

15 C.F.R. § 742.15(a) 9

15 C.F.R. § 742.15(b) 10, 51

15 C.F.R. § 742.15(b)(1)(i) 64

15 C.F.R. § 748(3) 10

15 C.F.R. § 748.3(a) 10

15 C.F.R. § 748.3(b)-(c) 10

15 C.F.R. § 750.4(a)(1) 58

15 C.F.R. 756.2(c)(1) 58

15 C.F.R. § 764.3 10, 49

15 C.F.R. § 764.3(a)(1) 10

15 C.F.R. § 764.3(b)(1) 10

15 C.F.R. § 764.3(b)(2) 10

15 C.F.R. § 768(1)(b) 8

15 C.F.R. § Part 742(a)(2)(ii),(b)(5)(E)(3), Supplement No. 6 62

15 C.F.R. Part 748 10, 49

15 C.F.R. Part 750 10, 49

15 C.F.R. Part 762 10, 49

15 C.F.R. Part 772 6, 7, 8, 38, 40

15 C.F.R. Part 774 6, 8

15 C.F.R. 774.1 6

22 C.F.R. 120 et seq 2

Executive Orders of the President

Exec. Order No. 13026 6, 59

Uncodified Federal Regulatory Law

61 Fed. Reg. 68575 44

61 Fed. Reg. 68579 8

63 Fed. Reg. 72156-58 10

63 Fed. Reg. 72162 62

63 Fed. Reg. 72164-72165. 62

Constitutional Provisions

U.S. CONST. amend. I passim

Secondary Authorities

Abelson and Sussman, Structure and Interpretation of Computer Programs, preface at xv (MIT Press 1985) 30

Emerson, Thomas, The Doctrine of Prior Restraint, 20 Law & Contemp. Probs. 648 (1955) 50

Knuth, Donald, The Art of Computer Programming: Fundamental Algorithms (1968) 32

Lewis, David, Languages and Language, PHILOSOPHICAL PAPERS Vol. I at 164-66 (1983) 28

National Research Council, Cryptography's Role in Securing the Information Society (1996) 4

McCarthy, John, Recursive Functions of Symbolic Expressions and their Computation by Machine, Part I, 3 COMMUNICATIONS OF THE ACM 185 (1960) 30

Smith, Don, Computerizing Computer Science, COMMUNICATIONS OF THE ACM, September 1, 1998 32

Stevens, John Paul, The Freedom of Speech, 102 Yale L.J. 1293 (1993) 23

Statement of Jurisdiction

The Court has jurisdiction over this appeal pursuant to 28 U.S.C. 1291. The District Court entered summary judgment in case No. 96 CV 1723 against Professor Junger on July 2, 1998. Under the Federal Rules of Appellate Procedure, a notice of appeal may be filed within sixty (60) days of the entry of judgment if the United States, its officers or agents are party to the suit. Fed. R. App. P. 4(a)(1). Professor Junger timely filed his notice of appeal on August 27, 1998. (R. 95). This case is now before this Court docketed as CA No. 98-4045.

Request for Oral Argument

This appeal presents a question of first impression for this Circuit, concerning the status of source code under the First Amendment, and the interplay of the First Amendment and United States export laws. Oral argument is required in order to clearly elucidate the regulatory scheme, and fully address the relevant First Amendment issues.

Statement of the Issues

This is a constitutional challenge to provisions of the Export Administration Regulations ("EAR"), 15 C.F.R. Parts 730-74, that prohibit the "export" of encryption software without prepublication review and a government license. This appeal is limited to the question of whether restrictions on the publication of encryption software on the Internet violate the First Amendment.

Specifically, the issues are (1) whether the District Court erred by holding that the publication of encryption source code on the Internet is not protected by the First Amendment, and (2) whether the District Court erred in holding that the licensing scheme imposed on the publication of encryption source code by the EAR survives First Amendment scrutiny.

Statement of the Case

This appeal by Plaintiff-Appellant Peter D. Junger, professor of law at Case Western Reserve University, challenges on facial and as-applied grounds U.S. export restrictions that make it a crime to publish encryption software on the Internet without a government license. Professor Junger appeals the July 2, 1998, decision of Hon. James S. Gwin, United States District Judge for the Northern District of Ohio, granting summary judgment for the defendants. (R. 93, 94)

The action was initially filed on August 7, 1996, and assigned to the Hon. Donald C. Nugent. On September 15, 1997, Professor Junger filed a Supplemental and Amended Complaint after the regulations controlling encryption (previously "cryptography") were transferred from the International Traffic in Arms Regulations ("ITAR"), 22 C.F.R. 120 et seq., to the EAR. (R. 75) Professor Junger and the government filed motions for summary judgment before Judge Nugent (R. 58, 62), who then assigned the case to Judge Gwin on December 19, 1997.

Judge Gwin granted the defendants' motion for summary judgment on the grounds that (1) the "export" of encryption software, which includes the publication of encryption software on the Internet, is not protected by the First Amendment; (2) the EAR controls on encryption software survive intermediate scrutiny; and (3) the EAR's implementing legislation, the International Emergency Powers Act ("IEEPA"), 50 U.S.C. 1701 et seq., does not violate the constitutional doctrine of separation of powers. Professor Junger appeals the grant of summary judgment on the first two grounds.

Statement of Facts

A. Cryptography and Encryption

"Cryptography" is derived from Greek words meaning "secret writing." National Research Council, Cryptography's Role in Securing the Information Society, 52 (1996). Cryptography concerns the processes of encryption, which involves converting original, "human-readable" text into text that a sender and receiver want to remain unintelligible to third parties (called "ciphertext"). (See R. 87 Stipulations of Fact ¶ 2, Apx. __.) Or, as one of the declarants describes it, "the aim of encryption is to turn an otherwise intelligible message into gibberish." (R. 20 Schneier Decl. ¶ 2, Apx. __.) Decryption is the opposite process, turning ciphertext back into plaintext. (See R. 87 Stipulations of Fact 2, Apx. __.)

Most cryptography today uses an algorithm--a mathematical transformation from plaintext to ciphertext or from ciphertext to plaintext or both--and a "key" -- "specific information (analogous to a password). (See id. 4; R. 20 Schneier Decl. ¶ 4, Apx. __.) In general, the security of a message depends on the algorithm and the strength of the key. (See id. ¶¶ 11-12; R. 87 Stipulations of Fact ¶ 4, Apx. __.)

"Modern cryptography is a mathematical science, and many cryptographers have Ph.D.'s in mathematics or computer science." (R. 20 Schneier Decl. ¶ 18, Apx. __.) Cryptography can be thought of as "the science of making and breaking algorithms." (Id. ¶ 28.) Algorithms that are not released publicly for other cryptographers to use are generally not trusted. (Id. ¶ 31.) This is why it is important for cryptographers to discover new algorithms to release them to their peers and the public, and it is how cryptographers develop their reputations. See id.

"Over 99%" of all uses of cryptography involve something other than keeping military secrets. (Id. ¶ 36.) Cryptography is used in many civil applications, such as protecting banking information, office buildings, lottery terminals and communication and data sent over the Internet. See id.

Beside its use by governments and business, cryptography is used by individuals and private groups, including human rights organizations, such as Human Rights Watch and Amnesty International. (See R. 69 Banisar Decl. ¶ 3.) Such groups, and political dissidents, reasonably fear repressive governments, and could not communicate freely over the Internet without cryptography. (See id. ¶ 6). "As a result, without encryption, some speech, pure political speech, would not occur over the Internet." Id. In the absence of cryptography, information sent via a computer is not secure, and may be viewed by persons other than the intended recipient. (See R. 87 Stipulations of Fact ¶ 2, Apx. __.)

B. Regulatory Background

The EAR is a comprehensive licensing scheme designed to control the export of non-military technology, software and commodities. It was originally promulgated under the Export Administration Act of 1979 ("EAA"), 50 U.S.C.A. 2401 et seq. When the EAA lapsed on August 24, 1994, the President extended it under IEEPA.

On November 15, 1996, the President issued an executive order in conjunction with a memorandum, which transferred export jurisdiction over nonmilitary encryption from the State Department (under ITAR) to the Commerce Department Bureau of Export Administration ("BXA")(under EAR). See Exec. Order No. 13026, 1996 WL 666567.

The EAR is structured around the Commodity Control List ("CCL"), which lists items subject to export control. See 15 C.F.R. Part 774; see id. § 738. Each item on the CCL is assigned an "Export Control Classification Number" ("ECCN"). See id. § 732.3(d). Encryption software, which is defined under the EAR as "computer programs that provide capability of encryption functions or confidentiality of information,"see id. Part 772, is classified as a Encryption Item ("EI") under ECCN 5D002, see id. Part 774. License requirements depend largely on an item's ECCN number.⁽¹⁾

Under the EAA, software was included within the definition of "technology," which was defined as "information and know-how . . . that can be used to design, produce, manufacture, utilize, or reconstruct goods, including computer software and technical data, but not the goods themselves." 50 U.S.C. App. § 2415(4) (emphasis added). Following the EAA, the EAR treats all software, except encryption software, like technology. By contrast, encryption software, in both source code⁽²⁾ and object code⁽³⁾ forms, is treated in the same manner as encryption hardware, and controlled for EI reasons. See 15 C.F.R. Part 774, Note following 5D002; 61 Fed. Reg. 68579 (to be codified at 15 C.F.R. § 738(d)(2)(I)(A)).

