984 1 UNITED STATES DISTRICT COURT SOUTHERN DISTRICT OF NEW YORK 2 ------------------------------x 3 UNIVERSAL CITY STUDIOS, INC., et al, 4 Plaintiffs, 5 v. 00 Civ. 277 (LAK) 6 SHAWN C. REIMERDES, et al, 7 Defendants. 8 ------------------------------x 9 July 25, 2000 10 9:00 a.m. 11 Before: 12 HON. LEWIS A. KAPLAN, 13 District Judge 14 APPEARANCES 15 PROSKAUER, ROSE, L.L.P. Attorneys for Plaintiffs 16 BY: LEON P. GOLD CHARLES S. SIMS 17 CARLA MILLER WILLIAM M. HART 18 MICHAEL MERVIS 19 FRANKFURT, GARBUS, KLEIN & SELZ Attorneys for Defendants 20 BY: MARTIN GARBUS ERNEST HERNSTADT 21 DAVID ATLAS JEFFREY ROLLINGS 22 23 24 25 [ deletions ... ] 1061 1 THE COURT: Okay. Let's proceed. 2 DAVID TOURETZKY, 3 called as a witness by the Defendant, 4 having been duly sworn, testified as follows: 5 THE COURT: Proceed, Mr. Atlas. 6 DIRECT EXAMINATION 7 BY MR. ATLAS: 8 Q. Good morning, Doctor. Can you tell us by whom you are 9 presently employed? 10 A. Carnegie Mellon University in Pittsburgh, Pennsylvania. 11 Q. And what do you do at Carnegie Mellon? 12 A. I do research and teaching computer science. 13 Q. Do you teach in any specialized area within the field of 14 computer science? 15 A. Artificial intelligence and computational neuroscience. 16 Q. Can you just describe briefly those areas in a little bit 17 greater detail? 18 A. Artificial intelligence is concerned with getting 19 computers to do things that normally require human-like 20 intelligence or animal-like sense. Computational neuroscience 21 is the use of computers to understand how brains work. 22 Q. And how long have you taught computer science at Carnegie 23 Mellon? 24 A. I've been on the faculty since 1984. 25 Q. Can you tell us what degrees you have and from which 1062 1 schools you obtained them? 2 A. I have a bachelors degree in computer science from Rutgers 3 University, and I have a masters and Ph.D. degree in computer 4 science from Carnegie Mellon. 5 Q. Do you serve on any boards? 6 A. I serve on several editorial boards: The Editorial Board 7 of Cognitive Science. It's a journal. The Editorial Board of 8 Neuro-Computation. I'm also on the board of directors of 9 NIPS. NIPS stands for neural information processing systems. 10 It's a foundation that manages an annual conference for 11 researchers in the field of neuro-networks. 12 Q. Have you published any books? 13 A. Yes. I've published three books. 14 Q. Have you published any articles? 15 A. Approximately two dozen journal articles and about 60 16 conference papers. 17 MR. ATLAS: May I approach, your Honor? 18 THE COURT: Yes. 19 Q. I'd like to show the witness what we've marked as 20 Defendants' Exhibit CCM, and if you could tell us what that 21 is. 22 A. This is a copy of my curriculum vitae as of, 23 approximately, April of this year. 24 Q. Are there any changes to your CV that are not reflected on 25 the April version? 1063 1 A. There are a few changes, a few articles that were listed 2 and perhaps now have page numbers. And I've added two items 3 to the CV that are relevant to this case. I now cite my 4 Gallery of CSS Descramblers as a publication and also a page 5 on the CSS descrambling algorithms. That's now included in my 6 vitae as a publication. 7 MR. ATLAS: If I may approach again. I'd like to 8 move Dr. Touretzky's CV into evidence. 9 THE COURT: Any objection, Mr. Mervis? 10 MR. MERVIS: Plaintiffs have no objection, your 11 Honor. 12 MR. ATLAS: If I may approach again, your Honor. 13 THE COURT: Go ahead. 14 BY MR. ATLAS: 15 Q. Dr. Touretzky, did you prepare a declaration in this case? 16 A. Yes, I did. 17 Q. I'd like to show the witness what we've marked as 18 Defendants' Exhibit BBC and ask the witness if that's the 19 declaration that you prepared and signed in this case? 20 A. Yes, it is. 21 MR. ATLAS: I'd like to move Dr. Touretzky's 22 declaration into evidence as well, your Honor. 23 MR. MERVIS: Your Honor, I suppose to the extent that 24 this won't help illustrate his testimony, I have an objection, 25 which is -- and it may be premature -- which is that the 1064 1 testimony is going to be cumulative from what you heard from 2 Dr. Felton. 3 Notwithstanding that, your Honor, I do have an 4 objection in specific to paragraph 7 of this declaration. The 5 grounds for that objection are as follows: Relevance, 6 hearsay, best evidence, and that, in fact, it does not involve 7 any expertise whatsoever. 8 THE COURT: What do you say, Mr. Atlas? 9 MR. ATLAS: I think Dr. Touretzky will be testifying 10 as an expert. I think he can opine on the matters he has in 11 paragraph 7, but I have no objection if the Court takes it for 12 everything other than paragraph 7. I think it will help for 13 the Court. 14 THE COURT: Dr. Touretzky, I take it that you're 15 telling us that the statements made in this document are true. 16 THE WITNESS: Yes. 17 THE COURT: Correct? 18 THE WITNESS: Correct. 19 THE COURT: All right. I will receive paragraphs 1 20 through 6 simply as a way of shortcutting his direct testimony 21 in open court. I do note that to the extent he expresses 22 opinions on legal subjects, I'm not going to consider those 23 opinions as evidence. Obviously some of these things are 24 matters for me to decide. 25 The one that caught my eye particularly is the 1065 1 assertion at the beginning of paragraph 3 that "source code is 2 expressive speech meriting the full protection of the First 3 Amendment." That is ultimately for me to decide. 4 This witness is not here as a law professor, but, 5 obviously, he may have things that are relevant to my 6 determination of that question to say. 7 MR. ATLAS: That's what I would like to focus this 8 witness' testimony on today, especially his gallery of CSS 9 descramblers which he's testified about. 