IN THE UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT DANIEL J. BERNSTEIN ) ) Plaintiff/Appellee, ) ) v. ) ) No. 97-16686 ) UNITED STATES DEPARTMENT OF ) C-95-0582 MHP STATE et al., ) (N.D. California, ) San Francisco) ) Defendants/Appellants ) ) ON APPEAL FROM THE UNITED STATES DISTRICT COURT FOR THE NORTHERN DISTRICT OF CALIFORNIA HONORABLE MARILYN HALL PATEL, PRESIDING APPELLEE'S OPPOSITION TO EMERGENCY MOTION FOR STAY CINDY A. COHN, ESQ.; SBN 145997 McGLASHAN & SARRAIL Professional Corporation 177 Bovet Road, Sixth Floor San Mateo, CA 94402 Tel: (415) 341-2585 Fax: (415) 341-1395 LEE TIEN, ESQ.; SBN 148216 1452 Curtis Street Berkeley, CA 94702 Tel: (510) 525-0817 JAMES WHEATON; SBN 115230 ELIZABETH PRITZKER; SBN 146267 FIRST AMENDMENT PROJECT 1736 Franklin, 8th Floor Oakland, CA 94612 Tel: (510) 208-7744 ROBERT CORN-REVERE, ESQ. JONATHAN FRANKLIN, ESQ. Hogan & Hartson, L.L.P. 555 Thirteenth Street, NW Washington, DC 20004 Tel: (202) 637-5600 Attorneys for Appellee Daniel J. Bernstein INTRODUCTION After conceding that source code is speech ("It is undeniably true that encryption source code, such as the source code for Snuffle . . . can be read and understood by computer scientists and programmers.")(Appellants' Stay Motion at 15), Appellants still assert that the prior restraint on this speech undeniably created by their regulatory scheme should be continued pending appeal. This request, which has been denied by the District Court, should again be denied. BACKGROUND For the past five years the government has restrained the Plaintiff/Appellee, Professor Daniel Bernstein, from publishing his academic research and computer source code in the field of cryptography. The First Amendment to the United States Constitution denies to the government the ability to impose prior restraints on speech except in those few "exceptional cases," Near v. Minnesota, 283 U.S. 697, 706 (1931), where publication will cause "direct, immediate and irreparable harm to the nation." New York Times Co. v. United States, 403 U.S. 701, 730 (1971) (Stewart, J., joined by White, J. concurring); see also id. at 726-27 (Brennan, J., concurring); id. at 719 (Black, J. and Douglas J., concurring). Nor may the government impose even lesser restrictions on expressive activity based on "speculation about serious harms." United States v. Nat'l Treasury Employees Union, 115 S. Ct. 1003, 1017 (1995). Instead, the government must "demonstrate that the recited harms are real, not merely conjectural." Turner Broadcasting Systems, Inc. v. FCC, 114 S. Ct. 2445, 2470 (1994). Despite these settled rules of constitutional law, Professor Bernstein has been precluded from engaging in traditional academic dialogue in his chosen field since 1992, when he first sought assurance from the government that he could publish his Snuffle 5.0 encryption software and related documents. The government steadfastly denied his request -- just as it is seeking to do now. During the intervening years, Professor Bernstein has been given erroneous information (that subsequently was disavowed by the government), had an academic paper about Snuffle wrongly classified as a "munition" (also disavowed by the government), and has been subjected to unconscionable delays. Bernstein v. U.S. Department of State, 922 F.Supp. 1426, 1434 (N.D. Cal. 1996) ("Bernstein I") ("item defendants now contend could not be subject to regulation was apparently categorized as a defense article and subject to licensing for nearly two years"). Professor Bernstein was forced to seek redress in federal court. The District Court held, in three separate opinions, that the Government's licensing regulations on encryption source code constitute a facial prior restraint on protected speech. In two of those opinions, the District Court held that the prior restraint was unconstitutional. In the last of those opinions, issued August 25, 1997, the District Court granted an injunction to protect Professor Bernstein and others from prosecution for discussing the science of cryptography, including in the international electronic forum of the Internet. This injunction was limited to Professor Bernstein and those who would interact with him and the cryptographic ideas he sought to publish and discuss. Shortly thereafter, in response to the Government's emergency stay motion in the District Court, that Court reluctantly stayed the injunctive relief in part, limiting it even further to protect only Professor Bernstein and only in the electronic publication of a two-page computer program called Snuffle 5.0 (Exhibit A filed herewith) and later versions of that program (hereinafter collectively referred to as "Snuffle") pending appeal. The District Court emphasized its extreme reluctance to grant even the partial stay, given what it felt were facts clearly undermining the Government's claim of harm. See Transcript made part of Order. The Government here challenges the remaining, extremely limited injunction left after the District Court's partial stay. They claim, without any concrete evidence, that electronic publication of Snuffle pending appeal of the broader question the