Since encryption software is not treated like other software, export control exemptions for software and technology that are "publicly available"do not apply.⁽⁴⁾ Public availability is defined as "generally accessible to the interested public in any form" and "already published or will be published; arising during, or result from fundamental research; are educational; or are included in certain patent applications." See id. Part 772.

Thus, the general public availability exemption (15 C.F.R. § 734.3(b)), the exemption for scientific research (15 C.F.R. § 734.3(b)(3)(ii), § 734.8), and the exemption for educational information (15 C.F.R. § 734.3(b)(3)(iii), § 734.9) do not apply to encryption software, either in source code or object code.

While encryption software in printed form is not subject to the EAR if publicly available, all publications of encryption software in electronic form, including computer diskettes, CD-ROMs and on the Internet, are subject to export control. 15 C.F.R. § 734.3(b)(2) and Note following § 734.(b)(3). Thus, encryption software in electronic form is always subject to export control, even if published for purely academic or scientific reasons.

The EAR defines "export" as the "actual shipment or transmission of items subject to the EAR out of the United States,"Id. § 734.2(b)(1), and includes the "release" of technology or software to foreign persons in the United States, see id. § 734.2(b)(1)-(3). The definition of "export" for encryption software also includes publication on the Internet:

(ii) . . . making such software available for transfer outside the United States, over wire, cable, radio, electromagnetic, photo optical, or photoelectric or other comparable communications facilities accessible to persons outside the United States, including transfers from electronic bulletin boards and Internet file transfer protocol and World Wide Web sites, unless the person making the software available takes precautions adequate to prevent unauthorized transfer of such code outside the United States.⁽⁵⁾

15 C.F.R. § 734.2(b)(9)(ii). Only encryption software is considered to be exported in these ways.

All encryption software controlled under 5D002 is subject to licensing, and a license is required before exporting an item to any destination other than Canada. See 15 C.F.R. § 742.15(a). License applications must be made in accordance with the EAR's licensing procedures and record keeping requirements, see id., Parts 748, 750 and 762, and are "reviewed by the BXA, in conjunction with other agencies to determine whether the export or reexport is consistent with U.S. national security and foreign policy interests."⁽⁶⁾ Id. § 742.15(b)

Under the EAR, it is the responsibility of the person intending to export an item to properly classify it. See id. § 748.3(a). If requested, BXA will provide information on whether an item is subject to the EAR, ECCN numbers and licensing questions through the "classification request" and "advisory opinion" procedures. See id. § 748.3(b)-(c). Responses from classification requests and advisory opinions, however, do not bind the BXA to issuing a license. See id. § 748(3).

If a person exports an item that requires a license without one, the person faces severe civil and criminal penalties. See id. § 764.3. Civil penalties can reach $10,000 for each violation, with a maximum penalty of $100,000. See id. § 764.3(a)(1). Criminal penalties for individuals range from $50,000 and five years in prison to $250,000 and ten years in prison for willful violations. See id. § 764.3(b)(1), (b)(2).

C. Factual Background Including Professor

Junger's Submissions

Appellant Peter Junger teaches a class in computers and the law at Case Western Reserve University Law School. In that course, he tries to teach students how to think about computers and what computers do. (R. 60 Junger Decl. ¶¶ 4-6, Apx. __.) He finds it extremely helpful to show his students computer programs, particularly encryption programs, because he finds that encryption programs help him to teach, and his students to understand many of the issues relating to the intersection of law and computer technology. (R. 60 Junger Decl. ¶ 8, Apx. __.)

Professor Junger has written computer programs, including a short encryption program called a "one-time pad," and with his students has used it to encrypt and decrypt secret messages. see id. He wishes to publish course materials containing that and other encryption programs on his web site. (See R. 1 Compl. ¶ 15.)

When Professor Junger that encryption programs might be governed by export laws, he contacted the State Department, which at that time had jurisdiction over most encryption. (R. 60 Junger Decl. ¶ 15, Apx. __.) Since he was not assured that he could freely publish encryption software on his web site, he has refrained from including encryption source code in a draft course book he is writing.

Fearing he could not teach foreign students about cryptography, and could not publish what he chose on his own web site, and without any recourse but to seek a license or continue to refrain from publishing, Professor Junger commenced this action in August 1996.

On January 2, 1997, counsel for Professor Junger sought written clarification from the Commerce Department regarding the transfer of jurisdiction over encryption from the State Department. (See R. 76 Tab B, letter from Commerce, January 29, 1997, Apx. __.) In his letter, counsel inquired, for example, whether books and articles, including law review articles containing encryption software were subject to export control under the EAR if electronically published. (See id. at 2.) On January, 29, 1997, James A. Lewis, Director of the Office of Strategic Trade and Foreign Policy Controls at the BXA responded to this inquiry in writing, but declined to answer that particular query because counsel had "not adequately describe[d] the nature of material Professor Junger proposed to export." (see R. 76 Tab B, letter from Commerce, January 29, 1997, at 4-5, Apx. __). Mr. Lewis further declined to answer other questions without Professor Junger submitting specific information regarding what he proposed to export. (See id. at 2.)

In order to receive specific information and classifications for some of the software he wanted to publish on his World Wide Web site, Professor Junger submitted three commodity classification application on June 12, 1997. (See R. 40 Tab C, Applications Nos. 082060, 082061, 082062, Apx. __.)

In the first application, Professor Junger requested classifications for a number of programs, including "Pretty Good Privacy" ("PGP") and the RSA algorithm written in the PERL programming language, which were downloaded from web sites outside the United States. (See R. 40 Tab C, letter to Applications No. 082060, Apx. __.) Professor Junger also requested classification for a "one-time pad" encryption program included in his draft course book, which he also wanted to publish on his web site. (See id.; see R. 4 Junger Decl. ¶¶ 8, 28, Apx. __.)

The BXA classified PGP and the Perl version of the RSA algorithm under ECCN 5D002, which under the regulations meant they could not published on the Internet without a license. (See R. 40 Tab D, Applications No. 082060 results, Apx. __.) The BXA, however, classified Professor Junger's one-time pad programs as EAR99, which meant that they could be made publicly available on the Internet without a license. (See id.)

In the second application, Professor Junger requested classification for three items, including specific web pages containing "hyperlinks" pointing directly to encryption programs located on foreign web sites. (See R. 40 Tab C, letter to Applications No. 082061, Apx. __.) The BXA was unable to classify these pages. (See R. 40 Tab D, Applications No. 082061 results, Apx. __.)

In the third application, Professor Junger requested classifications for all programs that implemented certain well known encryption algorithms,⁽⁷⁾ such as RC2 and RSA. (See R. 40 Tab C, letter to Applications No. 082061, Apx. __.) The BXA was unable to classify any of these items, even though specific examples were included with the application because "none of the requests seek a classification for a specific software item" "and "merely providing examples does not provide sufficient technical details of a specific item," (See R.__ Tab B, letter from Commerce, July 4, 1997, at 1, Apx. __.)

On July 18, 1997, and again on July 24, 1997, Appellant's attorney requested clarification of some of BXA's classification decisions. (See R. 40 Tab E, Apx. __.) The responses from the Director were quite telling. First, Professor Junger's course book was classified as EAR 99 because all of the programs contained within were classified as EAR 99. (See R. 40 Tab F, letter from Commerce, August 7, 1997, Apx. __.) Mr. Lewis, however, stated that if the book contained programs classified as 5D002 and the book was published in electronic form, the programs would remain classified for EI reasons, regardless of the surrounding contents in the book. (See R.40 Tab F, letter from Commerce, August 7, 1997, at 2, Apx. __.) Thus, Junger would have to remove any encryption program classified for EI reasons before publishing the book on his web site.

Second, regarding the web page with hyperlinks that BXA was unable to classify, Mr. Lewis gave a cryptic response: "While the use of html links by a person might, in some applications, involve an export, . . . the activity described by your submission is not an export activity." (See id.)

Finally, regarding the descriptions of programs that implement various encryption algorithms, Mr. Lewis reiterated the BXA's determination that descriptions of encryption programs could not be classified, . (See id. at 2.)

Since it was obvious that a controversy continued to exist between the Appellant and the government after the transfer of jurisdiction to the Commerce Department, the Appellant filed a supplemental and amended complaint. (See R. 75.)

Standard of Review

This Court reviews grants of summary judgment and First Amendment questions de novo. Montgomery v. Carr, 101 F.3d 1117, 1121 (6th Cir. 1996); Procter & Gamble Co. v. Bankers Trust Co., 78 F.3d 219, 227 (6th Cir. 1996) (citing Bose Corp. v. Consumers Union of United States, 466 U.S. 485 (1984)). Thus, the district court's decision may be upheld only if the record, when viewed in light most favorable to the plaintiff, contains no genuine issue of material fact and entitles the defendants to judgment as a matter of law. Montgomery, 101 F.3d at 1121.

Summary of the Argument

This is a case about the nature of protected expression. Encryption source code, in common with all source code, is a text, intelligible to those conversant with the language in which it is written. Like natural language texts, source code conveys meaning through a commonly understood set of meanings and representations. While the First Amendment extends its protection beyond the written word, linguistic expression is at the core of what the First Amendment protects. Source code is presumptively entitled to protection because it is an expressive text.

Source code is further entitled to First Amendment protection because it is an almost uniquely valuable of expression. Recognizing that the search for truth in a marketplace of ideas is a principal justification for protecting First Amendment freedoms, both the Supreme Court and this Court have adopted a broad conception of what is protected expression. Texts which are crucial to the exchange and exposition of ideas especially merit protection. Encryption source code is almost uniquely suited to the expression of certain ideas, which could no be expressed with adequate precision in its absence.