10 BY MR. ATLAS: 11 Q. Dr. Touretzky, do you have a website? 12 A. Yes, I do. 13 Q. Is the Gallery of CSS Descramblers that you testified a 14 moment ago posted on this website? 15 A. Yes. I have an extensive website about DeCSS, and one 16 portion of that website is the Gallery of CSS Descramblers. 17 THE COURT: I should note, Mr. Atlas, that 18 Defendants' Exhibit BBC does include internal references to 19 attached exhibits, which are not attached, one of which is a 20 printout of the gallery. 21 MR. ATLAS: I've broken them down separately, and I'm 22 going to take them as individual exhibits. 23 THE COURT: All right. 24 MR. ATLAS: If I could approach the witness. 25 THE COURT: All right. 1066 1 MR. ATLAS: It's Exhibit CCN. 2 THE COURT: CC. 3 MR. ATLAS: N. 4 THE COURT: N. 5 MR. ATLAS: Just so we're clear, I'd like to show the 6 witness -- the witness provided me this morning with a copy 7 that has various colors on it. He said it would be easier as 8 he goes through and describes what the gallery is if he can 9 refer to this one. The copies I have are black and white. Is 10 there an objection to my showing him the colored version? 11 MR. MERVIS: No. 12 BY MR. ATLAS: 13 Q. Dr. Touretzky, can you tell us what Defendants' Exhibit 14 CCN is? 15 A. This is a printout of the main page of my Gallery of CSS 16 Descramblers. 17 Q. Now, what is the Gallery of CSS Descramblers? 18 A. The gallery is a presentation of the CSS decryption of 19 algorithms in various forms in an attempt to show that these 20 forms are equivalent and that it's not possible to 21 discriminate between them. 22 Q. Why did you create the gallery? 23 A. Well, in reading reports of this case, it seemed to me 24 that the public discussion of some of these issues having to 25 do with legal status of computer code as speech could benefit 1067 1 from pointing out some of the things I try to point out in the 2 gallery. And since no one else seemed to be discussing it, I 3 thought I would create this website. 4 Q. If we could referring to Defendants' Exhibit CCN, which is 5 your gallery. If you could take us through, I guess, 6 beginning with the reference here to the anonymous C source 7 code, and just explain what it is you were trying to do. 8 A. Sure. So, the two points that the gallery tries to make 9 are, first of all, that source code has expressive content, 10 and, secondly, that you can't distinguish between different 11 forms of description of an algorithm, whether they're in 12 computer language like C or English or some other notation. 13 So the first two items in the gallery -- the first 14 item is the anonymous C source code, which is the source code 15 that was posted to the LiViD mailing list in October of 1999. 16 This is one of the items that is covered, I believe, by the 17 injunction issued by this Court. 18 The second item is another version of the same 19 algorithm. This.item is called CSS descramble.c, and this 20 comes from the CSS-auth software package that was authored by 21 Derek Fawcus. And the point of providing these two versions 22 is to show that it is possible to have different 23 implementation of an algorithm, and by looking at these 24 different implementations, one can gain knowledge that one 25 couldn't get from looking at other decryptions of the 1068 1 algorithms. 2 So, for example, if you compare CSS descramble.com 3 against the anonymous C source code, one of the things you see 4 is that Mr. Fawcus used fewer tables in his implementation, 5 and he unrolls one of the loops. 6 So, what he's telling us is, hey, there's more than 7 one way to implement this algorithm. You don't need to use so 8 many tables. You can use fewer tables. You don't have to do 9 this operation in a loop. Maybe it's more efficient to unroll 10 the loop and have the individual operations stated as separate 11 lines of code, and that's a message that Mr. Fawcus is 12 expressing in his chosen implementation of the algorithm. 13 You can't get that message, except by looking at the 14 source code and reading the source code. So that's the only 15 thing that can be expressed in the source code. 16 Q. By the way, if you could turn the page to page 2 of the 17 gallery. And can you explain to us what the first item on 18 page 2 is? 19 A. Yes. I should say that the next section of the gallery, 20 the items on page 2, they all have to do with the issue of 21 different forms the code can take and the fact that this Court 22 has enjoined some forms and not others, and I was puzzled by 23 that. It struck me as unusual that the Court would decide 24 that it was okay to ban, say, the C source code, but the Court 25 declined to ban discussion of the algorithm. 1069 1 Q. Why? 2 A. Because in my mind, these things are all equivalent, and 3 so I was wondering what your Honor was thinking when he 4 decided that this would be an effective remedy and why the 5 MPA's lawyers thought they won something when they got this 6 injunction because to my mind this didn't -- I couldn't make 7 sense of it. 8 So in an attempt to explore what the reasoning of the 9 Court and the plaintiffs' lawyers might have been, I decided 10 to investigate different forms that the code could take and 11 look at the implications of that. 12 So I began with the first two exhibits, which we've 13 already discussed, which are the actual C source code, but the 14 LiViD version and the CSS-auth version, which I believe have 15 been enjoined. 16 Now, it seemed to me that it could be that The 17 Court's reasoning was that these are files that could be fed 18 to a C compiler, whereas an English description, which the 19 Court declined to enjoin, could not be fed to a C compiler. 