![[Join EFF]](/Icons/home/joineff.gif){kind=icon}
![[Act Now]](/Icons/home/actnow.gif){kind=icon}
![[Sign Up]](/Icons/home/signup.gif){kind=icon}
![[About EFF]](/Icons/home/abouteff.gif){kind=icon}
See related files:
http://www.eff.org/IP/Video (EFF Archive)
http://cryptome.org/cryptout.htm#DVD-DeCSS
(Cryptome Archive)
http://www.2600.com/dvd/docs (2600 Archive)
http://eon.law.harvard.edu/openlaw/dvd/ (Harvard DVD OpenLaw Project)
1
VOLUME: I
PAGES: 1-95
EXHIBITS: 1-5
UNITED STATES DISTRICT COURT
SOUTHERN DISTRICT OF NEW YORK
Case No. 00 CIV 277 (LAK)(RLE)
UNIVERSAL CITY STUDIOS, INC.; )
PARAMOUNT PICTURES CORPORATION; )
METRO-GOLDWYN-MAYER STUDIOS, )
INC.; TRISTAN PICTURES, INC.; )
COLUMBIA PICTURES INDUSTRIES, )
INC.; TIME WARNER ENTERTAINMENT )
CO., L.P.; DISNEY ENTERPRISES, )
INC.; AND TWENTIETH CENTURY FOX )
FILM CORPORATION, )
Plaintiffs, )
)
VS. )
)
SHAWN C. REIMEREDES; ERIC CORLEY )
A/K/A "EMMANUEL GOLDSTEIN"; )
ROMAN KAZAN AND 2600 ENTERPRISES,)
INC. )
Defendants. )
DEPOSITION OF HAROLD ABELSON,
a witness called on behalf of the Plaintiffs,
taken pursuant to the provisions of the Federal
Rules of Civil Procedure, before Loretta
Hennessey, Registered Merit Reporter and Notary
Public in and for the Commonwealth of
Massachusetts, at the offices of Choate, Hall &
Stewart, 53 State Street, Boston, Massachusetts,
on Monday, July 5, 2000, commencing at 10:49 a.m.
2
1 APPEARANCES:
2
PROSKAUER ROSE LLP
3 (By William M. Hart, Esq.)
1585 Broadway
4 New York, New York 10036
for the Plaintiffs.
5
6
FRANKFURT GARBUS KLEIN & SELZ, P.C.
7 (By Martin Garbus, Esq.)
488 Madison Avenue
8 New York, New York 10022
for the Defendants.
9
10
____________
11
12
13
14
15
16
17
18
19
20
21
22
23
24
3
1 I N D E X
2 Witness Direct Cross Redirect Recross
3 HAROLD ABELSON
4 (By Mr. Hart) 4 82
(By Mr. Garbus) 44 --
5
6
7
E X H I B I T S
8
Exhibit No. Page
9
1 Vita of Harold Abelson. 6
10
2 Notice of Deposition. 10
11
3 Subpoena. 11
12
4 Declaration of Harold
13 Abelson. 17
14 5 Declaration of Michael
Shamos. 61
15
16
17 _______________
18
19
20
21
22
23
24
4
1 P R O C E E D I N G S
2 HAROLD ABELSON, Sworn
3 a witness called on behalf of the Plaintiffs,
4 having been duly sworn, was examined and
5 testified as follows:
6 DIRECT EXAMINATION
7 BY MR. HART:
8 Q. How shall I address you? Doctor? Professor?
9 A. How about Hal.
10 Q. Thank you.
11 Have you ever been deposed before?
12 A. No.
13 Q. Do you have an up-to-date curriculum vitae that
14 you could provide to me?
15 A. Yeah.
16 (Document exhibited to counsel.)
17 Q. Thank you.
18 Are you currently employed?
19 A. Yes.
20 Q. What do you do?
21 A. I teach at MIT.
22 Q. What do you teach?
23 A. Computer science.
24 Q. Is that a full-time job?
5
1 A. Yes.
2 Q. Do you also consult or do anything else in
3 addition to teaching full time?
4 A. Yes, I consult. I do most of my consulting right
5 now for Hewlett Packard.
6 Q. In what areas or area?
7 A. I work on Internet technology and some of their
8 electronic publishing.
9 Q. When you say "Internet technology," could you
10 briefly describe what that means?
11 A. It has to do with how to make an infrastructure
12 that lets people mount services on the Internet;
13 also some more specific technology to make very
14 efficient Internet servers.
15 Q. And what particular areas of experience,
16 knowledge or expertise do you contribute to the
17 area that you just described?
18 A. For MIT or for --
19 Q. No, I'm sorry, in your consulting capacities.
20 A. In my consulting? I was part of organizing
21 Hewlett Packard's Internet Technology Division,
22 part of their general management of what kinds of
23 projects they'll pick, what kinds of, what kinds
24 of services they'd like to develop, and general
6
1 kinds of discussions about what kinds of research
2 Hewlett Packard Laboratories ought to be doing.
3 Q. Let me just get this straight. I'm trying to
4 clarify this, so I apologize if I get it wrong.
5 Is your consulting at Hewlett Packard
6 Hewlett Packard to enable Hewlett Packard to put
7 up material on a web site to promote what Hewlett
8 Packard does, or is it to enable Hewlett Packard
9 to offer technology and/or services to others in
10 connection with the delivery of content onto the
11 web?
12 A. Mostly the second.
13 MR. HART: Let's mark this, if you
14 don't mind, Exhibit 1.
15 (Document marked as Exhibit 1
16 for identification.)
17 Q. Again, I apologize in advance if I misstate it,
18 but in connection with the consulting activities
19 you do for Hewlett Packard, does that involve the
20 delivery of content via the Internet?
21 A. It involves making the infrastructure that would
22 let other people do it.
23 Q. Okay. And by "infrastructure," what are we
24 referring to?
7
1 A. We're referring to, first of all, the hardware
2 infrastructure, the network infrastructure,
3 indexing, directory searching, tracking, content
4 protection in some cases.
5 Q. And when you say "content protection," what do
6 you mean?
7 A. Things like authentication, making sure that only
8 people who are authorized to get at various web
9 sites can do it.
10 Q. What about encryption as a form of protection?
11 A. Some encryption.
12 Q. Is part of the goal generally in the area of
13 content delivery on the Internet to increase the
14 speed at which large files can be delivered?
15 A. Repeat that.
16 MR. HART: Let me have her read it
17 back. She may get it more correctly than I do if
18 I have to restate it.
19 (Question read.)
20 A. Oh, I'm sorry. Yes.
21 Q. And is it fair to say that that's done through a
22 combination of compression technology and broader
23 bandwidth on the Internet?
24 A. Yes, compression technology, broader bandwidth,
8
1 different ways of arranging switching, sometimes
2 storing files in multiple locations. There's all
3 sorts of things.