facial invalidity of the encryption regulations will do "irreparable harm" to the security of the United States. Relying on declarations of government officials that say precious little about the subject of the instant Motion -- the ability to publish Snuffle -- Defendants instead base their stay request upon a general recitation of the history of military cryptography. "Encryption has long been a tool in the conduct of military and foreign affairs," Defendants state, Stay Mot. at 3, and they contend therefore that this Court must prolong the prior restraint without asking too many questions. See Crowell Decl. at para. 10. At bottom, Defendants assert that a stay is necessary because: (1) the District Court erred in holding that Plaintiff has been improperly restrained from engaging in fully protected speech, and (2) the nation will be irreparably harmed unless publication is restrained. Appellants' essential claim is that the keeping of messages secret is inherently dangerous. But "the American Founding Fathers did not believe codes and ciphers were employed from purposes of evil and cruelty." Rather, they viewed secret writing as an essential instrument for protecting critical information not only in wartime, but in peacetime, as well. During the period when our Constitution was written, "cipher was employed extensively not only in public correspondence but in the private correspondence of public men as well." Thomas Jefferson was one of the more prolific users of cryptography among the Constitution's framers. The use of cryptography, as well as free academic inquiry, have always been part of America's tradition of free speech. The government itself has acknowledged this in the past, concluding that restrictions on the academic exchange of cryptographic technical data were a prior restraint. The invention and use of computers does not alter the inherent First Amendment dimension of this issue. See Reno v. ACLU, 117 S. Ct. 2329, 2344 (1997). Here, the government's efforts to downplay the First Amendment intrusions created by its encryption software regulations are exceeded only by its attempt to inflate its national security assertions. Neither of these efforts is sufficient to require the granting of a stay to prevent the publication of Snuffle. ARGUMENT In order to obtain a stay pending appeal, the Government must demonstrate either that it will suffer irreparable harm pending appeal combined with probable success on the merits, or that serious questions exists on appeal and "the balance of hardships tips sharply in [its] favor." Tribal Village of Akutan v. Hodel, 859 F.2d 662, 663 (9th Cir. 1988). Moreover, the District Court's denial of a request for a complete stay pending appeal is entitled to substantial deference. In particular, a District Court order enjoining unconstitutional government action that violates the First Amendment may be set aside only if the court abused its discretion. Christian Science Reading Room v. City and County of San Francisco, 784 F.2d 1010 (9th Cir. 1986), cert. denied, 479 U.S. 1066 (1987); Regents of the University of California v. ABC, Inc., 747 F.2d 511, 522 n.7 (9th Cir. 1984) (denial of stay request is not an abuse of discretion). The Government has satisfied neither standard under which a stay could be issued to prevent the publication of Snuffle. They have presented only conclusory allegations that electronic publication of Snuffle will harm national security in some undefined way -- allegations that only underscore the Government's lack of concrete justification for the prior restraint it seeks to impose. Balanced against these unsupported assertions is the very real, and necessarily irreparable, harm that Professor Bernstein will suffer through the continued prior restraint on the publication of his academic work. Finally, even if the Government could show that it would suffer irreparable harm in the absence of a stay, or that the balance of hardships tips "sharply" in its favor -- which it cannot -- the Government has failed to show the requisite likelihood of success on the merits. In sum, the District Court plainly did not abuse its discretion in holding that a stay should not be granted to prevent the publication of Snuffle pending appeal. The Government's regulations significantly impact Professor Bernstein's free speech rights, and that the Government's prior restraint contains none of the safeguards required by the First Amendment. Finally, the public interest -- far from militating in favor of the continued restraint -- in fact supports the District Court's injunction allowing Professor Bernstein to continue his academic work without further Government censorship. I. THE BALANCE OF DEMONSTRATED HARDSHIPS TIPS OVERWHELMINGLY IN PROFESSOR BERNSTEIN'S FAVOR A. The Government Has Failed To Prove Any Harm From the Publication of Snuffle, Much Less Irreparable Harm The Government cannot receive a stay based on merely speculative assertions of harm. See, e.g., Caribbean Marine Servs. Co. v. Baldridge, 844 F.2d 668, 674 (9th Cir. 1988) (preliminary injunction standards). Yet the Government has made no concrete showing that the