20 So if that was the Court's reasoning, what I've shown in this 21 exhibit on the top of page 2 is a screen dump of the CSS 22 descrambling code. So this is the CSS code written as a 23 binary image file. So in this form, the code cannot be feed 24 to a C compiler, and, perhaps, therefore, it would fall into 25 the category of protective speech. 1070 1 But if that was the case, the problem here is that 2 any human could look at this image and just type the 3 characters that they see into a text editor, and then they 4 would have the C source code again. Furthermore, there are 5 today, there are OCR programs. OCR stands for optical 6 character recognition, which can take a binary image file like 7 this and turn it automatically into text form. 8 So it's not clear to me then if these binary images, 9 if these pictures of the source code -- that's what they are, 10 pictures -- if the pictures themselves are really illegal 11 circumvention devices under the DMCA, but since they can't be 12 feed to a C compiler directly, maybe they're protected speech. 13 Q. Did you reach any conclusion in terms of your examination 14 of this particular description of the code? 15 A. Well, my conclusion is that if these images are protected 16 speech, then really the Court has provided the plaintiffs no 17 protection at all because it was trivial to turn these back 18 into text files. So, since the plaintiffs seem to think that 19 they won something with this injunction, it must be that these 20 images are not protected speech. They must be circumvention 21 devices. 22 Q. Why is that? 23 A. Well, because they're trivially convertible into 24 circumvention devices by looking at the picture and typing the 25 words. 1071 1 Q. Moving down to the second item. 2 THE COURT: So, is the net of your view that if the 3 preliminary injunction decision was in substance correct, as a 4 matter of law, the problem is that the injunction is far too 5 narrow? Is that one way of putting your conclusion? 6 THE WITNESS: Yes; I think that's correct. 7 THE COURT: Okay. I thought that's where you were 8 going. 9 BY MR. ATLAS: 10 Q. The second item on page 2. 11 A. The second item is a file called new-language.txt, and I 12 believe that you have a printout of that that you can 13 introduce as a separate exhibit. 14 Q. Actually before I get to that, I'd like to explore 15 something the judge had just asked you if you don't mind. 16 Why, in your view, would it be too narrow, the injunction, as 17 it currently stands? 18 A. Well, I think what the Court was trying to do in the 19 injunction is prevent people from obtaining technology that 20 could be used to decrypt DVDs without the authority and 21 permission of the motion picture industry, and simply 22 restricting a particular form of the C source code will not be 23 an effective way of preventing that. 24 And so if you're going to be effective about it, if 25 you're really going to stop people from doing it, you've got 1072 1 to stop them from hearing the message that the defendants were 2 trying to convey. You've got to stop them from any 3 description of the algorithm, which could be used to construct 4 one of these devices. 5 Q. Would that say that could include an oral conversation 6 that one would have in sufficient detail in discussing the 7 DeCSS program? 8 A. Yes, it would. I could certainly sit down with someone 9 who had the appropriate degree of knowledge of C programming 10 and of encryption technology, and with an oral conversation -- 11 assuming that person had a good memory -- they could acquire 12 the knowledge necessary to reconstruct this program, yes. 13 Q. Let's draw your attention to back to page 2, which you 14 said I might have a copy of that. One moment. 15 MR. ATLAS: If I may approach, your Honor. This is 16 Exhibit CCQ, Defendants' Exhibit CCQ. 17 BY MR. ATLAS: 18 Q. What is that document, Dr. Touretzky? 19 A. This is a printout of the second exhibit on page 2 of the 20 gallery, which is called new-language.txt. And what I did 21 with this exhibit, I tried to further explore this issue of 22 the legal status of computer code. So if these pictures of 23 the code in the previous gallery exhibit, if the Court had a 24 problem with distributing those pictures of the C code, maybe 25 that wasn't -- maybe that wasn't protected speech. 1073 1 If they have a problem with that, maybe the problem 2 is that these are pictures of code expressed in a language for 3 which there is a compiler. There are C compilers available 4 for free, and so if you had a picture of the code, you could 5 type the code in, just literally typing what you see on the 6 screen, and then you can run it through the C compiler. 7 So if that was the problem, then maybe it would be 8 okay to describe the algorithm in another computer language 9 that had all the precision of C, but for which there did not 10 exist a compiler. So what I did is I created a computer 11 language, which has the same semantics as a C programming 12 language, but perhaps has a different syntax, and I 13 recompressed the DeCSS algorithm in this new language. 14 Now, there is no compiler for this language, and so 15 it could be that the Court would find that since you can't 16 feed this to a compiler, because there isn't a compiler, then 17 this form of the algorithm might be protected speech, might be 18 okay. But I have to say I'm pretty worried about this because 19 with this version of the code out on the web, I have no way of 20 preventing another person from writing a compiler for my 21 language. 