4 Q. From your vantage point, could you describe how
5 you see the pace of developments in those areas?
6 MR. GARBUS: I'd object unless you
7 feel --
8 THE WITNESS: Excuse me?
9 MR. GARBUS: I'd object unless you
10 feel that's an area of your expertise; if it is,
11 then so state.
12 MR. HART: You've got to keep your
13 voice up, Marty, because I don't know if the
14 court reporter can hear you, I know I can't. It
15 may just be that I'm deaf.
16 A. I couldn't characterize how fast it's getting
17 faster, but it's improving, mostly through
18 different kinds of network technologies.
19 Q. By "network technologies," can you just generally
20 describe what you mean?
21 A. Mostly increased use of optical networking and
22 advanced kind of switching, but I'm not an expert
23 enough to characterize that in any detail.
24 Q. Expert in network design?
9
1 A. Network design.
2 Q. Do you ever use DeCSS?
3 A. No.
4 Q. Do you know what it is?
5 A. Yes, I know.
6 Q. Can you give us a plain English description of
7 what it is?
8 A. My understanding is that it's an algorithm that
9 encrypts the Content Scrambling System that was,
10 that's used on videos.
11 Q. When you say "an algorithm," just for us
12 nonengineering types in the audience, you're
13 referring to a formula?
14 A. A formula, a recipe, what you build a computer
15 program to make.
16 Q. And are you aware whether a computer program
17 embodying DeCSS algorithm, formula, recipe has in
18 fact been constructed?
19 A. I'm told it has. I've never looked at one
20 myself.
21 Q. Do you regard there being a difference between an
22 algorithm and an executable utility?
23 A. A difference?
24 Q. Yes.
10
1 A. A difference for what purpose? Certainly things
2 are the same and things are different.
3 Q. Okay. Could you explain the differences between
4 an algorithm and an executive utility embodying
5 that algorithm?
6 A. An algorithm is generally described at different
7 levels of abstraction than computer utility.
8 Some algorithms you can translate quite directly
9 and automatically into computer utilities. Some
10 algorithms are best expressed as computer code in
11 the first place. And sometimes an actual program
12 will have other details that are different from
13 the algorithm itself.
14 Q. Have you examined the DeCSS algorithm?
15 A. No.
16 Q. Have you ever examined any form of executable
17 utility called DeCSS or embodying DeCSS?
18 A. No.
19 MR. HART: I'm going to mark a couple
20 of things here. Ms. Reporter, if you would,
21 that's No. 2.
22 (Document marked as Exhibit 2
23 for identification.)
24 MR. HART: That will be No. 3.
11
1 (Document marked as Exhibit 3
2 for identification.)
3 MR. HART: We're on the record for
4 just one moment.
5 Q. And I'm reluctant to call you Hal, although you
6 seem like a very nice man, and I'm not
7 necessarily a formal person, but I'm going to
8 show you what's just been marked by the reporter
9 as Exhibits 2 and 3. And my question to you,
10 after you've looked at them -- you can take a
11 moment to look through each of the pages if you
12 like -- my question is: Have you ever seen any
13 of those documents before? And in answering,
14 please indicate which document by exhibit number.
15 (Documents exhibited to witness.)
16 Q. It may help if I give you the ones that have been
17 marked by the reporter because they have the
18 exhibit numbers on them.
19 (Documents exhibited to witness.)
20 A. Okay.
21 Q. Have you seen either Exhibits 2 or 3 before?
22 A. I've seen 3 before. 3 is the subpoena I was
23 served with. 2 looks like it's the same, but I
24 haven't seen it before.
12
1 Q. Fair enough. You can actually hand those back.
2 (Documents exhibited to counsel.)
3 Q. Did you collect any documents to bring with you
4 here today?
5 A. No.
6 Q. Were you aware that the subpoena called for you
7 to produce certain documents as listed on the
8 Schedule A to the subpoena?
9 A. Yes. There was nothing relevant.
10 Q. I'm sorry?
11 A. I didn't see anything relevant in the subpoena.
12 Q. What do you mean by that?
13 A. It asked if I had written things about DeCSS or
14 articles about things. I don't have any.
15 Q. Is it your testimony that you had nothing in your
16 possession or control that responded to any of
17 the areas in the Schedule A of the subpoena,
18 Exhibit 3?
19 A. Yes, that's my testimony.
20 Q. Okay. It's difficult when you use lawyer terms
21 like "relevant." I have to find out what you
22 mean.
23 A. Sorry.
24 Q. How did you first get involved in this case?
13
1 A. I believe either Marty or Ed Hernstadt phoned me
2 and asked if they could use a copy of the
3 deposition I made in another case.
4 Q. Okay. You were deposed in another case?
5 A. I'm sorry, declaration. I'm sorry.
6 Q. And what other case are you referring to?
7 A. This is Bernstein versus U.S.
8 Q. And what was the gist of your testimony in
9 Bernstein versus U.S.?
10 A. It's the same as the declaration I've submitted
11 in this case: It said that computer code is
12 expressive, and people use computer code as a
13 means of expression.
14 Actually, I have it.
15 Q. I'm sorry, just because the record is blind, it
16 didn't see you reaching into your bag and pull
17 out a document.
18 A. I'm pulling out a copy of the declaration I made.
19 Q. You now have in front of you a copy of the
20 declaration that you filed in this case?
21 A. In this case.
22 Q. I prefer you put that to the side right now, if
23 you don't mind.
24 A. Okay. (Witness complies.)
14
1 Q. We're going to go through that in a few minutes.
2 Thank you.
3 Did you have any awareness of this
4 case prior to the time that you were contacted by
5 Mr. Garbus or Mr. Hernstadt?
6 A. Only a minor awareness that it was going on, but
7 I didn't know any of the details of it.
8 Q. What did you know before you heard from Mr.
9 Garbus and Mr. Hernstadt?
10 A. Let's see if I can remember.
11 Well, I knew that some sort of
12 complaint had been filed having to do with DeCSS,
13 but I didn't know the details of it.
14 Q. And were you informed in that first discussion,
15 whether it was a discussion with Mr. Garbus or
16 Mr. Hernstadt, were you informed more about the
17 case at that time during that discussion?
18 A. Only a little bit. What I was told is my
19 declaration in the Bernstein case was relevant to
20 this, could they use it.
21 Q. Did Mr. Garbus or Mr. Hernstadt describe to you
22 their view of the case?
23 A. No, not at that time.
24 Q. Did there come a time when they did?
15
1 A. I asked Marty about it this morning as we were
2 waiting. We haven't discussed things much other
3 than that.
4 Q. Apart from the first call that you got either
5 from Mr. Garbus or Mr. Hernstadt to ask you to
6 put in a declaration in this case, and your
7 meeting here today with Mr. Garbus, how many
8 other contacts or communications have you had
9 with anyone at the defendants' law firm? And by
10 that I mean Mr. Garbus or Mr. Hernstadt or any of
11 their associates or colleagues.