The district court ignored the controversy central to this case by failing to apprehend the unique utility of source code, and by focusing on its supposed functionality, a distinguishing trait of object code.

The district court also erred by reasoning that source code is not widely understood, and that most software users are concerned only with functionality and not expression. The concerns of the majority of users, and the fact that source code may be compiled into object code, are of no constitutional moment. The district court recognized that source code is a form of expression, and then ignored its own observation in holding it to be no part of the exposition of ideas.

The district court also erred in recognizing that the export controls at bar impose an impermissible prior restraint of on expression, in the archetypical form of a licensing scheme. The system of prepublication review lacks the procedural safeguards required by the First Amendment. The scheme is also content based, in that it distinguishes encryption source code from all other source code when it deals with encryption. Because the scheme cannot survive strict scrutiny, it should be invalidated under the First Amendment.

Law & Argument

This is a case about speech and about publication, the central question of which is this: Whether, consistent with the First Amendment, a law professor may be required to obtain a government licence in order to publish sections of text, when that text is written in an arcane and technical language, and destined to be published in a global medium the First Amendment context of which is undisputed.

Appellant Peter Junger is a law professor, who teaches a course in, and has written materials regarding, computers and the law. Junger has drafted a casebook on that topic, significant portions of which concern the legal questions surrounding computer programs, also known as software.

Junger believes that any sound knowledge of computer law requires an understanding of how programs are written. He also believe that encryption programs, which are designed to protect communications privacy, are especially important for lawyers to understand -- not only because they are essential to computer privacy, data security, and the protection of privilege communications -- but because they illustrate the deep nexus between programming, language and thought. For this reason, academic articles written by Junger contain examples of encryption programs written in "source code," that is, in various high-level computer programming languages which are widely written, read and understood by computer programmers.

While Professor Junger is free to publish his work on paper, he may not do so on the Internet without obtaining prior government approval. Any publication of encryption software "online" is subject to export controls, and requires a government licence before publication on the Internet. See: Bernstein v. United States Dep't of State (Bernstein III), 974 F.Supp. 1288, 1306 (N.D. Calif. 1997) (appeal docketed); 15 C.F.R. § 734.2(b)(9)(ii).

This is true despite the fact -- recognized by the district court below -- that Junger may freely publish his text abroad in books, magazines and other printed media. 8 F.Supp.2d 708, 720. And yet this distinction, between the Internet and more traditional media, runs afoul of the holding in Reno v. ACLU, 117 S.Ct. 2329, 2334 (1997), which held the Internet to be a traditional public forum, publication in which is entitled to the fullest measure of First Amendment protection. The export regulations at bar consequently restrict the use of the Internet as a medium of expression and exchange for computer programmers, scientists, researchers and other academics, who use programming languages to express and communicate their ideas about encryption.

Were the restrictions at issue applicable to ordinary printed matter, or to a familiar medium like film, they would unquestionable be subject to the familiar set of procedural safeguards which the First Amendment imposes on all such licencing schemes. See Freedman v. Maryland, 380 U.S. 51 (1965). Here, they are not.

Were the restrictions at bar applicable to the publication of articles in ordinary journals, then the distinction drawn by the regulations between source code regarding encryption, and source code regarding every other topic, would certainly merit the strictest constitutional scrutiny, as a content-based restriction on expression. Simon & Schuster Inc. v. New York State Crime Victims Board, 502 U.S. 105 (1991)(invalidating as content based a statutory scheme seizing profits earned through publication of criminal memoirs); Carey v. Brown, 447 U.S. 455 (1980)(invalidating as content based restrictions which allowed picketing on only certain subjects). Despite the obvious difference between their treatment of encryption source code and all other source code, the district court subjected the regulations at bar to no such scrutiny. 8 F.Supp.2d at 719-20.

Despite substantial evidence in the record to support the fact that computer source code is the lingua franca of programmers and others involved in the disciplines of computer science, the district court held, without so much as a smidgen of support, that source code is not "sufficiently communicative" to merit the protection of the First Amendment, consigning a unique medium of expression to the categorical dustbin of First Amendment discards along with obscenity and fighting words. 8 F.Supp.2d at 715-16, 718.

It is from this decision that Professor Junger appeals. The argument presented below, in support of that appeal, has essentially four parts.

First, Professor Junger contends that encryption source code, like source code generally, is a form of expression entitled to the full protection of the First Amendment, and that it is not only a tool for the exposition of ideas, but a tool so uniquely suited to the exposition of certain ideas, that without it, those ideas could not be adequately expressed.

Second, Professor Junger argues that the district court applied a deeply flawed analysis in order to arrive at the conclusion that encryption source code is more functional than expressive, and that this should somehow deprive source code, which the district court none-the-less recognized as a form of expression, of the fullest degree of First Amendment protection. In fact, Professor Junger argues, in the context relevant to this action, source code is profoundly, and uniquely expressive.

Third, Professor Junger contends that the regulatory scheme at issue in this case is fatally flawed in its methodology, for two distinct reasons: (a) because it imposes an archetypical prior restraint, a licensing scheme, on protected expression, and; (b) because it does so without providing the sorts of procedural safeguards and opportunities for prompt review required of such schemes by the First Amendment.

Fourth, Professor Junger contends that the regulations at issue are, on their face, content based regulations on free expression, which not being necessary to serve a compelling state interest cannot survive strict constitutional scrutiny.

I. The Regulations Impermissibly Restrict the Publication Of Expression Protected by the First Amendment.

The First Amendment protects expression about many subjects, including science, computers, and encryption. It also protects expression in many forms. As one Supreme Court Justice has noted, "constitutionally protected forms of expression include parades, dances, artistic expression, picketing, wearing arm bands, burning flags and crosses, commercial advertising, charitable solicitation, rock music, some libelous false statements, and perhaps even sleeping in a public park." John Paul Stevens, The Freedom of Speech, 102 Yale L.J. 1293, 1298 (1993).

The First Amendment "does not end at the spoken or written word," Texas v. Johnson, 491 U.S. 397, 404 (1989), but extends to the wide variety of forms taken by human expression. This not withstanding, linguistic expression -- the realm of the spoken and written word -- lies near the doctrinal heart of the First Amendment, and is central to the issues presented in this case.

In the argument which follows, Professor Junger will demonstrate that computer source code written in a programming language is a form of text, in the most traditional sense, and a form of expression fully entitled to the protection of the First Amendment. The district court erred in that it both recognized and then disregarded the communicative aspect of source code, focusing on the functional capabilities of software, as if those somehow robbed source code of its expression.

A. The Publication of Source Code Is Expression Essential to the Exposition of Many Technical and Scientific Ideas.

Foremost among the theoretical justifications for protecting expression is the notion that the search for truth is best conducted in an open "marketplace"where the "free trade" in ideas winnows truth from falsehood. Board of Education of the Island Trees Union Free School Dist. v. Pico, 457 U.S. 853, 866 (1982); Abrams v. United States, 250 U.S. 616, 630 (1919) (Holmes, J., dissenting). On the strength of this theory, both the Supreme Court and this Court have recognized a wide variety of expression as entitled to the protection of the First Amendment. See Hurley v. Irish-American Gay, Lesbian & Bisexual Group, 515 U.S. 557, 569 (1995) (recognizing parades, abstract painting, atonal music and nonsense verse as "unquestionably shielded" by the First Amendment); Dambrot v. Central Michigan University, 55 F.3d 1177, 1188 (6th Cir. 1995) (recognizing that the First Amendment protects "the public expression of ideas, narratives, concepts, imagery, opinions -- scientific, political, [and] aesthetic") (quoting Swank v. Smart, 898 F.2d 1247, 1250 (7th Cir. 1990)).

Expression in traditional media, like speech or text, is implicitly within the protective ambit of the First Amendment. Cf. Spence v. State of Washington, 418 U.S. 405, 409 (1974) (reasoning that it was "necessary to determine whether [appellant's] activity was sufficiently imbued with elements of communication" because the "appellant did not choose to articulate his views through spoken or written words"); see also Yniguez v. Arizonans for Official English, 69 F.3d 920, 934-35 (9th Cir. 1995), vacated on other grounds, 117 S.Ct. 1055 (1997) (holding that language is by definition speech).

The Circuit has taken a similar, broad view of linguistic expression, holding that the "only type of language that is denied First Amendment protection is 'fighting words,'" Sandul v. Larion, 119 F.3d 1250, 1255 (6th Cir.), cert. dismissed, 118 S.Ct. 439 (1997). Fighting words themselves are unprotected because they play "no essential part [in the] exposition of ideas." Id. (quoting Chaplinsky v. New Hampshire, 315 U.S. 568, 572 (1942)). Linguistic expression which plays some part in the exposition of ideas is, accordingly, protected by the First Amendment. For reasons elaborated below, Professor Junger argues that encryption programs written in source code are, by their very nature, a form of textual expression entitled to the protections of the First Amendment.

1. Like other statements expressed in language, writings in source code should be considered a protected form of expression within the meaning of the First Amendment.

Central to this case are a set of texts, examples of linguistic expression of a type both common and yet special. The encryption programs at issue are written in source code, the language of computer programmers.