22 And having done that, they will then make me in the 23 position of having provided code that's now compilable even 24 though I didn't create the compiler, and, in fact, there was 25 no compiler when the code was originally written. So I think 1074 1 this raises problems, too, that expressing the code in a 2 language that is as precise as C, but for which there is no 3 compiler, really you can't give that the same protected status 4 as other kinds of speech. You have to view it as a device 5 because it could become compilable at any time. 6 Q. Moving down to the next item on your gallery. I think I 7 have a copy of that. 8 MR. ATLAS: It's Exhibit CCO. 9 Q. Is this the next step in your analysis, Dr. Touretzky? 10 A. Yes. This is a copy of the file plain-english.html. 11 Q. What was the purpose of doing that? 12 THE COURT: I'm delighted, Doctor, that this is plain 13 English to someone. 14 THE WITNESS: I did my best. 15 THE COURT: I wasn't finding fault. 16 A. This is the third gallery item on page 2 of the gallery 17 printout. So what I tried to do here was, again, to further 18 explore this issue of the status of source code, and if it's 19 the case that the previous file, my new-language.txt -- if 20 that was, perhaps, not deserving of First Amendment protection 21 because somebody could write a compiler for it -- and the 22 Court had already refused to enjoin mere discussion of the 23 algorithm -- I know that from reading the transcript of the 24 preliminary injunction hearing and from reading the memorandum 25 that the judge subsequently wrote -- it seemed to me that the 1075 1 next thing one could consider was translation of the algorithm 2 in English. 3 Now, there's two ways to do that. The most 4 mechanical way is to simply translate the C code directly into 5 English and make no other changes, and I decided not to bother 6 with that. I'll leave that as an exercise for an 7 undergraduate, but I thought a more interesting way to do it 8 would be to translate the C code into English, but, in 9 addition, include some explanatory text and some editorial 10 comments. And that's what I've done in this version. 11 So if you look at this version, the first couple of 12 paragraphs are introductory text, and then that's followed by 13 an image, which I borrowed from a website in Norway that 14 explains the big picture of how DVDs -- how movies on DVDs are 15 descrambled. And after that image, there's another paragraph 16 of text where I explain what the image conveys. 17 And following that, the remainder of this document is 18 my translation of the C code into plain English, and that 19 translation is done pretty much line by line, but it contains 20 certain informalities of language and minor editorial 21 comments, which are the kinds of things that a human could do, 22 but a computer program typically could not do. 23 Q. And did you reach any conclusions in preparing this 24 plain-english.html version? 25 A. My conclusion is that this version is substantially 1076 1 equivalent to the original C source code that was the first 2 very exhibit in the gallery. That any person with a 3 rudimentary knowledge of C would be able to go from this 4 English description back to the C source code and reproduce 5 the C source code flawlessly. 6 So it seems to me that this document ought to be 7 enjoined for the same reason that all the others are, is that 8 it's providing exactly the same information. It's providing a 9 description of the algorithm with enough precision that 10 another person could reproduce the software. 11 Q. Moving down to the fourth item on page 2 of your gallery. 12 It's the english-and-c.html version. 13 MR. ATLAS: I have a copy of that as well. It's 14 Defendants' Exhibit CCP. 15 Q. I'm showing you what's been marked as Defendants' Exhibit 16 CCP. Could you tell us what that is? 17 A. Yes. This is a printout of the fourth exhibit on page 2 18 of the gallery. This is another version of that essay that we 19 just discussed. 20 Q. Why dis you create this version of the essay? 21 A. Well, I wanted to further explore the issue of plain 22 english descriptions being protected. Given that the Court 23 declined to enjoin discussion of the algorithm in plain 24 english, it could be that the previous exhibit, 25 plain-english.html, perhaps that was protected speech, and if 1077 1 it was protected speech, then what would happen if one simply 2 made a little more explicit the mapping between the English 3 description of the algorithm and the C code. 4 So for people who haven't yet acquired a rudimentary 5 knowledge of the C programming language, I thought I could 6 give them a little help in converting the English back to C. 7 And so what I've done is, beginning on page 2 of this 8 nine-page document, for each English paragraph that 9 corresponds to a line of C code, I have presented an inset 10 that gives the translation of that English into C. 11 And in the original document, that inset is done in a 12 light blue background to make it visually distinct, and it's 13 also indented. The indentation surveyed the copying process 14 here, but the blue background did not. 15 So in going through this, you see for each line of 16 the algorithm described in English, there is the corresponding 17 expression of that algorithm in C code. All I'm doing is 18 translating my English into statements in the C language. So 19 an interesting thing about this document is that it contains 20 within it the source code for the CSS descrambling program, 21 which the Court has enjoined. 22 But you cannot feed this file to a C compiler because 23 that C source code is interspersed with both English and with 24 HTML formatting commands, which provides the blue background 25 and information and so on. So you can't compile this thing, 1078 1 but you can read it, even if you're not a very good C 2 programmer. But what you learn from reading this document 3 would allow you to recreate with perfect accuracy the original 4 descrambling algorithm. 