12 A. Perhaps four.
13 Q. I'm sorry?
14 A. Perhaps four. Only with Mr. Garbus and Mr.
15 Hernstadt, mostly having to do with scheduling
16 this deposition.
17 Q. Okay. Were there any discussions concerning the
18 substance of the case?
19 A. We talked a little bit about computer code as
20 means of expression. We talked a little bit this
21 morning about general issues of copyright, but
22 nothing in depth.
23 Q. Have you followed the legislative progress and
24 enactment of the Digital Millennium Copyright
16
1 Act?
2 A. Yes, I have.
3 Q. Have you ever submitted any papers, views,
4 testimony or the like in connection with the
5 legislation, its enactment or any rule making
6 related to that law?
7 A. Nothing formal. On the other hand, I am a
8 constituent of Barney Frank's and I had
9 correspondence with him when he was on the
10 Judiciary Committee when that law was before the
11 Judiciary Committee.
12 Q. And this correspondence included your views about
13 the proposed enactment?
14 A. Yes, it did.
15 Q. You say your full-time job is teaching about
16 computers and computer programs at MIT?
17 A. Teaching and research.
18 Q. Can you generally describe for us what specific
19 areas within computers and computer programming
20 you teach or research in?
21 A. I do research in artificial intelligence and in
22 general to the interaction between the artificial
23 intelligence and computing for scientific and
24 engineering uses.
17
1 I teach -- for many years I taught
2 MIT's major introductory programming course. And
3 more recently I've been teaching a joint course
4 between MIT and Harvard Law School having to do
5 with the interaction between policy and
6 technology issues involving the Internet.
7 Q. Has this case come up in the course of that
8 class?
9 A. No.
10 MR. HART: Sorry?
11 MR. GARBUS: I didn't say anything.
12 MR. HART: I'd like to mark your
13 declaration now. We're up to 4.
14 (Document marked as Exhibit 4
15 for identification.)
16 Q. I show you what the reporter has just marked as
17 Exhibit 4 and ask you to identify that for us,
18 please.
19 (Document exhibited to witness.)
20 A. Yes, that's my declaration.
21 Q. Thank you.
22 MR. GARBUS: You may want to keep it
23 in front of you. He'll be asking you questions
24 about it.
18
1 Q. Were there any prior drafts of this declaration;
2 that is, prior to the one that we see in front of
3 us marked as Exhibit 4?
4 A. There's the declaration I gave in the Bernstein
5 case. I don't think there are any substantial
6 differences between them.
7 Q. Between Exhibit 4 --
8 A. Between this one and the previous one.
9 Q. Between this one and the one you filed in
10 Bernstein?
11 A. Right.
12 Q. I notice that the penultimate page stops at
13 Paragraph 12, kind of in the middle of the page,
14 and there's a blank, and you turn to the next
15 page for signature. Do you see what I'm saying?
16 A. No, there's nothing....
17 Q. In other words --
18 A. There was nothing extra there. I think the only
19 difference is that when -- I'm not sure. I
20 believe when I did the Bernstein declaration I
21 was still teaching the course, and now I'm not
22 anymore, so Paragraph 3, I think, is, Paragraph
23 3, I think has changed, but I'm not positive on
24 that.
19
1 Q. I'm sorry, but you're not?
2 A. But I'm not positive, I'm not even positive of
3 that.
4 Q. Why aren't you teaching that course anymore?
5 A. I just -- people move on, teach different
6 courses, and I've done it since 1979. It seemed
7 like....
8 Q. Are you getting paid anything for your
9 involvement in this case?
10 A. No. I got a witness fee attached to this.
11 Q. That was from me.
12 A. That was from you? Thank you. I'm told the
13 going rate in Boston is $7.
14 Q. You're welcome.
15 I understand you're going to be away
16 during the trial of this case; is that right?
17 A. I'm planning to be on the West Coast next week
18 and the week after.
19 Q. And how long have those plans been in place?
20 A. The plan for next week has been in place about a
21 month, and the plan for the week after is up in
22 the air at the moment. It has to do with my
23 consulting with Hewlett Packard --
24 Q. I see.
20
1 A. -- what I have to do there.
2 Q. I see. But until it became -- when did it become
3 up in the air?
4 A. I spend the summer going back and forth between
5 Cambridge and Palo Alto, and in general I'm not
6 sure which coast I'm going to be on which week.
7 Q. I see. And were you ever informed by Mr. Garbus
8 or Mr. Hernstadt about the trial date in this
9 case?
10 A. I was, but I don't remember. I thought you said
11 next week.
12 Q. Do you recall when you were first informed by Mr.
13 Garbus or Mr. Hernstadt about the trial date in
14 this case?
15 A. One of our discussions about the depositions.
16 Probably sometime in the last three weeks, but
17 I'm not positive.
18 Q. Okay. What kind of Internet connection do you
19 have in your office?
20 A. In my office?
21 Q. Uh-huh.
22 A. I have a hundred megabit per second Ethernet that
23 attaches through the MIT artificial intelligence
24 laboratory to MIT's major, what they call the
21
1 backbone network at MIT, which itself has roughly
2 a hundred megabit connection to what's called
3 NERnet.
4 Q. NERnet?
5 A. The New England Regional Network, I believe, and
6 from there to one of the major backbone networks.
7 Q. Do you have any knowledge about the type, speed
8 or bandwidth of the Internet connections that are
9 available to students at MIT generally, whether
10 in student facilities or dorm rooms or the like?
11 A. What I have is typical of MIT. I think all of
12 our dormitory rooms have Ethernet connections. I
13 don't remember whether all of them are hundred
14 megabit. Some of them may be 10 megabit.
15 Q. I'm going to ask you a couple of questions that
16 may require you to do some rough computations.
17 A. Oh, boy. A test.
18 Q. I apologize in advance.
19 A. A test.
20 Q. No, it's not a test.
21 With that type of Internet connection,
22 what's the effective transfer rate of, say, a one
23 gig file within that network?
24 A. Okay. I'm going to embarrass myself.
22
1 Q. Please. We've all done it.
2 A. So a one gig file? Gigabyte, I assume?
3 Q. Yes.
4 A. So that's 8 gigabits. So 8 times 10 to the 10th
5 megabits.
6 You want to know how long to transfer
7 a one gigabyte file?
8 Ten megabits per second.
9 MR. GARBUS: Are you factoring the
10 time of day of this transmission? Midnight?
11 MR. HART: Please, Marty. Whether you
12 were saying that in jest or not, I would
13 appreciate you confining your comments at this
14 point.