limited injunction now in place will harm national security in any way. As the District Court found, the government has offered nothing more than bare allegations of harm, which are clearly insufficient to meet the requirements of harm necessary to justify a prior restraint on speech. Bernstein III, slip op at 26:3-20. This same bare allegation is insufficient here as well, where the government seeks to stay an injunction based on alleged national security concerns. See Armstrong v. Executive Office of the President, 877 F. Supp. 750 (D.D.C. 1995) (National Security Council denied stay pending appeal where it could not show likelihood of success on the merits or irreparable injury). The Government adds nothing new in their Motion for Stay or the supporting declarations, merely repeating the conclusory assertions that the use of encryption "can impede intelligence gathering," "can jeopardize" foreign policy interests and "can threaten" the safety of U.S. citizens. Stay Mem at 19. Once again, however, the Government offers no evidence whatsoever that any actual harm has or will result from the export of encryption programs that are freely available in this country in any form and freely exportable in paper versions, and it certainly has demonstrated no specific harm that will result from the posting of Snuffle on the Internet. Where the deprivation of First Amendment rights is at issue, it is an abuse of discretion to grant a stay absent a compelling showing by the government that such a stay is necessary. See Wildmon v. Berwick Universal Pictures, 983 F.2d 21, 23-24 (5th Cir. 1992) (stay of injunctive relief issued to "preserve the status quo" reversed where district court had granted injunction on First Amendment grounds). The Government's conclusory, unsupported allegations of harm merely underscore the patent invalidity of the prior restraint it seeks to impose on Professor Bernstein's speech. It seeks to review in advance, and censor, the electronic communications of Professor Bernstein and others in his field, yet presents no evidence of any specific national security interests that would be harmed by such communications. Moreover, it is plain that not even the Government believes its own assertions here regarding national security concerns for cryptography on the Internet. As Vice Admiral J. M. McConnell of the National Security Agency testified before Congress, "[e]ncryption software distribution via Internet, bulletin board or modem does not undermine the effectiveness of encryption export controls." Even Appellants' current declarant, William Crowell, told Congress as recently as last year that the Internet did not pose a problem for export controls because, "serious users of security products don't obtain them from the Internet." The Government's own testimony therefore confirms that they have failed to demonstrate irreparable harm from web site postings of encryption software in general, much less the specific posting of Snuffle. The Pentagon Papers case, which the District Court relied on below, further supports denial of the Government's motion. See United States v. New York Times Co., 328 F. Supp. 324, 330 (S.D.N.Y.) (injunction denied where government could not demonstrate irreparable harm), aff'd, 403 U.S. 713 (1971). There, the Second Circuit granted a very limited, three-day stay of the District Court's decision to permit publication of the Pentagon Papers despite the government's national security allegations. Yet the Supreme Court reversed even this short, targeted restriction on publication. United States v. New York Times, 444 F.2d 544 (2d Cir. 1971), rev'd, 403 U.S. 713 (1971). In a parallel case, the D.C. Circuit declined to overturn the District Court's denial of an injunction with respect to publication in the Washington Post, and the Supreme Court upheld that decision. United States v. The Washington Post Co., 446 F.2d 1327 (D.C. Cir. 1971) (en banc), aff'd, 403 U.S. 713 (1971). It is particularly instructive that the government argued in the Pentagon Papers case that suppression of speech was necessary to preserve secrets relating to cryptography. In a hearing at the U.S. Court of Appeals for the D.C. Circuit, an NSA official argued that, in its "worst case" scenario, publication of the Pentagon Papers would reveal that the United States had the ability to decode North Vietnamese communications. This assertion, however, was insufficient to support the proposed restraint. The Supreme Court similarly found the national security claims unsubstantiated, holding that the government must demonstrate that publication would cause direct, immediate and irreparable harm to the nation. As in the Pentagon Papers cases, it would be an abuse of discretion for this Court to grant this stay based upon nothing more than assertions of a national security interest. Cf. Clinton v. Jones, 117 S. Ct. 1636, 1651-2 (1997). With respect to national security, it should be emphasized that the injunctive relief presently in force is very narrow. All that Professor Bernstein can do under the remaining injunction is to publish electronically two pages of source code, Exhibit