5 Q. When did you put up the gallery? When did you post the 6 gallery? 7 A. In March of this year. 8 Q. Did anyone involved in this case suggest or encourage you 9 to create this gallery that you've been testifying about? 10 A. The creation of the gallery was my own idea. I did 11 discuss it with my graduate students, who thought it was a 12 pretty good idea and that I should go ahead with it. 13 Q. Why did you create and post it? 14 A. I wanted to spare public discussion of these issues. This 15 case raises what for me are very serious concerns about the 16 future of computer science and my ability to function as a 17 computer scientist. 18 Q. Why? 19 A. Because what the plaintiffs are trying to do is carve a 20 hole in the First Amendment and say, this kind of speech you 21 can engage in and that kind of speech you can't. And as a 22 computer scientist, this is my livelihood. I've been 23 programming computers since I was 12 years old, and I'm very 24 concerned when events take place that threaten my ability to 25 express myself. 1079 1 THE COURT: Mr. Mervis. 2 MR. MERVIS: I move to strike that portion of the 3 answer which Dr. Touretzky purports to describe what the 4 plaintiffs are trying to do in this case and to the extent he 5 is, also, offering opinion about legal matters. 6 MR. ATLAS: I think it's the witness' perception of 7 how the plaintiffs' position in this case impacts what it is 8 that he does and how he communicates with his peers and his 9 students, and I think it's acceptable testimony. 10 THE COURT: To the very limited extent of the motion, 11 it's granted. I'm not taking Dr. Touretzky's view of the 12 First Amendment as evidence. His testimony presupposes a view 13 of the scope of the First Amendment on which he is not an 14 expert, which is not a proper subject for testimony and which 15 is ultimately for me and appellate courts to decide. 16 There is no doubt that yelling "fire" in crowded 17 theatre is something that is problematic and probably impedes 18 expression by certain twisted individuals. I'm not drawing a 19 one-to-one analogy. You know how dangerous one-to-one 20 relationships are this morning, but you see the point. 21 MR. ATLAS: I think it explains why the witness is 22 here, but I may be -- 23 THE COURT: I understand that. I understand why he's 24 here. 25 BY MR. ATLAS: 1080 1 Q. By the way, the different ways that you have approached 2 communicating the DeCSS source code or the DeCSS code, would 3 those be the types of communication you may -- strike that. 4 These different ways of approaching the DeCSS code, 5 are those ways that you communicate with the people in -- your 6 peers, the other professors, the other faculty members? 7 A. Yes. We mix these techniques freely, and even on the same 8 page of a paper, you might have descriptions in English. You 9 might have mathematical equations, bits of code in a 10 acceptable programming language, bits of code in a language I 11 just made up because it's particularly convenient for trying 12 to express what I'm trying to express. We use all of these 13 forms of expression all the time. 14 Q. Turning to the last page of your gallery, page 3 -- 15 MR. ATLAS: Actually before we get to that, what I'd 16 like to do, your Honor, at this time is move into evidence the 17 gallery, which is Exhibit CCN and then the three different 18 types of entries on the gallery, which I think we've marked as 19 CCO, CCP and CCQ. 20 MR. MERVIS: No objection. 21 THE COURT: They're received. 22 (Defendants' Exhibits CCN, CCO, CCP, and CCQ 23 received in evidence) 24 BY MR. ATLAS: 25 Q. Turning on the top of page 3, the Cryptanalysis of CSS. 1081 1 Can you explain what the entry of this refers to? 2 A. Yes. This entry is actually a collection of writings by 3 Frank Stevenson, rather than a single document. And in those 4 writings, which I believe have been entered in this case 5 already -- in those writings, Mr. Stevenson talks about 6 various cryptographic attacks on pieces of the CSS encryption 7 scheme, and some of these documents are English text and some 8 of them are little bits of C code, which illustrate the point 9 he's trying to make. 10 The interesting thing about this exhibit in the 11 gallery is that this is not -- the software provided here 12 would not allow you to immediately decrypt the DVD. What's 13 here is software that illustrates the effectiveness of the 14 attacks that Mr. Stevenson developed. And, yet, again, to 15 provide the plaintiffs' effective protection, one would have 16 to somehow prevent the discussion of this material because in 17 the hands of a knowledgeable person, it could be used to 18 access the material on the DVD. 19 Q. The next item, the DeCSS T-shirt, why did you post this on 20 the website on the gallery? 21 A. Well, this is a photograph of a T-shirt that's offered for 22 sale by an outfit called CopyLeft, and I purchased one of 23 those shirts myself. And the point of including it here is it 24 seems to me that there is some confusion among all the parties 25 in this case about whether something is speech or not. 1082 1 And my reaction is if you can put it on a T-shirt, 2 it's speech. And so the point of showing the T-shirt was to 3 illustrate that and, also, to raise the question if this 4 T-shirt, itself, would have to be prohibited, then I wonder 5 what would happen to me if I wore the shirt in public. 6 Wearing the shirt in public could, perhaps, be interpreted as 7 engaging in trafficking a circumvention device. 