15 MR. GARBUS: It was in jest.
16 A. That's 8,000 megabits, roughly. One megabit per
17 second, that would be 8,000 seconds; ten, that
18 would be 800 seconds; a hundred megabits would be
19 80 seconds. So if I take ten megabits per
20 second, that's -- a hundred megabits per second,
21 which is roughly a minute and a half for a one
22 gigabyte file. I hope I'm getting this right.
23 Q. Okay.
24 A. A ten gigabyte file?
23
1 Q. No, one gigabyte file.
2 A. Looks to me to be at a hundred megabits a second,
3 it's a minute and a half; and at ten megabits per
4 second, it's about 15 minutes.
5 That's theoretical capacity.
6 Q. Right.
7 A. Now, state the question again so I can understand
8 the assumptions. That's theoretical capacity.
9 Q. I understand.
10 A. If I were doing it at MIT, for example.
11 Q. Right.
12 A. So if I were doing it at MIT, MIT's backbone as a
13 whole would tolerate a hundred megabits per
14 second, so if there were ten people trying to use
15 the network, I'd get maybe a 10th of that.
16 Q. Is that always true of network connections, or
17 does that depend on how they're switched?
18 A. It depends on how they're switched, and it
19 depends on just lots and lots and lots of
20 details.
21 Q. So let's say on your average day, taking into
22 account the humidity and the other factors Marty
23 was alluding to, in the real world, what transfer
24 rate are you getting on a hundred meg network at
24
1 MIT? What's your real yield, roughly?
2 A. I would have to guess around 25.
3 You have to understand, for very short
4 periods of time, I can get close to the maximum.
5 If you're trying to do something over a long
6 period of time, you're sharing the connection.
7 Q. What do you call a long period of time in the
8 Internet world?
9 A. A minute.
10 Q. Taking our hypothetical one gig file and
11 factoring real world factors into it, what do you
12 think is a real world case for the transfer of
13 that one gig file over your hundred meg network
14 at MIT?
15 A. I'd have to guess around 15 or 20 minutes.
16 Q. Are you familiar with how large a feature length
17 movie file is as embodied on a DVD?
18 A. Not really.
19 Q. Do you have any anecdotal information?
20 A. Let me think. DVD is 600 megabytes. DVD -- no,
21 I just don't. Something like -- it's multiple
22 gigabytes, but....
23 Q. I'm not asking you to guess, I just wanted to
24 know what you knew.
25
1 Are you familiar with video
2 compression technologies, generally?
3 A. Only a little bit.
4 Q. Do you have any idea --
5 MR. GARBUS: Bill, do you want to
6 stipulate as to a general number in case you want
7 the doctor to figure it out? Just take a number
8 that we can agree is kind of an average or
9 working number so he can do the rest of the
10 computation?
11 MR. HART: No.
12 Q. We're not going to do a whole lot more math here,
13 Doctor. But I thank you for that computation a
14 minute ago.
15 Do you have any knowledge of the kinds
16 of compression ratios that are currently
17 available under, for example, the MPEG-4 standard
18 for video?
19 A. I would guess, this is just a guess, a factor of
20 ten, but it depends on quality.
21 Q. Is that also something that is developing as a
22 science or technology, that is, there are
23 improvements being wrought on a continual basis?
24 A. Well, I mean, MPEG-4 is fairly recent. I'm not
26
1 aware of things that are going on beyond that,
2 but I know there are people working on it.
3 Q. This is just part of the nature of this beast,
4 because the record gets cluttered. I'll try not
5 to interrupt you.
6 How do you know about MPEG-4?
7 A. Sort of general reading in the technology
8 literature.
9 Q. To your knowledge, is it a commonly available
10 piece of software?
11 A. I don't know. I know it's used.
12 Q. Have you ever heard of Napster?
13 A. Oh, yeah.
14 Q. How did you hear of it?
15 A. More general reading both in technology
16 literature and newspapers; read an article about
17 it this morning in the Wall Street Journal while
18 I was waiting.
19 Q. Do you know if MIT has had a problem with Napster
20 in terms of student usage?
21 A. No. MIT's noticed it, and I know that other
22 universities have. MIT in general has enough
23 capacity that Napster use wasn't a problem,
24 although it was at the level where if it were
27
1 much more, it would become a serious problem.
2 Q. And that's because the bandwidth was being used
3 for so-called file sharing of MP3 audio files?
4 A. Yes. Just because MIT has a lot of internal
5 capacity on its network.
6 Q. What kind of Internet hook-up do you have at
7 home?
8 A. Connection through Media One.
9 Q. What's the effective bandwidth through that?
10 A. I believe it's 300 kilobits upstream, and a
11 little over a megabyte downstream.
12 Q. And by "upstream" and "downstream," you mean that
13 you could get a megabit in if you were
14 downloading something from the Net?
15 A. You could get a megabit -- well, again,
16 theoretically.
17 Q. Right. I just wanted to define upstream and
18 downstream.
19 A. Yes, you can get stuff from the Internet about
20 three or times as fast as you could upload it.
21 Q. Gotcha.
22 CONFIDENTIAL
23
24 Q. Do you have any view of what the sort of standard
28
1 bandwidth is that's available at major
2 universities and colleges throughout the country?
3 A. I tend to think it's -- are you talking about
4 bandwidth in the university or inside the
5 university to the outside?
6 Q. Let's do them both.
7 A. From is usually either 10 or a hundred meg.
8 Q. Meg?
9 A. Megabits per second. Internally, it's about the
10 same, 10 or a hundred. Again, with the external
11 connections, there tends to be bottlenecks and
12 things.
13 Q. Just so that I'm clear, you may have had it
14 perfectly in your own mind, were we speaking of
15 the 10 or 100 meg that we were speaking of when
16 we were talking about MIT a few minutes ago,
17 megabytes?
18 A. Megabits per second.
19 Q. As opposed to megabytes?
20 A. Right. We're talking networks, you usually talk
21 about megabits.
22 Q. So just to be clear, again -- it may be my fault
23 in terms of how I ask the question, I apologize
24 for that -- when we were talking about what the
29
1 typical network bandwidth is at major colleges
2 and universities throughout the country
3 internally --
4 A. Right.
5 Q. -- what is that bandwidth rate, approximately in
6 your view?
7 A. No, my guess, it's either 10 or a hundred.
8 Q. 10 or a hundred what?
9 A. Oh, I'm sorry. I'm sorry. 10 or a hundred
10 megabits per second.
11 Q. Thank you.
12 A. Theoretical maximum transfer.
13 Q. Okay. And that theoretical is subject to the
14 same variables you described a few minutes ago
15 when we were talking about MIT's network, right?
16 A. Right. The way to understand that is that
17 maximum transfer rate means that's as much as you
18 have for everyone who might be trying to do this
19 at one particular time at the university.
20 Q. But doesn't that also depend on how it's
21 switched, that is, how the network is switched?
22 A. Yes, it depends in complicated ways that I'm not
23 expert enough to really characterize.