As argued below, these texts should be considered a form of protected expression for two reasons: (a) because they are linguistic creations in the same sense as more traditional texts, and; (b) because they are intimately tied up with the expression of certain sorts of ideas.

a. Texts expressed in source code are linguistic creations in the same sense as more traditional texts.

As an initial matter, it is important to distinguish here between the "form" in which a given expression takes place and the "medium" in which the expression occurs. While the "medium" of expression usually refers to the physical or mechanical vehicle through which the expression is made manifest (on canvas, on television, on paper, or in marble bas relief, for example), the "form" of expression is better understood to refer to the type of expression through which meaning is made manifest (through the use of written words, symbols, song, or color). See, e.g., 15 C.F.R. § 734.2(b)(9)(ii). While certain forms may be inappropriate for certain media (for example, one cannot express color well on radio), it is easy to imagine other forms which lend themselves to a wide variety of media (the written word, or a particular symbol, might be displayed on paper, the computer screen, or on a marble entablature. The media differ, but the form of expression through which meaning is made manifest, writing, remains the same in each).

The significance of this distinction for source code becomes clear when one considers the regulations at issue, which permit the unlicenced distribution of source code -- a form of expression uniquely suited to conveying certain ideas -- in certain media (such as print) but not in others (such as in digital form online). Regardless of the medium in question, the text at issue has the same information content, much as the brief you are now reading has the same information content as the digital form saved on the hard drive of the author's computer.

In either their written or electronic form, then, programs written in source code share the same information content. In either medium, programs written in source code also share many of the linguistic elements of texts written in natural languages, like German or French. It was the similarity of source code to natural languages that lead the court in Bernstein v. Department of State, 922 F.Supp. 1426, 1435 (N.D.Cal. 1996)("Bernstein I") to conclude that source code is a form of expression protected by the First Amendment. The district court below rejected the Bernstein court's equation of "language with protected speech,"and in doing so, misapprehended the significance of language to the First Amendment. 8 F.Supp.2d at 716.

For First Amendment purposes, the use of language is highly significant. Languages are made up of conventions, regular ways of expressing and communicating thoughts and ideas shared by a group of speakers. See David Lewis, Languages and Language, Philosophical Papers Vol. I at 164-66 (1983).

Language is by nature a shared or community phenomenon, which separates the idiosyncratic from the public. As this Court recognized in Dambrot, the First Amendment protects the right to speak to an intended audience. 55 F.3d at 1188. A common language is what makes speech to such an audience possible; indeed, it may be what defines the audience, and vice versa. Speaker and audience, sharing a common language, communicate ideas to one another.

Language can be technical, and even arcane, but that does not make it any less like language. A native speaker of English, untrained in medicine, might have difficulty understanding the discussion of a cancer patient's progress. The vocabulary is different even if the grammar is not. Musicians use artificial notation for writing and reading music. Mathematicians and scientists do the same to express concepts important in their fields, with grammar and other conventions that may be wholly incomprehensible to the uninitiated.

For every shared language, natural or artificial, however, there is a community of speakers and listeners (or readers) for whom the language is both comprehensible and for which the language is its conventional form of communication. To belong to a community is to speak its language. In this light, the district court's dismissal of the relationship between language and protected expression central to Bernstein I is troubling, because the First Amendment serves to facilitate the exchange of ideas in a common marketplace, in which language is the common currency of exchange.

In short, to recognize the characteristics of language in a certain utterance is to recognize the potential for the exchange of ideas. Source code in this light becomes especially significant, as it not only facilitates the exposition of certain ideas, but does so in a way that makes its use practically indispensable.

b. What is special about computer programming languages.

Like all languages, programming languages "participate in a complex system of understood meanings within specific communities." Bernstein I, 922 F.Supp. at 1435 (quoting Yniguez, 69 F.3d at 934-35). Within the community of computer scientists and programmers, such languages are "a novel formal medium for expressing ideas about methodology." (R.60 Abelson Decl. ¶ 8, Apx. __.) The programming language LISP, for example, was invented in the late Nineteen Fifties as a formal means of reasoning about certain kinds of logical expressions, called recursion equations, and as a model for computation.⁽⁸⁾

This, of course, is how programming languages are different from other languages. They are used to express procedures (and ideas about procedures) with as little ambiguity as possible. "Just as everyday thoughts are expressed in natural language, and formal deductions are expressed in mathematical languages, methodological thoughts are expressed in programming languages." Abelson and Sussman, Structure and Interpretation of Computer Programs, preface at xv (MIT Press 1985). Thus, a fundamental use of programming languages, within the community of computer scientists and programmers, is to communicate ideas. (See R. 60 Ellison Decl. ¶¶ 8-9, Apx. __.)

A computer program, written in a programming language, communicates ideas on two different levels: first, as the district court below recognized, 8 F.Supp.2d at 716, it describes the instructions given a particular machine, and is thus an expression of procedures, but; second, it also communicates ideas about the writing of instructions, and in this sense is a tool for communication between professionals, computer scientists, programmers and enthusiasts.

Programming languages are not generally used to express "everyday thoughts," but they are an important form of expression. They are "the natural and best means of communication of some kinds of ideas, specifically mathematical concepts in the form of algorithms."( R. 60 Ellison Decl. ¶ 8, Apx. __.) And while at a highly abstract level, one might say that expressions in one language can be translated into another, such translations usually come at a price, and certain meanings may be impossible to convey with the desired nuance and precision after translation.

A prosaic example of this problem is presented by the familiar case of Cohen v. California, 403 U.S. 15 (1971), which upheld on First Amendment grounds the right to display a jacket bearing the epithet "F--- the Draft" in a courthouse hallway. It is possible to imagine Cohen translating his disapproval of the Selective Service into other terms, choosing perhaps to "Damn" or even "Curse the Draft." But while the basic message of such statements would be the same as the original at one level of abstraction (registering Cohen's disapproval of the draft), one is pressed to find a translation of the original which would express, with the same nuanced precision, Cohen's boundless contempt.

Programming languages may be the only practical means of expressing certain algorithms and other ideas that must be precisely defined.⁽⁹⁾ Descriptions in English or mathematics may not be sufficient to describe such ideas, certainly not for computer scientists and cryptographers. (See R. 60, Goldberg Decl. ¶¶ 10-11, Apx. __; R. 60 Appel Decl. ¶ 6, Apx. __.)

As Professor Knuth noted, programming languages avoid difficulties that English has in describing algorithms: "Each step of an algorithm must be rigorously and unambiguously specified for each case. The algorithms in this book will hopefully meet this criterion, but since they are specified in English, there is a possibility that reader might not understand exactly what the author intended." Donald Knuth, The Art of Computer Programming: Fundamental Algorithms (1968).

Thus, the very fact that the expression at issue in this case is in a precise sort of language matters, because language is what makes it possible to express particularized meaning to others. And programming languages make it possible to express exceeding precise, particularized meanings. In short, computer programs are statements or sets of statements in languages peculiarly suited for expressing procedures and ideas about procedures. In this way, they are almost uniquely suited to the expression of certain ideas.

2. Computer Programs Written in Source Code Are an Essential Part of the Exposition of Ideas.

Even source code which has been published or distributed so that it can run on a computer remains expressive, and can be used as an important means of scientific and technical communication. (See R. 60 Ellison Decl. ¶ 24, Apx. __ .)

Because programming languages allow the expression of ideas, books and articles in computer science and cryptography are commonly published that contain computer code, including student textbooks and scientific journal articles.⁽¹⁰⁾ Moreover, since computer programming has gained such widespread interest outside of research and academic institutions, major retail bookstores carry what must be hundreds of general programming books in a variety of programming languages.⁽¹¹⁾

The publication of computer code is indispensable for the advancement of computer science and cryptography. Like many areas of mathematics (statistics, for example), computer science and cryptography are applied sciences, and in the applied sciences, "ideas are not just expressed in abstract, theoretical terms, but in precise applications." Bernstein III, 974 F.Supp. at 1305. By publishing code, computer scientists and cryptographers share research, evaluate existing algorithms and introduce new ones. (See generally R. 60 Appel Decl. ¶¶ 3-10, Apx. __.)

The publication of encryption source code is essential to cryptography, an applied science, which deals with real world computer security issues. (See id. ¶ 10, 17, Apx.__; R.60 Goldberg Decl. ¶ 10, Apx.) In confronting those issues, cryptographers communicate in source code because computer security problems "are often so subtle that the only way to discover a security flaw is to be able to read the source code" itself. (Id. ¶ 11. Apx. __.)

Nor is the importance of publishing encryption source code limited to cryptographers and mathematicians. Professor Junger, for example, seeks to publish encryption source code in legal articles because "[c]ryptographic programs are by far the best tool that I know of for illustrating almost every important issue in law and computing, from the basic understanding of what it is that computers do to the complexities of intellectual property, privacy, and data security." (R.69 Junger (Third) Decl. ¶ 5, Apx. __.)

The district court dismissed cryptographic source code as something with "occasional" communicative elements, that is "rarely expressive, in limited circumstances," and thus more akin to fighting words or deceptive advertising than to protected expression. 8 F.Supp. at 717. This conclusion is deeply flawed.

B. The District Court's Analysis Is Internally Inconsistent, And Does Not Support Its Conclusion That Source Code is Unprotected Expression.