8 So if one really wants to afford the plaintiffs the 9 protection that they seek, I think I would only be able to 10 wear my shirt in the privacy of my own home and must not go 11 outdoors with it. 12 Q. The final item on here, can you tell us what this refers 13 to? 14 A. Yes. This is an image file. It was constructed by Andrea 15 Gnesutta, who was the winner of a contest for the best way to 16 distribute the DVD source code. And what she did was she 17 created an image file that looks like a bumper sticker. It 18 says, DVD Source Code Distribution Contest. But inside this 19 image file is the source code. And although you can't see the 20 source code, when you display the image on your web browser, 21 the mere fact that this image appears on this screen -- if 22 you're looking at a website and this image comes up means that 23 the source code has been downloaded onto your computer. 24 And Ms. Gnesutta gives the technique necessary to 25 extract the source code from the binary image file. It's a 1083 1 couple of simple operations can pull out those bytes, and then 2 one can easily obtain the text file from that. 3 (Continued on next page) 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 1084 1 Q. So, summing up, looking at all the different items you 2 have on your gallery of CSS -- do you have a difference -- 3 strike that. 4 Do you have an opinion on the differences, if any, 5 between them? 6 A. My opinion is that there's no effective difference between 7 any of these forms in that if you want to accomplish what the 8 Court set out to do in its preliminary injunction, you have to 9 enjoin all of these forms because if any one of them is 10 allowed, then it's a trivial matter to transform it into any 11 of the others. They're all interconvertible. 12 Q. Do you have any opinion on the effect within your field, 13 the computer science field, if the Court were to issue an 14 injunction first over the DeCSS executable resource code? 15 MR. COOPER: Objection, your Honor; relevance. 16 THE COURT: Overruled. 17 A. Well, yeah, I see this as having a chilling effect on my 18 ability as a computer scientist to express myself. In 19 comparison, I know I could publish the formula for LSD using 20 chemical notation and even though I'm not allowed to possess 21 LSD, no one would try to stop me from publishing the formula. 22 And I could publish the schematic for the timing 23 device for a bomb, and even though I'm not allowed to possess 24 a bomb, no one would try to stop me from publishing those 25 schematics. 1085 1 The language that -- in which I try to express myself 2 is the computer language and if the Court upholds this 3 injunction, what would happen is that certain uses of computer 4 language, my preferred means of expression would illegal. 5 They'd be enjoined. And that means that anything I do from 6 now on has the potential to be ruled illegal. 7 And furthermore, what I've tried to show with the 8 gallery, even if I just decided never to try another line of 9 computer code again, I wouldn't be any better off because 10 simply writing English, if I wrote it with the same precision 11 as I used in my plain English on an HTML file would accomplish 12 the same thing. 13 So, I could merely by using the kind of precision 14 that I've been trained to use as a computer scientist simply 15 by expressing myself that way, I could be opening myself up to 16 legal liability. 17 THE COURT: Dr. Touretzky, do people in your field of 18 endeavor sometimes make bits of hardware, computer hardware? 19 THE WITNESS: Yes, they do. 20 THE COURT: Are there circumstances in which it might 21 be helpful to a computer scientist to make a bit of hardware 22 as to which somebody else holds the patent? 23 THE WITNESS: Yes, I think so. 24 THE COURT: O.K. Mr. Atlas? 25 Q. Dr. Touretzky, would a declaration by the Court that the 1086 1 CSS descriptions are illegal have a similar chilling effect 2 compared to an injunction? 3 MR. COOPER: I have the same relevance objection, 4 your Honor. 5 THE COURT: Overruled. 6 A. Well, it would certainly. It would certainly make me 7 afraid to publish things like that; yes. Any indication by 8 the Court that this was illegal would concern me; yes. 9 Q. By the way, are you being compensated for your role as an 10 expert in this case? 11 A. No, I'm not. 12 Q. Why are you rendering services if you're not being 13 compensated here? 14 A. I think the issues being decided here have tremendous 15 importance for the field of computer science and it's actually 16 I consider it a privilege to be allowed to come here and 17 express my views and hope to influence the outcome. 18 Q. One other thing, Doctor. I believe in your deposition an 19 article that you had prepared was marked -- or actually I 20 believe it's Exhibit 2 at your deposition, we've marked it as 21 Exhibit BBE, it's an article entitled "Source Versus Object 22 Code." 23 MR. ATLAS: May I approach, your Honor? 24 THE COURT: Yes. 25 Q. Dr. Touretzky, if I may, why did you write this article? 1087 1 A. I wrote this article because after reading various 2 depositions in this case, it occurred to me that the both 3 sides were grappling with the issue of source versus object 4 code. And I could see that several of the other defense 5 witnesses had tried in depositions to explain that source and 6 object code are really the same thing, that there isn't any 7 rational way to distinguish one from the other, but I'm not 8 sure that the lawyers were getting it. 9 And I think it's kind of hard to do that in 10 deposition. So, by constructing this essay, including the 11 illustrative appendices at the end, I hoped to assist both 12 sides in understanding this. 13 Q. That's actually what I'd like to refer your attention to 14 appendices beginning on page 4 of 6 of your article. Appendix 15 A, Appendix C, and D. I was wondering if you could explain 16 the relationship between these appendices? 