A, as well as any improvements which he has made to that code in the many years since he first began asking the government for permission to publish. No other person can republish Professor Bernstein's source code, even to comment on it or improve upon it as part of the normal scientific process. Furthermore, Professor Bernstein cannot publish any of his other ideas in the field or comment upon the ideas of others on the Internet. Even if the Government could demonstrate harm arising from Professor Bernstein's publication of cryptographic information in general -- which it cannot -- it plainly has shown no harm arising from the narrow relief that is presently at issue. Moreover, as noted, the Government's claim of irreparable harm claim is particularly undermined because of the exemption in the regulations for source code printed on paper. The District Court found that this exemption deprived the Government's national security claims of credibility, holding that: Defendants claim that encryption poses unique and serious threats to national security, yet the printed matter exception belies this rationale by making encryption freely available to only those foreigners are technologically sophisticated. Defendants conceded at oral argument that the effect of this dichotomy would be to make it more difficult only for the inept. Bernstein III, slip op. at 25. This finding was not an abuse of discretion. Not only is the distinction between Internet publication and paper publication clearly untenable in light of the Supreme Court's recent decision in Reno v. American Civil Liberties Union, 117 S.Ct. 2239 (1997), but it is not necessary to grant a stay to "preserve the status quo" where, as here, cryptographic information is freely available to foreign entities in other forms. Additionally, the Government's claims are undermined by the fact that a wide range of cryptographic software is already available overseas. See Declaration of David Balenson, with cryptography study attached (Exhibit G). When 570 cryptographic products are already available abroad, using strong cryptographic software and algorithms developed first in the U.S. as well as others developed by foreign sources, the government's claims of irreparable harm due to the publication of the simple Snuffle program begin to look quite ridiculous. Defendants' claim that a stay is necessary to preserve the status quo rests most uneasily next to their statement that "[o]nce Snuffle is exported to a person or group abroad, whether via the Internet or in some other way, it is impossible to control either the persons who receive it or the use that they make of it." Stay Mot. at 21. By their own assertions, this already is the status quo, since the government has claimed that Snuffle has been made available on "publicly accessible computers overseas." Stay Mot. at 21, Reinch Decl. 8. Under settled law, no injunction preventing publication can be retained under these circumstances, regardless of the government's national security assertions. B. Professor Bernstein Will Suffer Irreparable Harm to His First Amendment Rights if the Prior Restraint is Continued Pending Appeal Professor Bernstein has been restrained from the electronic publication of his protected expression for more than five years. And despite the District Court's December, 1996 holding that the restraint is unconstitutional and its pending issuance of a final injunction, he has continued to be restrained. The continued restraint on him is now plainly unacceptable. It is well-settled that " [t]he loss of First Amendment freedoms, for even minimal periods of time, unquestionably constitutes irreparable injury.' " American-Arab Anti-Discrimination Comm. v. Reno, 70 F.3d 1045, 1057 (9th Cir. 1995) (quoting Elrod v. Burns, 427 U.S. 347, 373 (1976) (plurality opinion)). Accord, Topanga Press, Inc. v. City of Los Angeles, 989 F.2d 1524, 1528-29 (9th Cir. 1993); San Diego Comm. Against Registration and the Draft v. Governing Bd. of the Grossmont Union High School Dist., 790 F.2d 1471, 1473 n.3 (9th Cir. 1986). Accordingly, a stay of District Court's injunction pending appeal, regardless of whether the appeal is expedited, would irreparably harm Professor Bernstein by preventing him from engaging in activities protected by the First Amendment. The Government nevertheless asserts that since it has successfully restrained Professor Bernstein for so long pending the District Court's final injunction, continuing that restraint "can hardly be said to impose a significant burden on Bernstein." Stay Mot. 22:11. This argument turns First Amendment prior restraint doctrine on its head. The fact that the Government has been able to censor Professor Bernstein's publications to date weighs not in favor of further restraint, but rather in favor of ending the restraint. Professor Bernstein has been waiting more than five years to be free from the threat of criminal prosecution for engaging in protected speech. Now that the District Court has finally ruled in his favor, he should not be forced to wait any longer. II. THE GOVERNMENT HAS NOT SHOWN THE REQUISITE LIKELIHOOD OF SUCCESS