The district court acknowledged that source code is a form of expression for those literate in programming languages, including computer scientists and programmers; that it may communicate ideas, and; that even the "export" of source code can have communicative aspects. Id. Despite these conclusions, the district court held that encryption source code is "inherently functional," and that because its communicative nature was neither "'overwhelmingly apparent'" nor "'unmistakable,'" its export (that is, publication on the Internet) was not protected conduct under the First Amendment. Id. at 716-17 (quoting Johnson, 491 U.S. at 406; Tinker v. Des Moines Independent Community School Dist., 393 U.S. 503 (1969)). The holding below ultimately turned on the district court's conclusion that encryption source code is "inherently functional." But, as shown below, that conclusion cannot support the holding that encryption source code is unprotected expression, unworthy of the First Amendment.

1. The District Court's Distinction Between Inherently Functional and Inherently Expressive Software Is Inherently Unsound.

The district court began its analysis by distinguishing "inherently expressive" and "inherently functional" software. 8 F.Supp.2d at 716. "Certain software is inherently expressive" because it "contains an 'exposition of ideas.'" Id. Other software is "inherently functional" because users care mainly about what the software does and not about "the methods employed" or the "language used to control such methods." Id. Encryption software "is especially functional. . . . [it] does not merely explain a cryptographic theory or describe how the software functions. . . . [it] carries out the function of encryption" and in so doing "is indistinguishable from dedicated hardware that does encryption." Id.

The district court then relied upon Chaplinsky for the notion that the hallmark of protected expression is the "exposition of ideas," and concluded (without the benefit of citation to anything) that since most users look to encryption software for its function, and not the content of its source code, the publication of that source code is "no essential part of any exposition of ideas." 8 F.Supp.2d at 717 (quoting Chaplinsky, 315 U.S. at 572). In doing so, the district court both misapplied the law and misunderstood the facts, arriving at a conclusion unwarranted by the record.

a. The district court misapplied Chaplinsky v. New Hampshire.

For expression to be cast outside the protection of the First Amendment under Chaplinsky, it must be "of such slight social value as a step to truth that any benefit that may be derived from them is clearly outweighed by the social interest in order and morality." Chaplinsky, 315 U.S. at 572. Of course, on the record below, encryption source code is hardly so worthless. It is a powerful, almost uniquely suitable means of communicating subtle ideas with exceeding precision, vital to professional communication in the field of computer science.

b. The district court used an flawed analysis, mischaracterizing the expressive activity at issue.

The district court not only ignored facts in the record, but applied the law in such a fashion as to ignore the First Amendment activity central to this case, the publication and dissemination of source code among academics, programmers and scientists, choosing instead to focus on how software is viewed, and used, by the public at large:

Like much computer software, encryption source code is inherently functional; it is designed to enable a computer to do a designated task. Encryption source code does not merely explain a cryptographic theory or describe how the software functions. More than describing encryption, the software carries out the function of encryption. The software is essential to carry out the function of encryption. In doing this function, the encryption software is indistinguishable from dedicated computer hardware that does encryption.

* * *

In the overwhelming majority of circumstances, encryption source code is exported to transfer functions, not to communicate ideas.

8 F.Supp.2d at 716.

In a quiet shift, the district court here moves subtly from the discussion of source code, which even the district court recognized as expressive, to a conclusion based on the predominant mass market uses of software, or object code, without acknowledging the shift. But the shift is significant, and even dispositive, because the characteristics upon which the district court relied to find encryption source code outside the First Amendment are the characteristics of object code, and not the attributes of the source code Professor Junger seeks to publish.⁽¹²⁾ The predominance of function over form, the focus on the practical use of a program, rather than on its textual makeup, the similarity to embedded circuitry and the references to users "looking to performance" all describe "equipment executable" object code. Id. at 716; see also 15 C.F.R. Part 772.

This focus on a similar, and arguably more easily regulated, form of activity corollary to the expression at bar is not unique to the opinion below: it is exactly the sort of misdirection rejected by the Supreme Court in City of Lakewood v. Plain Dealer, 468 U.S. 750 (1988). In Lakewood, the High Court invalidated a licensing scheme which limited the installation of newsracks on city sidewalks. Rejecting an dissenting argument that found "no constitutional right to place newsracks on city sidewalks over the objections of the city," id. at 784 (White, J., dissenting), the Court recognized that "[t]he actual 'activity' at issue here is the circulation of newspapers," id. at 768, and not the placement of newsracks.

Like the dissent in Lakewood, the district court below has chosen to focus not on the actual activity at issue in this case -- the publication of source code -- but on related activities, like the distribution, use and performance of encryption object code in the form of software. In short, the district court substituted into a discussion of source code reasoning based on the ways in which object code may be used. Such a substitution ignores the expression at issue, in favor of regulating cases not before the court. Thus:

The right to demonstrate becomes the right to demonstrate at noise levels proscribed by law; the right to parade becomes the right to parade anywhere in the city 24 hours a day; and the right to circulate newspapers becomes the right to circulate newspapers by way of newsracks placed on public property.

Id. at 768. By focusing on the medium of dissemination (the Internet) and a similar but distinct form of expression (object code), the district court has made the same error, evaluating not the publication of source code, but the publication of something else entirely.

2. The District Court Misunderstood the Communicative Aspects of Publishing Programs in Source Code Form.

The district court was quick to categorize encryption software as "inherently functional," but gave no example of what software might be "inherently expressive." See 8 F.Supp.2d at 718. This omission is striking given the importance of this categorization, and fatal given the obstacle it creates to meaningful appellate review on a question of law.

The problem is that the district court wholly ignores the dual nature of software. Whether or not it expresses ideas or information when it is executed (i.e., run on a computer) software in source code form expresses ideas or information as a text written in programming languages.

As a text, source code, defined as "the convenient expression of a program,"⁽¹³⁾ is meant as much to convey ideas and information as it is to be converted by a compiler or interpreter to make computers perform specific operations. (See R. 60 Ellison Decl. ¶ 8, Apx. ___.) Programmers have even written source code with no practical application, which cannot run on a computer, because no appropriate compiler or interpreter exists. (See R. 60 Ellison Decl. 8, 20 Apx. ___.)

It is of no significance that programming languages are "unintelligible to most people."⁽¹⁴⁾ 8 F.Supp.2d at 718. In describing one of the primary purposes of the First Amendment, this Court has recognized that speech is geared to an audience: "The purpose of the free speech clause . . . is to protect the market in ideas, broadly understood as the public expression of ideas . . . to an audience whom the speaker seeks to inform, edify, or entertain." Dambrot, 55 F.3d at 1188 (emphasis added). What matters for First Amendment purposes is that for Professor Junger's intended audience -- those who understand programming languages -- source code is intelligible.

If there were a First Amendment requirement that is met only if the majority of people understand the speaker's message, then a speech in Navaho, the musical score of Beethoven's Ninth Symphony, and the proof of Fermat's Last Theorem,⁽¹⁵⁾ all of which are understood by relatively few people, would fall beyond the protection of the First Amendment.

The Constitution, however, has never countenanced a preference for one language over another, or required that a language be shared by a majority of the people in order to merit protection. See Meyer v. Nebraska, 262 U.S. 390 (1923) (declaring unconstitutional a law that prohibited school instruction in German rather than English); Yu Cong Eng v. Trinidad, 271 U.S. 500 (1926) (invalidating a law that prohibited accounting books in any language other than English or Spanish); Asian-American Business Group v. City of Pomona, 716 F.Supp. 1328 (C.D. Cal. 1989) (invalidating down a law restricting the use of non-English alphanumeric characters). Thus, for First Amendment purposes, it makes no difference that relatively few people understand programming languages.

3. The District Court Drew a Distinction Between Print and Electronic Form That Makes No First Amendment Difference.

Finally, the district court found that source code is functional "in a way instructions, manuals, and recipes are not," revealing a belief that source code is functional in the same sense as the parts in a machine, a belief supported by the court's analogy to the circuitry in a telephone. See 8 F.Supp.2d at 717.

Yet the district court also found that "in print, encryption source code is simply a description of instructions." Id., at 721. This statement renders unclear the lower court's understanding of source code, and incorrectly suggests that source code is functional in a way different from instructions communicated in language.

The district court may have meant that source code is functionally different than other kinds of instructions because it can be made to run on a machine. This certainly is a distinction, but it is a distinction without a First Amendment difference.

Source code displayed on a computer screen is as descriptive and expressive as it is printed on paper. In both forms, source code can be read and understood. Just as scientific journal articles and court opinions communicate the same ideas, whether published in print or electronic form, a program written in source code communicates the same ideas, whether published in a printed journal or on a World Wide Web site. In no sense is one more or less expressive than the other. There is no meaningful First Amendment distinction between them. Moreover, source code is not directly executable in either print or electronic form.

Source code must always be compiled or interpreted before a computer can execute its instructions. (See R. 87, Stipulations of Fact ¶¶ 10-11, Apx. ___.) The only "functional" difference between source code rendered in print and in an electronic form is that the latter is one step closer to being compiled, and thus executed, than in print form.⁽¹⁶⁾ The difference is one of degree, and not one of kind.⁽¹⁷⁾ For the district court to focus on the medium of expression as a meaningful distinction betrays a misunderstanding of the fact that source code by definition is not executable, and no more functional than plain language instruction like recipes and instruction manuals.

C. The Regulations Restrict Expressive Activity on the Internet

The marketplace of ideas is no longer confined to print. It now extends into "cyberspace," where the encryption regulations have their most deleterious effect on the free exchange of ideas. The Internet is "the most participatory marketplace of mass speech that this country -- and indeed the world -- has yet seen." ACLU v. Reno, 929 F.Supp. 824, 881 (E.D. Pa. 1996), aff'd, Reno v. ACLU, 117 S.Ct. 2329 (1997), and entitled to the fullest First Amendment protection.