24 Q. I'm sorry, I didn't mean to interrupt you. In
30
1 general terms, do you have any understanding of
2 the difference between a shared network and a
3 switched network?
4 A. A shared network and a switched network? No, I
5 don't.
6 Q. Are there networks, to your knowledge, that can
7 make the bandwidth that you mentioned, let's say,
8 10 meg, available to all users no matter how many
9 users are on it?
10 A. Not that I'm aware of.
11 Q. Okay. Can you read object code?
12 A. Sure.
13 Q. Is there a difference, in your view, in the
14 expressive content between object code and source
15 code?
16 A. It depends very much on the circumstances. In
17 general, each one will express different things.
18 Q. But you have not looked at either the object code
19 or source code versions of DeCSS?
20 A. No.
21 MR. GARBUS: Can you take a second? I
22 just want to make one call.
23 MR. HART: Let the record reflect Mr.
24 Garbus is making a phone call. I don't think
31
1 there's a question pending right now. We can
2 take five.
3 (Discussion off the record.)
4 MR. HART: Okay. We're back.
5 (Question and answer read.)
6 Q. What expressive value does object code have?
7 A. Well, object code can tell me a tremendous amount
8 about efficiency. Source code is generally
9 written to be machine independent.
10 Q. Source code is?
11 A. Source code is. And then when you put that
12 through a translator to translate it into
13 something for a specific machine, then the object
14 code is telling you about, in general, the
15 interactions of the original source program with
16 specific details of the architecture of the
17 machine.
18 Q. The object code is?
19 A. That's what the object code will tell you. Then
20 for some applications that will be absolutely
21 critical.
22 Q. In other words, to see how efficiently the
23 program runs, object code has expressive value?
24 MR. GARBUS: I object to what you
32
1 said. He said more than that. But go ahead.
2 A. It's a little bit more than efficiency. Again,
3 it's interactions of the detailed, details of
4 this program execution with the specific
5 architecture of the machine, and that can be not
6 only efficiency but things like reliability,
7 things like how it will interact with maybe other
8 programs that are running on the machine at the
9 same time.
10 Q. What, to your knowledge, was the issue in dispute
11 in the Bernstein case?
12 A. The issue in the Bernstein case was that the
13 government had claimed that posting of various
14 kinds of encryption algorithms, or even in the
15 Bernstein case, a technical paper describing
16 them, violated the Export -- initially it was
17 ITAR, but later the Export Control Act.
18 Q. What is your understanding of the Export Control
19 Act? I'm not looking for a legal dissertation
20 here.
21 A. My understanding is that it's horribly
22 complicated.
23 Q. Is it fair to say until recently there were
24 restrictions on the bit length of certain
33
1 encryption codes that were put into place by
2 government regulation?
3 A. There were certain restrictions on bit length.
4 There were other restrictions as well, but I
5 can't remember them. It's a very convoluted and
6 complicated act, as I recall.
7 Q. Do you remember the bit length restriction that
8 was in place at the time of the Bernstein case?
9 Was it a 40-bit limitation?
10 A. At one point I believe it was 40 bits. I think
11 at some point it got changed to 56.
12 Q. Okay. Have you ever gone to the 2600 web site?
13 A. Yes, I did, when I was looking at this case.
14 Eventually I read it because 2600 had a
15 discussion of the case, so I looked at it, but
16 just only briefly.
17 Q. Was that before or after you got the call from
18 Messrs. Garbus and Hernstadt?
19 A. After.
20 Q. I'm just having to complete the record.
21 A. Yes, it was after.
22 Q. Did you ever have occasion to read the judge's
23 opinion or opinions in this case involving the
24 preliminary injunction?
34
1 A. I skimmed them, again, recently. I skimmed them.
2 Q. Now, if I can try and sum up your declaration in
3 one sentence.
4 MR. GARBUS: That might be a problem.
5 MR. HART: It might be, but let me try
6 it.
7 Q. Would it be fair to say that it's your view that
8 computer programs are expressive in nature and
9 convey ideas that are useful for people that are
10 involved in the evaluation or study of computer
11 programs?
12 MR. GARBUS: I'll object. I think he
13 didn't say it that narrowly.
14 MR. HART: Okay.
15 MR. GARBUS: Why don't you let the
16 witness --
17 Q. You can correct it. I'm trying to move it along.
18 MR. GARBUS: Why doesn't the witness
19 say what he intends to say in his affidavit.
20 MR. HART: You're going to object that
21 the declaration will speak for itself.
22 MR. GARBUS: Why don't you just tell
23 us....
24 A. I guess it's my declaration that computer
35
1 programs are mediums of expression for many
2 purposes, and they convey ideas. Often a
3 computer program will be the most appropriate way
4 to express a particular idea about how things are
5 done, and it's quite independent of any
6 particular purpose that someone wants to use it
7 for.
8 MR. GARBUS: Could you just mark that
9 answer because I'm going to want to go back to
10 it.
11 Q. When you said at the end of your answer, in rough
12 ideas, independent of the use to which the
13 program can be put --
14 A. Yes.
15 Q. -- could we fit into that general statement,
16 independent, say, for example, of the use of a
17 decryption utility?
18 A. Wait.
19 MR. GARBUS: I object to the question.
20 Too many independents.
21 A. Try again. I got lost on you.
22 Q. Okay. I'm sorry. That was my fault, I
23 apologize.
24 Even though you haven't examined or
36
1 analyzed DeCSS in either source or object code
2 form --
3 A. Right, that's correct.
4 Q. -- is it your position that as a general
5 proposition DeCSS must have some expressive value
6 in its object or source code iterations?
7 A. I'm not sure about "must have," but I would
8 certainly expect that it does.
9 MR. GARBUS: Do you have a copy of the
10 codes with you? Maybe we can show it to the
11 witness.
12 MR. HART: You might want to do that.
13 I'll continue my line of questioning, Marty,
14 though, if you don't mind.
15 Q. I believe in your previous answer, which I made a
16 garbled question over, and I apologize for that,
17 I believe you said that a program can convey
18 ideas independent of whatever use it's for?
19 A. Yes.
20 Q. I think we should go back -- is that right?
21 A. Yes, that's right.
22 Q. So I guess this is a syllogism. You deal with
23 this more than I do.
24 A. Okay.
37
1 Q. Does that mean that DeCSS, as a decryption
2 utility, would, in your view, have value to
3 communicate ideas totally independent of its use
4 as a decryption utility?
5 A. Oh, yes, certainly.
6 Q. Is that essentially your testimony?
7 A. I can give you an example. It's CSS was
8 incompetently designed. So that's an idea that
9 you can get from looking at DeCSS, I would
10 assume.
11 Q. It's also an idea you could express just in the
12 manner you did, isn't it?