17 A. Yes, these appendices are four different forms of the same 18 program. I'd like to go through them in order and then back 19 up. 20 So, starting with Appendix A, Appendix A is a source 21 file written in the C language for a simple program that 22 multiplies the numbers 1 through 5 together and then prints 23 out the result. In mathematical results, we'd say the program 24 computes the quantity 5 factorial. 25 I should say that Appendix A is the same form as the 1088 1 version of the DeCSS code that the Court has enjoined. The 2 source code written in the C program language just like the 3 anonymous C source code that the Court has enjoined. 4 What I next did was I compiled this source code. 5 Now, a compilation is actually a complicated process with a 6 number of different steps which are typically invisible to the 7 lay person, but computer scientists know about them. 8 So, when you feed the C source code to a C compiler, 9 let's say GCC, the new C compiler, it's a very commonly-used C 10 compiler. What the C compiler does it translates that source 11 code into a language called RTL, which is an abstract machine 12 language. Then it translates the RTL into assembly language 13 for whatever processor you're compiling for. In this case, I 14 was compiling for a spark processor for the sun work station 15 in my office. 16 Then an assembler takes the assembly language code 17 and translates that into machine executable code. So, what we 18 have in Appendix B is an intermediate stage in that process. 19 What I asked the C compiler to do was when it 20 translated the C to RTL, and then the RTL into assembly 21 language, I asked it to tell me and give me back the assembly 22 language, the intermediate product. So, from the C compiler's 23 point of view, Appendix B is the object code is a result of 24 what the C compiler does to the source code, but from the 25 assembler's point of view, assembler being the next guy in 1089 1 line that's going to process this, the assembler used this as 2 a source code, but as the assembler is going to turn it into 3 by their machine instructions. 4 So, if we proceed to Appendix C, what I next did was 5 I told the C compiler, go ahead and compile this program, and 6 instead of stopping at the assembly language level, stop at 7 the level of the binary machine code. What's called the .0 8 file or the object file and then I took that .0 file and I 9 displayed it using a utility program that displays binary 10 files. 11 And so, what you see are in Appendix C, the column of 12 numbers on the left, the leftmost column are the equivalent of 13 line numbers in a deposition. They're basically keeping track 14 of the position in the file and the remaining columns of 15 numbers are hexadecimal encryption codings of the binary bits 16 that make up the object file. 17 Finally, in Appendix D, I took that object file and 18 fed it to a program called a disassembler. And the object -- 19 the disassembler translated the object file back into symbolic 20 assembly language form, and that's what we see in Appendix D. 21 Now, so the point I want to make here is that the 22 same ideas that were expressed in my statement to you that 23 five factorial is the product of the numbers 1 through 5 are 24 also expressed in all four of these appendices and they're 25 expressed in ways that are transparent to a computer 1090 1 scientist. And I'd like to show you that starting at Appendix 2 D, if you look at the line labeled 20 in Appendix D, it's 20 3 and a colon. 4 That line in the middle of that line, you see the 5 letters CMP, which stands for compare. So, this line is a 6 line that's telling the computer to compare two numbers. 7 And if you look to the right of the CMP, what it's 8 saying is compare the contents of register 00 with the number 9 5. So, this is a piece of symbolic assembly language code 10 which computer scientists can read just fine. 11 And if you look to the left of the CMP, you'll see 12 four groups of numbers. These are hexadecimal numbers. The 13 numbers are 80822005. This is the encoding of that compare 14 instruction in the binary machine language of the spark 15 processor. 16 So now, if you go back to Appendix C, and if you look 17 at in Appendix C, the line labeled 220, so it's about 10 lines 18 at the -- down from the top, the line labeled -- it's actually 19 0000220. 20 And if you look about halfway in that line, you'll 21 see the sequence 80822005. So, this is -- again, this is the 22 binary machine instruction that's that compare instruction 23 that we saw in Appendix D. 24 Now, we can go back once more to Appendix B. And if 25 you look at Appendix B, this is again symbolic assembly 1091 1 language. And so, on the left side of Appendix B, you'll see 2 some labels. There's a label main and a little bit further 3 down there's a label that says .LL3 on the left-hand side. 4 And if you look two lines below that .LL3, we see 5 again the compare instruction. CMP percent 00, 5. It's 6 comparing the contents of register 00 with the No. 5. We can 7 go back one more step and look at Appendix A, my C source 8 code. 9 And in Appendix A, about 5 lines down, there's a four 10 alone. It begins with the word "four" and then it says: Left 11 (I equals 1; I less than 6). That "I less than 6" that is 12 exactly what that compare instruction was turned into by the C 13 compiler. 14 So, the compare instruction is the expression of this 15 concept, "I less than 6" in the machine's assembly language. 16 Q. So, do I understand that these four appendices are 17 different ways of expressing the same idea? 18 A. Yes, it's exactly the same idea and you might ask, well, 19 why does "I less than 6" and yet in the assembly language 20 code, we are comparing the register against the No. 5. And 21 the reason is that it reflects a certain strategy used by the 22 C compiler to encryption code this 4 loop and that the only 23 way you could know about that strategy is to read the assembly 24 language. 