The Internet's predecessor, the ARPANET, was designed initially to help the military, defense contractors and universities share defense-related research. See ACLU, 929 F.Supp. at 831. The World Wide Web, or simply "Web," had its origin at CERN, the European Particle Physics Laboratory, and was designed to share information and research on high-level particle physics. See id. at 836. Since their origins as means of exchanging information and research, the Internet and the Web have experienced phenomenal growth,⁽¹⁸⁾ but remain an integral means of scientific communication.

The Internet is fast becoming the preferred medium of communication for the scientific and academic communities and may someday replace print. (See R. 60 ACM letter at 1-2, Apx. __.) Scientific and academic papers made available online can swiftly reach a large audience for review and comment.⁽¹⁹⁾

Many universities have set up web sites covering a range of subjects, providing students with class information, course materials (including source code) and homework assignments. Some universities even require that student theses and dissertations be published on the Web.⁽²⁰⁾

The Internet is no longer strictly the domain of scientists, academics and researchers. Information is made available online by and for private individuals, non-profit organizations, businesses and governments in almost all fields and disciplines.

Yet of all the speech occurring on the Internet, only one kind is restricted under the EAR. Through their definition of "export" for encryption software (see 15 C.F.R. § 734.2(b)(9)(ii)), the regulations restrict any Internet publication or discourse which includes encryption source code.

The regulations provide general exemptions from export control for software and technology that are publicly available, including software and information that is the result of scientific research or academic activity,⁽²¹⁾ but none of these exemptions apply to encryption source code.⁽²²⁾

According to the William Reinsch, Undersecretary of Export Administration, if the general academic exemption, for example, applied to encryption source code, it would create a "significant loophole in export controls." (See R. 63 Reinsch Decl. § 21, Apx. __.). And so a wide range of scientific and academic discussion is restricted on the Internet. Without prior governmental approval, doctoral students cannot publish dissertations in cryptography on their universities' web sites;⁽²³⁾ Professor Junger cannot publish parts of his textbook and articles on Computing and the Law on his own web site;⁽²⁴⁾ organizations like the Association for Computing Machinery ("ACM") and the American Association for the Advancement of Science ("AAAS") cannot publish electronic versions of their journals on the Internet;⁽²⁵⁾  and computer programmers and scientists cannot use the Internet to collaborate on encryption research projects.⁽²⁶⁾

The encryption regulations "effectively eliminate all communications on electronic media that describe or discuss cryptographic source code." (R. 60 ACM letter at 3, Apx. __.)

Because the encryption regulations restrict publication on the Internet, the regulations restrict not only what U.S. persons can publish, but also what they can receive. The right to receive information is a well established corollary to the right of free speech. See Pico, 457 U.S. at 867; Pacific Gas & Electric Co. v. Pacific Utilities Comm'n of California, 475 U.S. 1, 8 (1986). If an article cannot be placed on the Internet, it cannot be received by either foreign or American citizens. The encryption regulations thus are not merely restrictions on the "export" of ideas and information to persons outside the United States, but restrictions on the domestic dissemination of ideas and information.⁽²⁷⁾

II. The Regulations Create a Licensing Scheme That Violates The First Amendment.

Before Professor Junger or any United States citizen or permanent resident can publish encryption source code on the Internet, he or she must either submit the source code with a formal classification request to the BXA⁽²⁸⁾ or apply directly for a license under a complicated set of application and record keeping rules.⁽²⁹⁾ The person could forego these procedures and publish without a license, but then risks severe civil and criminal penalties if the BXA classifies the source code for "EI" reasons. See 15 C.F.R. § 764.3. The right to publish is thus conditioned on the advance approval of government officials.

As a licensing scheme on expressive activity, the regulations create a prior restraint that can be challenged facially and can only survive First Amendment scrutiny by providing procedural safeguards and requiring the government to meet the strict burden of showing "direct, immediate and irreparable" harm.

A. The Licensing Scheme Is An Unconstitutional Prior Restraint on Free Expression.

1. The scheme is a prior restraint on publication.

Prior restraints are prohibitions on speech and expressive activity imposed in advance of expression or publication. See Southeastern Promotions, Ltd. v. Conrad, 420 U.S. 546, 552-53 (1975); see also Ward v. Rock Against Racism, 491 U.S. 781, 795 n.5 (1989). The doctrine of prior restraint in its clearest form "arises in those situations where the government . . . undertakes to prevent future publication or other communication without advance approval of an executive official." Thomas Emerson, The Doctrine of Prior Restraint, 20 Law & Contemp. Probs. 648, 655 (1955). As the Court noted in Ward, "the regulations we have found invalid as prior restraints have 'had this in common: they gave public officials the power to deny the use of a forum in advance of actual expression.'" Id. (quoting Southeastern Promotions , 420 U.S. at 553).

The encryption regulations strictly condition use of the Internet on advance government approval. All encryption software controlled under ECCN 5D002 (i.e., for "EI" reasons), including all academic and scientific publications that contain encryption source code, require a license before they can be published on the Internet. The regulations require prepublication review and expressly state that licensing decisions are made on a "case-by-case" basis.⁽³⁰⁾ If a person publishes encryption software on the Internet without a license, that person has committed a crime, and the crime is not contingent on harming or threatening national security, but simply publishing without a license.

Notwithstanding this, the district court held that the encryption regulations do not constitute a prior restraint because "exporting encryption source code is not an activity that is 'commonly associated with expression.'" 8 F.Supp.2d at 718-19. The court relied on the assumption that encryption source code is only occasionally distributed for expressive reasons and on a Ninth Circuit case, Roulette v. City of Seattle, 97 F.3d 300 (9th Cir. 1996), which involved an anti-sitting ordinance. This analysis fails for two reasons.

First, there is simply no support in the record for the court's position that encryption source code is "rarely expressive" or only occasionally "exported" for expressive purposes. Rather, as shown, the publication of encryption source code on the Internet is expressive activity and an essential part of academic and scientific discourse.

Second, the court's reliance on Roulette is completely misplaced because the ordinance in that case was neither a licensing scheme nor a prior restraint. Roulette concerned a city ordinance that generally prohibited people from sitting on public sidewalks during most of the day in the city's commercial district. See 97 F.3d at 302. The ordinance imposed a civil penalty of a $50 fine or performance of community service, see Roulette v. City of Seattle, 850 F.Supp. 1442, 1444 (W.D. Wash. 1994), rev'd by, Roulette, 97 F.3d 300, but was not a licensing scheme in that it in no way conditioned expressive activity, such as performances by street musicians,⁽³¹⁾ on the prior approval of government officials. Roulette did not involve any kind of prior restraint nor any kind of abuse of discretion.

Prior restraint analysis and facial challenges, however, are triggered by licensing schemes that condition expressive activity on the discretion of government officials. See Lakewood v. Plain Dealer Publish Co., 486 U.S. 750, 757 (1988).

2. A facial challenge under Lakewood is proper because the scheme vests excessive discretion in government officials.

It is well established that when a licensing scheme "allegedly vests unbridled discretion in a government official over whether to permit or deny expressive activity, one who is subject to the law may challenge it facially without the necessity of first applying for, and being denied a license." Lakewood, 486 U.S. at 755-56.

The encryption regulations condition the publication of encryption source code on the essentially unlimited discretion of BXA and other government officials. The regulations appear to be quite detailed, but licensing decisions in the final analysis are made on a "case-by-case" basis taking into account national security and foreign policy interests, which are inherently imprecise.

The regulations, moreover, require prepublication review without providing clear, discernable standards to restrain the discretion of administrating officials. Thus, for example, the BXA would not classify "one-time pad" programs written by Professor Junger that had the capability to encrypt (and thus met the regulatory definition) until he formally submitted them.⁽³²⁾ Once he did, the BXA classified them as something other than encryption software .⁽³³⁾ On the other hand, government officials have threatened to classify software that did not have any encryption capability as encryption software.⁽³⁴⁾

The lack of discernable standards leave licensing and classification decisions to the appraisal and judgment of government officials, who make decisions on the basis of national security and foreign policy--two inherently vague reasons for decisions.

If a licensing scheme "involves appraisal of facts, the exercise of judgment, and the formation of an opinion by the licensing authority, the danger of censorship . . . is too great to be permitted." Forsyth County v. The Nationalist Movement, 505 U.S. 123, 131 (1992) (citations and internal quotations omitted). The risk of censorship, even if not exercised by government officials, was the major justification for permitting facial challenges in Lakewood. See 486 U.S. 757-58. The Court in Lakewood recognized that censorship could be self-imposed and if the risk of self-censorship was substantial, a facial challenge would be permitted to avoid a chilling effect on expression and expressive activity. See id. at 757.

The risk of self-censorship as a result of the encryption regulations is both real and substantial. Professor Junger has refrained from publishing encryption source code on his web site for almost three years.⁽³⁵⁾ Several declarants have noted that they and others have refrained from using the Internet to publish, teach and exchange research as a direct result of the challenged regulations.⁽³⁶⁾ Scientific journals will have to decide between publishing articles in cryptography without including source code or not publishing the articles at all. (See R. 60 AAAS letter, February 7, 1997, at 1-2, Apx. __.)

These activities are at the core of the First Amendment and all involve the publication of encryption source code, the kind of expressive activity directly and specifically targeted by the regulations. Thus, the regulations are "precisely the kind of law that risks self-censorship on the part of those that must apply for licenses and censorship on the part of the decisionmaker." See Bernstein III, 974 F.Supp. at 1305.