13 A. (Witness nods.)
14 MR. GARBUS: Objection.
15 Q. You better verbalize --
16 A. It would say more. Yes. It would tell me more
17 details about that particular kind of
18 incompetence. And, sure, there are other ways
19 one could say that, but often computer code is
20 the most succinct way to do that.
21 Q. Now, let me pose this to you, because I want to
22 see where you come out in terms of potential for
23 abuse.
24 A. Abuse?
38
1 Q. Yeah. Let's take --
2 MR. GARBUS: He's not abusive.
3 THE WITNESS: I'm not sure who's being
4 abused here.
5 MR. HART: Well, I hope no one.
6 Q. Let's take some code that in the form of an
7 executable utility would enable anyone who
8 downloads it from the Internet to walk into
9 CitiBank and cause cash to be spit out of
10 someone's account.
11 A. Okay.
12 Q. And I guess I need your view as to where you
13 balance the interests in terms of interest and
14 information about the program versus the
15 proliferation of same on the Internet when that
16 risk exists.
17 MR. GARBUS: I would object. I would
18 allow the witness to answer, if you -- it seems
19 to me that's an endless dialogue, but go ahead.
20 A. Just ask again so I can --
21 Q. Sure. Here's the example: Someone develops a
22 utility that if downloaded from the Internet
23 enables you to basically go to any CitiBank ATM
24 machine and take money out of someone else's
39
1 account.
2 A. Right.
3 Q. And in view of what I've said about the sort of
4 expressive content of computer code and interest
5 in examining that for whatever reason --
6 A. Right.
7 Q. -- I guess I'm asking you, Hal Abelson, where you
8 come out professionally and personally on the
9 issue of promulgating that executable utility on
10 the Internet for academic instructive value if
11 there is the risk of abuse that people could
12 download it and take money out of CitiBank using
13 it. Do you understand my question?
14 A. I'd say in that particular case, I would tell
15 CitiBank about it, wait a couple of days, and
16 then make it very clear that either I or someone
17 else would promulgate that code. I think that's
18 probably the most -- I would argue that's
19 probably the most useful thing to do for
20 CitiBank. But, again, it's kind of up to them to
21 say.
22 Q. Based on the assumption that they could change
23 their system?
24 A. Based on the assumption either that they could
40
1 change their system or they certainly, they
2 certainly should know about that weakness, and
3 they certainly should know that that would be a
4 really serious weakness.
5 Q. And notifying CitiBank about that weakness is
6 different than promulgating that utility on the
7 Internet, yes?
8 A. It's different. Sometimes it's very difficult to
9 get people to change, it's very difficult for the
10 people who you inform of this thing to make
11 changes happen.
12 Q. Do you have a view as to the expressive value of
13 promulgating DeCSS on the Internet in this case?
14 A. View as to the expressive value?
15 MR. GARBUS: I object to the form of
16 the question, but I'll certainly let Professor
17 Abelson answer. He's already testified --
18 MR. HART: Please.
19 MR. GARBUS: Go ahead.
20 A. Well, I think it does quite well at expressing
21 the nature of the Content Scrambling System. I
22 think it -- I would venture to say that had the
23 people who designed the Content Scrambling System
24 had the benefit of the kinds of things I've read
41
1 on the Internet about DeCSS, they certainly would
2 have designed a different algorithm.
3 Q. You've read things about DeCSS on the Internet;
4 is that right?
5 A. Yes.
6 Q. And despite the fact that you're testifying in
7 this case as an expert, and despite your
8 expertise in computers and software, you've had
9 no occasion to actually look at the source or
10 object code in connection with your evaluations
11 or your opinion or your study?
12 A. I wasn't asked to do that.
13 Q. I see. But you didn't find it necessary as a
14 professional to do that?
15 A. Necessary for what?
16 Q. To understand DeCSS or its expressive value, all
17 the things you testified to.
18 A. Oh, in general?
19 Q. Yes.
20 A. I'm not particularly looking at those kind of
21 encryption systems.
22 Q. But we're discussing that kind of encryption
23 system in this case, are we not?
24 A. Well, you are. I'm testifying about the general
42
1 expressiveness of code.
2 Q. Okay. And I'm simply confirming with you, sir,
3 that in the course of your testifying about the
4 expressive nature of code, and in the course of
5 your reviewing things on the Internet about
6 DeCSS, you, as a professional, as an expert in
7 this area, saw no need to actually look at the
8 code, did you?
9 A. It's one of the very many algorithms I haven't
10 looked at.
11 Q. Is there anything else that you intend to testify
12 to in this case in connection with DeCSS, CSS, or
13 any of the issues as you understand them in the
14 case?
15 A. That I intend to testify about? I didn't come
16 here with intentions to testify. I'm just
17 responding to both of your questions.
18 Q. I see. Were there areas that were identified by
19 Mr. Garbus or Mr. Hernstadt, or anyone else, for
20 that matter, as to topics for you to testify on
21 in this case?
22 A. No.
23 Q. Do you have a web site?
24 A. Do I have a web site? My research group at MIT
43
1 has a web site.
2 Q. What's that group's web site called? If I were
3 to go look for it, how would I find it?
4 A. The project on mathematics and amorphous
5 computer -- I'm sorry. The MIT project on
6 mathematics and computing.
7 MR. HART: You know what might make
8 sense to expedite this, if you give me about ten
9 minutes, I'm see what I have left, it may enable
10 me to consolidate.
11 MR. GARBUS: I'm going to ask some
12 questions, so what you may want to do, it might
13 make things go faster, after I ask questions, I
14 presume you're going to want to ask some also.
15 If you want, I can go now, or we can do as you
16 wish; namely, give you a few minutes for you to
17 do it.
18 MR. HART: Why don't we give me a few
19 minutes, then we'll --
20 MR. GARBUS: Do you have with you
21 copies of the DeCSS source and object code? If
22 so --
23 MR. HART: I don't think so.
24 MR. GARBUS: We'll take ten minutes.
44
1 (Brief recess.)
2 BY MR. HART:
3 Q. Have you prepared any kind of report or summary
4 of your views or possible testimony in connection
5 with this case?
6 A. No.
7 MR. HART: I don't think I have
8 anything more for you, Dr. Abelson. I thank you
9 for your time and for your candor
10 CROSS EXAMINATION
11 BY MR. GARBUS:
12 Q. Dr. Abelson, on your site do you link to the site
13 of David Touretzky?
14 A. Yes, I put a link to that site after I noticed it
15 when I was looking at the various things about
16 the DeCSS.
17 Q. Do you know whether the Touretzky site has the
18 object or source code for DeCSS?
19 MR. HART: Object, form.
20 A. I don't. I don't know. I know it had a whole
21 bunch of different forms of DeCSS.
22 Q. Now, before you used the express "architecture of
23 the machine." Can you tell me what you mean by
24 that? You were talking about how -- do you
45
1 recall your reference to that?