ON THE MERITS The Government's failure to identify any concrete harm that would befall it in the absence of a stay is fatal to its motion. See Bernard v. Air Line Pilots Ass'n, 873 F.2d 213 (9th Cir. 1989). In addition, the Government has failed to demonstrate the requisite likelihood of success on the merits. The District Court quite properly held that the encryption regulations are plainly unconstitutional on their face. They regulate protected speech; they are a prior restraint and they contain none of the procedural protections required by Freedman v. Maryland, 380 U.S. 51 (1965), and its progeny. Nothing in the government's brief casts doubt on these conclusions. A. The Regulations Specifically Target Speech Contrary to the government's current claim, the export controls relating to encryption are "specifically directed at speech protected by the First Amendment." Bernstein III, slip op. at 24. Just as it argued below, the government now asserts that the export licensing requirements at issue do not act as a prior restraint because they are not aimed at "preventing the free exchange of information and ideas" but only at the "functional capacity" of encryption software. Stay Mot. at 14, 15. Defendants' claim is wrong for several reasons. First, by their terms, the export regulations are not limited to controlling only source code or software, but apply to a range of expressive activities--including here to the "instructions" for Snuffle written by Professor Bernstein in English--whether or not they have "functionality." As noted below, in the realm of academic speech, restricting the exchange of source code inherently limits the exchange of "information and ideas." Moreover, by singling out software on the subject of cryptography from other software, the regulations are plainly content-based. The District Court found that: [A]s made explicit by the new regulations, export includes publication where the publication is or could be made electronic and even where the information to be published is already publicly available. In fact, in spite of the disclaimers regarding functionality and the exception for printed materials, the encryption regulations issued by the BXA appear to be even less friendly to speech interests than the ITAR. Here, encryption software is singled out and treated differently than other software regulated under the EAR. Bernstein III, slip op. at 24. B. Encryption Source Code, Like All Source Code, Is Protected Expression As noted above, the Government agrees that it is "undeniably true that encryption source code, such as the source code for Snuffle . . . can be read and understood by computer scientists and programmers" but asserts that restricting such academic exchange is not a limitation on speech because export controls are directed at the "non-expressive function of cryptographic software" and not at "whatever information may be claimed to be embodied and conveyed." Stay Mot. at 15-16. This is nothing more than sophistry, and Defendants cite no case law to support this new category of lesser protected speech they seek to have recognized. Moreover, as the District Court held, "functionality" cannot be restricted without censoring the "information . . . embodied and conveyed": By the very terms of the encryption regulations, the most common expressive activities of scholars -- teaching a class, publishing their ideas, speaking at conferences, or writing colleagues over the Internet -- are subject to a prior restraint by the export controls when they involve cryptographic source code or computer programs. In the field of applied science ideas are not just expressed in abstract, theoretical terms, but in precise applications. Those applications are subject to licensing under the encryption regulations and are excluded from the exemptions for fundamental research and educational information. Bernstein III, slip op. at 23. The District Court's findings are confirmed by M.I.T. computer science professor and author of the leading introductory textbook on computer science, Harold Abelson (see Exhibit E), among many others in the Declarations and exhibits submitted below. Professor Abelson quoted the introduction to his textbook: First, we want to establish the idea that a computer language is not just a way of getting a computer to perform operations but rather that it is a novel formal medium for expressing ideas about methodology. Thus, programs must be written for people to read, and only incidentally for machines to execute. Abelson Declaration, Exhibit E, at para. 8, quoting preface to Abelson & Sussman, "Structure and Interpretation of Computer Programs," MIT Press and McGraw-Hill Book Company (1985) Further support is found in scholarly legal analysis and in the fact that computer programs are protected under copyright law just as other forms of speech: As the Office of Technology Assessment noted, "Like other copyrightable works, programs symbolize information to human beings, and can be read and understood by programmers. The CONTU report stressed that programs, like other copyrightable works, communicate to those who can read them." Office