25 And I hope what I've shown is that it's not that hard 1092 1 to do that. 2 Q. In terms of taking one of the items? 3 THE COURT: We are all going to get into 4 post-assembly language this afternoon. 5 THE WITNESS: Excuse me? 6 THE COURT: We are all going to get to post-assembly 7 language this afternoon. 8 Q. Doctor, can you tell me the skill level it would take to 9 take any one of the identities that appear on your gallery of 10 CSS descramblers and turn that into any one of the other 11 identities that appear on your gallery of descramblers, what 12 skill level would be involved? 13 MR. MERVIS: I object to the form of the question. 14 THE COURT: Overruled. 15 I think we can understand it. 16 A. Yes, since I've been involved in teaching computer science 17 for quite a few years, I'd say that anyone who had taken an 18 undergraduate level C programming course should be able to 19 convert between any of these forms. 20 MR. ATLAS: The final thing I'd like to move into 21 evidence, Exhibit BBE, which is the article and the appendices 22 that the doctor was just talking about. 23 MR. MERVIS: No objection, your Honor. 24 THE COURT: Received. 25 (Defendant's Exhibit BBE received in evidence) 1093 1 MR. ATLAS: I have no further questions for Dr. 2 Touretzky. 3 THE COURT: Thank you. 4 Any cross-examination, Mr. Mervis? 5 MR. MERVIS: Unless the professor is going to give us 6 all a quiz, the plaintiffs have no questions. 7 THE COURT: Well, Dr. Touretzky, let me just tell you 8 that this was illuminating and important. I was hoping we 9 were going to hear something like this through the whole 10 trial. I appreciate your having come. 11 THE WITNESS: Thank you, your Honor. 12 THE COURT: Thank you. 13 (Witness excused) 14 THE COURT: The witness is excused. [ deletions... ] 1153 12 THE COURT: We're going to cut the baby in half. 13 We're going to make it August 8th. And a 35-page limit. And 14 let me give you a little guidance. 15 I haven't really had a chance during the course of 16 the last week to go back over the opinion in January in light 17 of the evidence at trial. Obviously, there was no record 18 before me in January. There was nothing from the defense at 19 all, and there was very little from the plaintiffs compared to 20 what's there now. 21 But my tentative sense, from having listened this 22 week and last with a lot of care, is that probably nothing 23 much has changed with respect to the analysis of the DMCA 24 itself. I may be wrong in that, but that's my gut reaction 25 here. 1154 1 I think one thing probably has changed with respect 2 to the constitutional analysis, and that is that subject to 3 thinking about it some more, I really find what Professor 4 Touretzky had to say today extremely persuasive and 5 educational about computer code. 6 Now, of course, which way that cuts is another 7 matter, but I just don't think it's likely ultimately to prove 8 tenable to say that computer code of any kind has no 9 expressive content, source or object. Which then gets you to 10 the question of how then do you deal with it under the First 11 Amendment. I found myself wondering and I'm thinking out loud 12 here, because I really want both sides to understand where my 13 mind is going on this, because then they're going to focus 14 more closely on what concerns me. 15 I'm really in doubt as to whether saying that 16 computer code is constitutionally protected speech goes very 17 far toward answering the questions in this case. 18 In the past, the Supreme Court has dealt with 19 questions involving the regulation of expression essentially 20 by fitting the particular kind of expression into various 21 pigeon holes, fighting words, obscenity, conduct as contrasted 22 with speech, commercial speech, and they've developed over the 23 years various kinds of definitions that to one degree or 24 another separate those different categories. 25 I'm really not sure that paradigm fits this case. Or 1155 1 if it does, where exactly -- which is the right pigeon hole 2 for this case. I reread the draft card burning case over the 3 lunch hour, and the Supreme Court there said that although the 4 burning of the draft card was obviously intended as 5 expression, the government had the right to prohibit it, to 6 criminalize it because of the governmental interest in running 7 the draft. I remember what I thought about that in 1968. Not 8 that I'm telling anybody. 9 The speech conduct distinction they draw in that 10 case, I think, one fairly can say is one that is debatable, 11 and they put the expressive draft card burning purely into the 12 conduct box, so that it could be regulated even though it was 13 obviously expressive. There's some kind of analogy here. So 14 I share those views. They're very unformed. They're very 15 tentative. But I guess what I'm intending to communicate by 16 this is that rhetoric is not really likely to cut it very 17 much. I'm really very interested in the guts of this, and for 18 what it's worth, there you have it. 19 Okay. Anything else? I do think it appropriate to 20 thank all counsel because I know what a huge personal effort 21 getting ready has been. Nobody should think that I was ever 22 unmindful of that, and I appreciate that for everybody. 23 Okay, our business is concluded. I'll hear from you 24 on these other odds and ends of documents from you tomorrow. 25 Now, to whatever extent I don't have the exhibits, I need 1156 1 them. 2 MR. GOLD: Thank you, very much. 3 MR. ATLAS: Thank you, your Honor. 4 THE COURT: Thank you, folks. 5 o0o 6