2 A. Yes.
3 Well, in general it's the
4 architecture, it's the structure by which the
5 machine is put together. So it's such things as
6 the number of registers that there will be, the
7 number of bits in the data paths that connect
8 them, the number and kinds of operations that can
9 happen simultaneously, the size of various
10 registers and portions of memory.
11 Q. And that, you indicated before, was communicated
12 by object code; is that right?
13 A. Yes. Well, you'll see that immediately
14 reflected. I should say that differently. A
15 piece of source code could potentially translate
16 into many different kinds of object code. And
17 then depending on the sophistication of the
18 computer, you'll try to take advantage of
19 different elements of the machine, and you'll see
20 that reflected in the object code.
21 Q. When you say source code can be translated to
22 different object codes, can you describe that for
23 me?
24 A. Sure. There's a translation process. In general
46
1 there are many levels in which languages will be
2 translated. So in the very simplest case, you'll
3 have a program that takes in source code as
4 prepared by a human programmer and produce a code
5 called ASSEMBLY code, which is rather close to
6 the machine, although it's symbolic. And then
7 ASSEMBLY code will be further translated by
8 something called an assembler to something that's
9 really at the level of the machine.
10 In more general cases, you'll have a
11 whole bunch of levels of intermediate code; and
12 even more generally, a program can be prepared in
13 a multiplicity of languages. So it would be --
14 it's often possible to write much of an algorithm
15 in a very high level language, then write
16 critical details, in fact dropping down to the
17 ASSEMBLY level.
18 Q. When you say a multiplicity of languages, can you
19 explain that for me?
20 A. Yeah, a good example would be something like
21 JAVA, so JAVA people typically write code in the
22 JAVA language. That is compiled or, quote,
23 compiled into something called JAVA bite code.
24 JAVA bite code is a lower level language than
47
1 JAVA source code. However, it's designed to be
2 machine independent. When people ship around
3 little JAVA applets, it's the JAVA bite code
4 that's getting shipped around the network.
5 Then when you get that to the
6 recipient machine, the bite code could either be
7 translated on the fly, which is a process called
8 interpretation, or for other purposes it itself
9 can be compiled by a compiler that's specific for
10 the target machine. So that's an example where
11 you have three levels. And in that case
12 something like JAVA bite code would be called an
13 intermediate language.
14 Q. So you would have source code, intermediate
15 language and then object code?
16 A. Then object code.
17 Q. Now --
18 A. Although I should say it depends on who's
19 looking. So you'll often find JAVA bite code
20 described as object code. It really is kind of
21 in the eye of the beholder.
22 Q. When you say it's "in the eye of the beholder,"
23 then, you also talked about object code showing
24 you operations that are critical to how the code
48
1 operates, how the machine operates. Would you
2 call that expression? Would you explain that?
3 A. Sure, I'll give you a --
4 MR. HART: I'm going to object to
5 that, but go ahead.
6 Q. Let me rephrase the question so there's no
7 problem about it --
8 A. Okay.
9 Q. -- in case we --
10 You used the term "operations
11 critical" in response to Mr. Hart's question.
12 Can you expand on that?
13 A. I'll give you an example. So often an algorithm
14 will have a particular interlude which you want
15 to make run absolutely as fast as possible. One
16 of the typical ways to make things run fast is to
17 take advantage of a property of the machines
18 called pipeline, which effectively means they can
19 set up to do the next operation while they're
20 doing the first one. And often you can have as
21 many as, oh, ten different machine instructions
22 that are all in very different phases of
23 execution, which means your algorithm will run,
24 say, ten times as fast.
49
1 Now, whether or not you're actually
2 able to do that depends on very subtle details of
3 how the algorithm is designed and subtle details
4 of the compilation process.
5 So a particular -- just to give you a
6 particular example, if I have to fetch three
7 numbers from memory and do two multiplications,
8 the particular order in which you fetch things
9 from memory and do the multiplications will have
10 an impact on how well those things can be
11 pipelined.
12 Q. And --
13 A. So I was going to say when somebody is designing
14 the interlude of an algorithm, it's very, very
15 common that you will in fact write a piece of
16 this algorithm in source code, run it through the
17 compiler to look at what the computer -- look at
18 the object code that the compiler produces, and
19 then modify that object code and feed that into
20 your, feed that into your program. That's an
21 example of what I was talking about before when
22 you have an algorithm that is actually input to
23 the computer by a human as a mixture of source
24 and object code.
50
1 Q. Now, before Mr. Hart asked you about that
2 CitiBank analogy. Can you give me the basis for
3 your statement or your view that you would give
4 CitiBank a certain amount of time; and if they
5 did not respond, then release the information
6 publicly?
7 A. I think it's -- I mean, I don't quite know what
8 it looks like to be sitting at CitiBank and
9 someone calls you up and says there's a major
10 problem in your banking system. I don't know how
11 much you pay attention to that or how much you
12 would even consider it. I don't know that if you
13 were an employee at CitiBank who got such a call,
14 perhaps in the programming department, and you
15 have to convince some vice-president to make a
16 major investment in changing their banking
17 system, I don't know how much luck you would have
18 doing that unless there is some real visible
19 impact to them taking no action.
20 Q. What is the ethical or social value of you
21 releasing that information five days, ten days,
22 or whatever, if a CitiBank doesn't respond to
23 you?
24 MR. HART: Object, form, prior
51
1 testimony.
2 Q. Go ahead.
3 A. Say again.
4 Q. Why would you do it?
5 A. Why would I give them five days notice, or why
6 would I give them only five days notice?
7 Q. Why would you give them notice, and why would
8 you, if they did not respond, then release the
9 information?
10 A. Well, I would give them notice because I think
11 it's the polite thing to do. I mean, if my goal
12 is to actually make things, actually improve
13 things as opposed to my goal being to rob
14 CitiBank, then for one thing it demonstrates
15 intent. This is my goal, I've done it publicly,
16 it demonstrates my good will. Hopefully someone
17 now at CitiBank has seen this; if they know this
18 is going to happen, they would get some extra
19 warning.
20 Q. And if they do nothing, there's a possibility, is
21 there a possibility that people will then go in
22 and rob CitiBank on the basis of the information
23 you've given them?
24 A. Yes, but offhand it's -- in general if you find a
52
1 flaw in a security system, it's wildly unlikely
2 that you're the only person who has found that
3 flaw. And I think the real danger is that there
4 are people who are trying in fact to rob CitiBank
5 and are looking at this stuff very, very
6 carefully. I think by the time this stuff gets
7 out to the public, they're unlikely to be the
8 first to know.
9 Q. So is it your testimony that you're being
10 protective, in a way, of CitiBank by releasing
11 the information?
12 A. I think I'm doing them a service. I'm not
13 exactly sure -- I'm not sure they would
14 characterize it as protective of them.