of Technology Assessment, Intellectual Property Rights in an Age of Electronics and Information 80 (1986). Since the first amendment protects an individual's right to communicate with others, it must protect a computer programmer's right to communicate with others in a special language. . . . . Just as a mathematics text or written music communicates to a specially trained group of readers, a computer program communicates to its own group of readers. When seen in this light, the first amendment implications of computer programs are no different from those of many other copyrightable texts. Alfred C. Yen, "A First Amendment Perspective on Idea/Expression Dichotomy and Copyright in a Work's Total Concept and Feel," 38 Emory L.J. 393, 430 n.190 (1989). Examples of easily readable source code abound. See, e.g., Exhibit F, Decl. of Bernstein in Opposition to Defendants' Second Motion for Summary Judgment. Yet even if lay people are unable to read more complex programs, just as a lay person may not be able to read a symphony score, such programs are readable and are read by computer scientists and others are protected speech. C. The Regulations Are A Prior Restraint On Protected Speech The government acknowledges that its regulations prevent a professor of mathematics from publishing his own ideas, expressed in computer programming language, on the Internet. The government further admits, as it must, that such publication would be fully protected speech if accomplished in a book. Yet despite these admissions, the government contends that its regulations do not have a close enough nexus to expression or to conduct commonly associated with expression, to pose a real and substantial threat of censorship. City of Lakewood v. Plain Dealer Publishing Co., 486 U.S. 750 (1988). Instead, Defendants compare the encryption controls to restrictions on sitting or lying on a sidewalk, and cite as their primary authority Roulette v. City of Seattle, 97 F.3d 300 (9th Cir. 1996). The government made the same assertion below, without success. The District Court found that there is no legitimate comparison between the ordinance in Roulette and the licensing scheme, and it quite properly rejected the government's reliance on symbolic speech cases. Bernstein III, slip op. at 23-24 ("The encryption regulations . . . are much more like the regulation of newspaper racks than lying or sitting"); Bernstein I, 922 F. Supp. at 1435 ("A computer program is so unlike flag burning and nude dancing that defendants' reliance on conduct cases is misplaced."). Indeed, in every case where this Court has considered a licensing scheme directed at expression, like the one at issue here, it has applied the body of law applicable to prior restraints, and not the symbolic speech cases. This is nothing more than a straightforward application of the bedrock principle that the most rigorous First Amendment protections apply to licensing schemes that vest government officials with authority to review and limit speech in advance. Contrary to the government's assertions here, it is impossible to separate the communicative from the noncommunicative attributes because speech about cryptography (e.g., technology and technical data) is expression, algorithms and source code are an integral part of academic speech, and, as noted below, encrypted messages are themselves still speech. The government's argument is not that these regulations are unrelated to speech -- it is that encrypted speech, if used by the wrong people for bad purposes, is too dangerous. Defendants are free to try to persuade the Court that this is so, but it will require a far more rigorous constitutional analysis than that set out in Roulette. If conduct cases are to be triggered at all here, they can only be so with the observation that any issues of "speech" and "conduct" are inseparable. In cases such as this, the courts have rejected attempts to uphold licensing schemes despite the argument that the license applied to "a whole range of easily identifiable and constitutionally proscribable conduct." These cases rejected the notion that a licensing scheme could be upheld by separating the "noncommunicative elements" from the communicative elements of an activity. They held that the activities were "intertwined" so that regulation of one involved regulation of the other. Riley, 467 U.S. at 963 n.11; Schaumberg, 444 U.S. at 632; Gaudiya Vaishnava Society, 952 F.2d at 1065; See Lakewood, 486 U.S. at 762 (holding as "meaningless" the distinction of distributing newspapers "by a machine rather than by hand"). D. Functional Speech is Still Speech As noted above, the government's assertions regarding "functionality" do not diminish the District Court's First Amendment findings. Indeed, Defendants made the same arguments below, and the District Court found that "functionality does not remove [encryption software] from the realm of speech" or "`negate' its expressiveness." Bernstein III, slip op. at 34-35 n.20. See also Bernstein I, 922 F. Supp. at 1435 ("the functionality of a language does not make it any less like speech"). Various kinds of instructions, recipes, manuals