Moreover, even if some exports of encryption source code fall outside the protection of the First Amendment, the distinction between those exports that are protected and those that are not cannot be left to the discretion of BXA and other government officials.⁽³⁷⁾ Allowing BXA and NSA officials to determine which exports are protected and which are not only exacerbates the risk of censorship. "[T]he line between legitimate and illegitimate speech is often so finely drawn that the risks of freewheeling censorship are formidable." Southeastern Promotions, 420 U.S. at 558-59 (citation omitted).

Thus, the Court has not permitted licensing officials free reign to determine beforehand whether expression is obscene See, e.g., Vance v. Universal Amusement Co., 445 U.S. 308 (1980); Freedman, 380 U.S. 51 (1965). And in the area of national security, the government was not permitted to suppress publication that at least one justice believed would have been able to sustain a conviction under the Espionage Act. See New York Times Co. v. United States, 403 U.S. 713, 734-35 (1971) (White, J., concurring).

The preference for subsequent punishment over a prior restraint is deeply rooted in our history: "a free society prefers to punish the few who abuse rights of speech after they break the law than to throttle them and all others beforehand." Southeastern Promotions, 420 U.S. at 558-59 (citation omitted). For this reason, the Court has subjected prior restraints to rigid procedural tests that these regulations cannot meet.

2. The scheme is unconstitutional because it fails to provide procedural safeguards and because it fails to meet the burden imposed on prior restraints under United States v. New York Times.

Prior restraints are "the most serious and least tolerable infringement on first amendment rights."Nebraska Press Assoc. v. Stuart, 427 U.S. 539, 559 (1976). As a result, "[a]ny system of prior restraints of expression comes to this Court bearing a heavy presumption against its constitutional validity." New York Times, 403 U.S. at 714 (per curiam) (quoting Bantam Books, Inc. v. Sullivan, 372 U.S. 58, 70 (1963) (citations omitted)); Proctor & Gamble, 78 F.3d at 224.

Any system of prior restraint fails to meet this burden if it lacks the procedural safeguards set forth in Freedman v. Maryland, 380 U.S. at 58-60. See FW/PBS, Inc. v. City of Dallas, 493 U.S. 215, 227-28 (1990); East Brooks Books, Inc. v. City of Memphis, 48 F.3d 220, 224 (6th Cir.), cert. denied, 516 U.S. 909 (1995).. If the scheme, moreover, is a restraint on publication, the government must satisfy the additional burden of establishing a "direct, immediate and irreparable" harm. See New York Times, 403 U.S. at 730 (Stewart, J., concurring); Proctor & Gamble, 78 F.3d at 228 (Martin, Jr., J., concurring); Nebraska Press, 427 U.S. at 545.

The licensing scheme created by the encryption regulations fails to meet both requirements.

Under Freedman, prior restraints must provide (1) definite and reasonable time limits in which the licensor must decide to issue or refuse to issue a license, (2) prompt judicial review of any adverse decision, and (3) a burden on the government to seek court review and justify an adverse licensing decision. See East Brooks Books, 48 F.3d at 224 (6th Cir. 1995).

The regulations require licensing decisions to be made within 90 days from the time an application is registered with the BXA unless the application is referred to the President. See 15 C.F.R. § 750.4(a)(1). If an application is referred to the President, there are no time limits on his decisions. If a license is denied, an appeal can be taken to the Undersecretary of the Department of Commerce, but no specific time limit is given for a decision by the Undersecretary. See id. § 756.2(c)(1).

The 90 day time period, which may be extended by referral to the President, and the lack of any specified time period in which an appeal must be decided, fail the first two Freedman requirements. The 90 day period and unlimited period of appeal are not reasonable time limitations that provide for prompt judicial review. See Teitel Film Corp. v. Cusak, 390 U.S. 139, 141-142 (1968) (per curium) (holding that 50-57 day period for obtaining a decision in a film censorship context is not a sufficiently "specified brief period"). Moreover, because the discretion of the administrative officials is excessive, the regulations create too real a risk of censorship. Therefore, the third Freedman requirement must also be satisfied. See FW/PBS, 493 U.S. at 228. The regulations, however, do not seem to provide for any judicial review, let alone prompt judicial review.⁽³⁸⁾

The government has also failed to meet its burden under New York Times. The standard of immediate and irreparable harm requires proof, not mere "surmise or conjecture that untoward consequences may result." See id. at 725-27 (Brennan, J. concurring) (quoted in WXYZ, Inc. v. Hand, 658 F.2d 420, 426 (6th Cir. 1981). Although national security is no doubt a compelling state interest, the government has not demonstrated in any way that immediate or irreparable harm would occur if the publication of encryption source code on the Internet was not restricted through this particular licensing scheme. The government has not introduced evidence that any particular publication or transfer of encryption software on the Internet has caused the kind of harm required by New York Times, and the government can hardly show that all Internet publication of encryption software poses such harm.

Thus, the encryption regulations fail to meet each and every standard required to justify a prior restraint and are therefore unconstitutional.

B. The Licensing Scheme Is Content-Based And Cannot Survive Strict Scrutiny

The encryption regulations offend the First Amendment in ways other than as an unconstitutional prior restraint. The regulations are also content-based restrictions that discriminate on the basis of subject matter (cryptography) and content and thus must be evaluated under strict scrutiny.

1. The regulations, on their face, discriminate on the basis of both subject matter and content and therefore the government's purpose is not controlling.

A regulation is content-based if it restricts expression based on "its message, its ideas, its subject matter, or its content." Arkansas Writer's Project v. Raglan, 481 U.S. 221, 229 (1987). The district court recognized that "the government subjects encryption software to heightened licensing regulations," 8 F.Supp.2d at 720, but held that the regulations were content-neutral because the government's purpose was controlling and was unrelated to the suppression of ideas. See id. The government's purpose, however, is not controlling.

In many cases, whether a law is content-based or content-neutral will depend on whether the government's purpose is to suppress or to favor certain ideas. See Ward, 491 U.S. at 791. But that is not true in all cases. "[W]hile a content-based purpose may be sufficient in certain circumstances to show that a regulation is content-based, it is not necessary to such a showing in all cases." Turner Broadcasting System, Inc. v. Federal Communications Comm'n, 114 S. Ct. 2445, 249 (1994). Where a government regulation discriminates on its face, the "mere assertion of a content-neutral purpose" will not be enough to otherwise save the regulation. See id. at 2459.

There is no question that the regulations at issue discriminate on their face against encryption source code. The most glaring example of discrimination is encryption source code's exclusion from the exemptions for public availability, 15 C.F.R. 734.3(b)(3), including fundamental research, id. § 734.8, and academic activity, id. § 734.9. Thus, while all other source code is a fortiori outside the scope of the EAR if it falls under one of those exemptions, encryption source code is still subject to control. The regulations accordingly discriminate on their face, and the government's assertion of a content-neutral purpose is neither dispositive, nor sufficient to save the regulations from being analyzed as content-based.

What is dispositive is that the encryption regulations effectively foreclose all discussions and publications on the Internet that involve encryption computer code. The result is that of all subjects that can be freely discussed on the Internet only cryptography is singled out for regulations and licensing. The regulations "prohibit the public discussion of an entire topic" in a medium fully protected by First Amendment and are thus properly seen as subject matter restrictions on speech. See Consolidated Edison Co. v. Public Service Comm'n of New York, 447 U.S. 530, 536-37 (1980) (invalidating restriction an utility company's inclusion of inserts in utility bills that discussed issues of public policy).

In addition to discriminating on the basis of subject matter, the encryption regulations also discriminate on the basis of content. The regulations require prepublication review before the BXA will issue a commodity classification or a license. (See, e. g., R. 76, Tab B letter form Commerce, January 29, 1997, at 2, Apx. __.) Prepublication review allows the BXA, NSA and other government agencies to review the content of the source code and make licensing decisions on the basis of that content. In fact, licensing decisions are made on the basis of content. The favorable licensing treatment given to mass market source code, for example, requires the prior submission of the source code and limits favorable treatment to source code that implements one of a few chosen mathematical algorithms.⁽³⁹⁾ Any distinction based on the choice of mathematical algorithm is a distinction based on "the content of ideas."

The district court's point that the regulations leave open publication in print does not alter the outcome. See 8 F.2d at 720. A similar point made by the government was explicitly rejected the Court in Reno.

In Reno, the government argued that restrictions imposed by the Computer Decency Act ("CDA") on speech in certain Internet modalities, left open "'a reasonable opportunity' for speakers to engage in restricted speech on the World Wide Web." See 117 S.Ct. at 2348. The Court found the government's argument to be "equivalent to arguing that a statute could ban leaflets on certain subjects as long as individuals were free to publish books." Id. at 2348-349. In rejecting the government's argument, the Court upheld the principle that, regardless of the content of a publication, a person's exercise of "his liberty of expression in appropriate places [cannot be] abridged on the plea that it may be exercised in some other place." Id. at 2349 (quoting Schneider v. State of N.J. (Town of Irvington), 308 U.S. 147, 163 (1939)). Similarly, restrictions on Internet publication cannot be justified because publication in print is permitted.

As content-based restrictions, the regulations must be evaluated under strict scrutiny, and under strict scrutiny, they are unconstitutional.