and technical information "are often purely functional; they are also speech." Id. For example, "[t]he music inscribed in code on the roll of a player piano is no less protected for being wholly functional." Id. The government here simply assumes that by labelling speech as a "function" it may dispense with constitutional protections, as if running an cryptographic program is akin to activating some kind of doomsday device. But running an encryption program does not create a weapon; it merely makes speech private. Accordingly, even under Defendants' analysis, the "function" of an encryption program is to ensure speech is transmitted only to the audience of the speaker's choosing. Thus, federal law describes cryptography as a "method of secret writing," and encryption systems as "[s]peech scramblers [and] privacy devices." In the final analysis, encrypted speech is still speech, and the Supreme Court has held repeatedly that the First Amendment protects private or anonymous communications. McIntyre v. Ohio Elections Commission, 115 S. Ct. 1511, 1516 (1995); City of Ladue v. Gilleo, 114 S. Ct. 2038, 2046 (1994); Talley v. California, 362 U.S. 60, 64 (1960). See Rosen v. City of Portland, 641 F.2d at 1250-1252. It is true that speech might be used to further bad ends, and encrypted speech or source code is no different. A pen and paper can be used to scribble a ransom note or a threat to assassinate the President. Indeed, "[p]aper is a dangerous medium": "Hitler's `Mein Kampf' was written on paper." Michael Kinsley, A Dangerous Medium, Washington Post, April 16, 1997. The government's theory of "functionality" would permit the licensing of typewriters because they might be used to produce subversive tracts. Thomas Jefferson's cryptography would certainly be at risk by virtue of its "functionality." Or, to take an example that would resonate with those who crushed the student protests at Tienamen Square, FAX machines should be controlled because they can function to aid movements considered dangerous to state security. Defendants seeks to exploit a fear of computer technology, but their message is alien to our constitutional traditions. In upholding First Amendment protection for anonymous speech, the Supreme Court has reaffirmed the principle that "our society accords greater weight to the value of free speech than to the dangers of its misuse." McIntyre, 115 S. Ct. at 1524, citing Abrams v. United States, 250 U.S. 616, 630-631 (1919) (Holmes, J., dissenting). In addition, the Supreme Court recently held that distribution of speech over the Internet does not make it more dangerous or diminish its level of constitutional protection. Reno v. ACLU, 117 S. Ct. at 2344. The District Court correctly applied that finding here, and found that the export controls, which place special burdens on communication over the Internet, violate the First Amendment. Bernstein III, slip op. at 25. The government's stay request does not challenge these sound conclusions. III. THE PUBLIC INTEREST DOES NOT FAVOR A STAY Finally, the Government is wrong to contend that the public interest militates in favor of granting a stay. In fact, precisely the opposite is true. The public interest would be served by preventing the government from imposing its standardless prior restraint on Professor Bernstein's speech, not by continuing the restriction. Indeed, restrictions on the First Amendment rights of citizens are inherently contrary to the public interest. ACLU v. Reno, 929 F. Supp. 824, 851 (E.D. Pa. 1996)(opinion of Sloviter). Professor Bernstein is a professor of mathematics who wishes to engage in scientific discussion and teaching activities in the field of cryptography, including by use of cryptographic source code and by use of the speech arena of the Internet. As the District Court found, the encryption licensing regulations at issue here impede these scientific activities. As the court explained with reference to a statement by the American Academy for the Advancement of Science, enforcement of the regulations threaten[s] to undermine the essential features of scientific freedom and the open exchange of information that are generally acknowledged as critical to innovation in science and technology and are responsible in large part for the preeminence of America's research and development enterprise. Bernstein III, slip op. at 24:2-4 (quoting AAAS statement). See also Declaration of Prof. Appel of Princeton, Exhibit D, paras 11-17 (Internet publication is important part of scientific development). Particularly given that the Government does not -- and cannot -- identify any specific harm that would arise from Professor Bernstein's publication of his scientific work, the public interest clearly falls on the side of allowing him to exercise his First Amendment rights. CONCLUSION For the foregoing reasons, the Government's emergency motion for a stay pending appeal should be denied, and this appeal should be expedited. Respectfully submitted, McGLASHAN & SARRAIL Professional Corporation By________________________ CINDY A. COHN Dated: September 17, 1997 Attorneys for Plaintiff-Appellee