2. The regulations cannot survive strict scrutiny.

Content-based regulations on speech are presumptively invalid, see R.A.V. v. St. Paul, 505 U.S. 377, 381 (1992), but can survive strict constitutional scrutiny if the government has a compelling interest and "chooses the least restrictive means to further the articulated interest." Sable Communications of California, Inc. v. FCC, 492 U.S. 115, 126 (1989).

The challenged regulations cannot be the least restrictive means the government has to further its national security interests. Regulations that restrict United States citizens from publishing the same encryption source code that is freely available on Internet sites outside the United States are not narrowly tailored. The restrictions placed on Internet publication, moreover, may not be effective at all given the NSA's Admiral McConnell's testimony to Congress that "encryption software distributed via Internet, bulletin board or modem does not undermine the effectiveness of encryption export controls." (R. 60 McConnell testimony, May 3, 1994, Apx.__.) Regulations that are not effective in furthering an asserted government interest yet restrict a broad range of First Amendment protected activity cannot survive strict scrutiny.

CONCLUSION

For the reasons given above, this Court should reverse the district court's decision.

Respectfully submitted,

_______________________

Raymond Vasvari, Jr.

Attorney for the Appellant

CERTIFICATE OF SERVICE

A copy of the foregoing was mailed by regular U.S. mail on this the 1^st day of March 1999 to

Scott R. McIntosh
Attorney, Appellate Staff
Civil Division, Room 9550
Department of Justice
601 D Street, N.W.
Washington, D.C. 20530-0001

_______________________

Raymond Vasvari, Jr.

Attorney for the Appellant

[Notes]

1. Items not specifically listed on the CCL may still be controlled for export, under the catch-all classification EAR99. See 15 C.F.R. § 732.3(d)(5); § 774.1.

2. "Source code" is defined as "[a] convenient expression of one or more processes that may be turned into . . . equipment executable form [i.e., object code]." 15 C.F.R. Part 772. "Encryption source code" is defined as "a precise set of operating instructions to a computer that, when compiled, allows for the execution of an encryption function on a computer." Id.

Source code in some programming languages may be converted into object code through the use of a another program called a "compiler." (See R. 87 Stipulations of Fact ¶ 11, Apx. __.) Software written in "interpreted" programming languages, like PERL, are converted into executable machine code by a program called an "'interpreter' at the same time the source code is executed by the computer, without creating a separate file of object code." (See id. ¶ 12.)

3. "Object code" is defined as the "equipment executable form of a convenient expression of one or more processes ("source code" (or source language)) that has been converted by a programming system." 15 C.F.R. Part 772. "Encryption object code" is defined as "[c]omputer programs containing an encryption source code that has been compiled into a form of code that can be directly executed by a computer to perform an encryption function." Id.

"Object code may be conveniently thought of as a series of binary digits--"ones" and "zeros"--that may be directly executed by a computer." (See R. 87 Stipulations of Fact ¶ 10, Apx. __.)

4. See also Part 774, Note following 5D002; § 734.3(b)(2)-(3) and following Note; § 734.4; § 734.7(c); § 734.8(a); 734.9; Part 734 Supplement No. 1; 768(1)(b).

5. The district court noted that "it is nearly impossible for most Internet users to carry out or verify the precautions" required under the definition. See 8 F.Supp.2d at 713; (see also R. 60 Blaze (Second) Decl. ¶¶ 28-29, Apx. __; Junger (Second) Decl. § 15, Apx. __; ACM letter, February 12, 1997, at 3, Apx. __.) Thus, for all practical purposes, the regulations preclude the publication and distribution of all encryption software subject to ECCN 5D002 on the Internet, including encryption source code published in scientific and academic articles.

6. The recent amendments to the encryption regulations create a new licence "exception" ("ENC") that gives somewhat favorable treatment to some exports of encryption software, such as exports for use by insurance companies, U.S. subsidiaries located overseas and on-line merchants, and some mass market and non mass market encryption software. See generally 63 Fed. Reg. 72156-58, December 31, 1998.

7. An "encryption" or "cryptographic algorithm" is "a set of instructions usually made up of mathematical functions or equations, that can be applied to encrypt plaintext into ciphertext." (See R. 87 Stipulations of Fact ¶ 8, Apx. __.)

8. See John McCarthy, Recursive Functions of Symbolic Expressions and their Computation by Machine, Part I, 3 COMMUNICATIONS OF THE ACM 185 (1960).

9. See Don Smith, Computerizing Computer Science, COMMUNICATIONS OF THE ACM, September 1, 1998, (available on WestLaw, 1998 WL 13668576) (suggesting that programming languages may be the lingua franca of the future since natural languages are "certainly not ideal for many descriptive and performative tasks, such as describing mathematical, graphical, and computational objects, and specifying complex actions").

10. (See R. 87 Stipulation of Fact ¶ 18, Apx. __; R.60 Abelson Decl. ¶¶ 8-9, Apx. __; R. 60 Appel Decl. ¶¶ 4-5, 8, Apx. __.)

11. A recent search on Amazon.com's web site at http://www.amazon.com (last visited February 15, 1998) came up with over fifteen thousand books under the subject "computer programming."

12. For a discussion of these distinctions, see supra this Brief, at Notes 2 and 3.

13. See 15 C.F.R. Part 772 (definition of "source code").

14. The district court also stated that "[i]n the overwhelming majority of circumstances, encryption source code is exported to transfer functions, not to communicate ideas." Mem. Op. at 13. There is no support anywhere in the record for the district court's assertion, which on its face appears to be a finding of fact.

15. Fermat's Last Theorem is described as there being no n such that aⁿ +bⁿ = cⁿ, where n > 3.

16. Although errors may occur when source code on paper is scanned into a computer using Optical Character Recognition Technology ("OCR"), errors can also occur in source code that is written, for example, directly on a computer with a word processing program. All errors in a source code program whether caused by OCR or sloppy typing must be detected and corrected before source code can be compiled or interpreted and used to perform specific tasks like encrypting information. (See R. __ Stipulations of Fact ¶ 17.)

17. The government recognizes that the difference is only a matter of degree by reserving the right to control printed source code. See 61 Fed. Reg. 68575.

18. In June 1997 when the Court decided Reno, it was estimated that over 200 million people would be using the Internet this year. See 117 S.Ct. at 2334.

19. (See R. 60 Goldberg ¶ 12, Apx. __; Ellison ¶ 30, Apx. __.)

20. (See R. 60 Appel Decl. ¶ 16, Apx. __; R. 87 Stipulations of Fact ¶ 23, Apx. __.)

21. See 15 C.F.R. § 734.3(b)(3), § 734.7, § 734.8 and § 734.9.

22. See 15 C.F.R. § 732.2(b), § 734.3(b)(3), § 734.7(c), § 734.8(a), § 734.9 and Supplement No. 1 to Part 734.

23. (see R. 60 Appel Decl. ¶ 17, Apx. __.)

24. (see R. 76 Tab B, letter from Commerce, January 29, 1997, at 2, Apx. __.)

25. (see R. 60 ACM letter at 2, Apx. __; AAAS letter February 7, 1997, at 1-2, Apx. __.)

26. (see R. 60 Goldberg Decl. ¶ 16, Apx. __ .)

27. This is not to imply that the First Amendment ends at our nation's borders. See Bullfrog Films, Inc. v. Wick, 847 F.2d 502, 509 n.2 (9th Cir. 1988) (First Amendment protection extended to communications with foreign audiences unless there is a "clear and direct threat to national security"); see also Reid v. Covert, 354 U.S. 1, 5-6 (1957) (First Amendment protection extended to U. S. citizens in foreign countries); (see also R. __ Washington Post, September 27, 1997, Apx. __) (displaying graphic of estimates of number of U.S. citizens living abroad).

28. (See R. 76 Tab B, letter from Commerce, January 29, 1997, at 2, Apx. __.)

29. See 15 C.F.R. Parts 748, 750, 762.

30. 15 C.F.R. § 742.15(b); (see also R. 76 Tab B, letter from Commerce, January 29, 1997, at 2, Apx. __.)

31. See 97 F.3d at 303 n.7.

32. (See R. 76 Tab B, letter from Commerce, January 29, 1997, at 2, Apx. __.)

33. (See R. 76 Tab D, letter to Application No. 082060, July 4, 1997, Apx. __.)

34. (See R. 60 Behlendorf Decl. 5-11, Apx. __.)

35. (See R. 1 Compl. ¶ 13, Apx. __.)

36. (See, e.g., R. 60 Blaze Decl. ¶ 33, Apx. __ ; Goldberg Decl. ¶¶ 16-17, Apx. __; Behlendorf Decl. ¶¶ 10-11, Apx. __.)

37. Laws that regulate both expressive activity and non-expressive activity are also not immune from prior restraint analysis. See Vance v. Universal Amusement Co., 445 U.S. 308 (1980) (striking down a statute on prior restraint grounds that regulated gambling, prostitution and bull fighting as well as obscenity); see also FW/PBS, Inc. v. City of Dallas, 493 U.S. 215, 240 (1990) (Brennan, J., concurring) (describing the statute in Vance as a "general public nuisance statute").

38. See Executive Order No. 13026, 1996 WL 666563, sec. 3.

39. (See 15 C.F.R. § 742.15(b)(1)(i), 63 Fed. Reg. 72162; see also Supplement No. 6 to Part 742(a)(2)(ii),(b)(5)(E)93), 63 Fed. Reg. 72164-72165.)

[End]

Conversion to HTML by JYA/Urban Deadline.