15 Q. Why do you think you're doing them a service?
16 Why do you think you're doing them a service?
17 A. Because I'm strongly encouraging them to improve
18 their system, with the likelihood that they are
19 guarding against security attacks that they do
20 not know about.
21 Q. Now, you say the DeCSS shows you how weak CSS is?
22 MR. HART: Object, form and prior
23 testimony.
24 Q. Can you expand on that?
53
1 MR. HART: Same objections.
2 A. Well, the descriptions -- be careful. The
3 descriptions I have read of DeCSS, based on other
4 people who have seen it, demonstrate that it is
5 extremely weak.
6 Q. Do you know how and do you know where?
7 A. Two primary ways: Apparently it's 40-bit
8 encryption. So given merely that you know it's
9 40-bit encryption, I would expect you could
10 break, you could break something in a couple of
11 minutes.
12 And the second way it's weak is that a
13 DeCSS recorder holds a decryption key -- I'm
14 sorry, DeCSS player holds a decryption key that
15 will decrypt any CD, so it's roughly -- it's
16 roughly as if, if I understand it correctly,
17 every recorder holds -- if I have a box that I
18 give everybody that has a key to your house, and
19 this key will open the lock on everybody's house,
20 so all somebody has to do is compromise one
21 recorder, and then you can decrypt every DVD.
22 Q. Do you know that in fact those recorders have
23 been compromised?
24 A. No, I don't know.
54
1 Q. Do you know when the first cracks of CSS were
2 published?
3 A. No, I don't know.
4 Q. You mentioned before, you used the term
5 "theoretical capacity" in response to Mr. Hart's
6 question. What are the variables that affect
7 theoretical capacity?
8 A. That affect the theoretical capacity?
9 Q. Yeah. Concerning the computations you gave
10 before.
11 A. I want to make sure I understand the question.
12 Q. Go ahead.
13 A. You're asking if I were looking at a network and
14 saying what is it's theoretical capacity?
15 Q. Yes. And what are the characteristics that you
16 consider when you get to its real capacity? In
17 other words, how is the theoretical capacity --
18 A. Okay. Those are two different questions.
19 Q. Pardon me.
20 A. So theoretical capacity has to do with sort of
21 the basic bandwidth of the individual pieces of
22 the network, as I might measure in the
23 laboratory. It might have to do with, if I took
24 one of the switches and drove it under ideal
55
1 circumstances, how fast it could switch.
2 Let me think what else.
3 Generally it doesn't consider
4 bottlenecks coming from the individual servers at
5 the end of the system.
6 Q. Can you describe to me what you mean by
7 bottlenecks from the servers at the end of the
8 system?
9 A. Quite simply, if my machine is trying to serve up
10 a file and trying to do 50 other things at the
11 same time, it may not be able to feed the network
12 as fast as it possibly could. Similarly with the
13 machine on the other side that's receiving.
14 Similarly, there are addressing
15 bottlenecks in the Internet. If I have to go
16 through -- if I'm connected through something to
17 something by multiple hops, and depending on how
18 those switches are configured, there can be
19 slowdowns there.
20 Q. Is distance a factor?
21 A. Well, distance in --
22 MR. HART: Object, form.
23 A. -- itself is not a factor. Distance in the sense
24 of the number of hops that you have to go through
56
1 is a major factor.
2 Q. Can you tell me why and how?
3 A. Because every time -- what happens is sort of a
4 piece of information goes over one segment of
5 network, and then gets effectively to a switch or
6 sometimes to a machine that acts as an
7 intermediate way station for it. And then the
8 rest of the path has to be recomputed. And that
9 takes a little bit of time. By the time you get
10 a couple of hops, those things add up.
11 MR. GARBUS: Now, Bill, I presume we
12 don't have to go through the affidavit line by
13 line? I presume we can stipulate if he were
14 asked questions with regard to each allegation in
15 the affidavit, he would confirm it as his
16 testimony. If you want, I'll show it to him and
17 say, "Is that your testimony?"
18 MR. HART: I think I understand what
19 you're asking me, and without waiving any
20 objection to the fact that you're trying to
21 perpetuate the testimony, if your question is,
22 would Dr. Abelson testify to the effect that he
23 set forth in his affidavit, I think you can just
24 ask that question and have him answer it.
57
1 BY MR. GARBUS:
2 Q. Would you, if I were to ask you questions about
3 your affidavit, testify that the allegations in
4 your affidavit are accurate and true today?
5 A. Yes.
6 Q. In the Bernstein case, do you know if it was
7 object, source code, or a combination?
8 A. I believe that Bernstein himself published source
9 code, and also I believe technical memos that
10 weren't encoded at all. And I believe the
11 government's injunction covered both of those.
12 There may have been arguments before
13 the Court about source and object code, but I
14 didn't look at those. I don't believe there was
15 a distinction --
16 Q. You don't --
17 MR. HART: I'm sorry, let the witness
18 complete his answer.
19 A. I don't believe there were distinctions made, but
20 I don't remember.
21 Q. In Younger, do you know if it was object code or
22 source code?
23 A. No, I don't. Under the impression --
24 Q. Put aside what the Court said. Do you know what
58
1 it was, in fact?
2 MR. HART: Object.
3 A. I don't quite remember exactly what the Court
4 said, but I don't believe people were talking
5 about those distinctions very much.
6 Q. Why?
7 A. Because they were just talking about, again --
8 MR. HART: Continuing objection to
9 this. Go ahead.
10 A. Again, I don't remember exactly what Younger was
11 doing, but he was talking about teaching stuff in
12 his class, and I believe it was just discussing
13 things at the algorithm level, but I'm not sure.
14 Q. Insofar as communicative expression, then, to
15 your mind, or to your, based on your education
16 and writing and teaching, are source code and
17 object code both covered as communicative
18 expression?
19 MR. HART: Object, form.
20 A. Well, both -- yes. I mean, both -- if I take a
21 particular program, and I look at it in its
22 source form or in its object form, those express,
23 those communicate, those communicate different
24 things, and depending on what I want to do, I'll
59
1 want to look at them in different forms.
2 Q. Does the fact that you haven't actually seen the
3 DeCSS source code or object code change your
4 opinion that DeCSS object and source code are
5 both covered as communicative expression?
6 MR. HART: Same objections and prior
7 testimony.
8 A. Change it from? I haven't looked at it so....
9 Q. Instead of change it from, is it your view that
10 even though you have not looked at it thus far,
11 that DeCSS source and object code both come
12 within communicative expression?
13 MR. HART: Same objections.
14 A. I would expect they do.
15 Q. Why?
16 A. As I said, I assume that someone inferred that
17 there's a 40-bit key from looking at the source
18 code, where the object is, as an example of
19 something expressed. Someone inferred that
20 there's a global master key that is sitting on a
21 DeCSS recorder, from looking at either the source
22 or object code. So that certainly would count as
23 something that was expressed.
24 I would imagine that if I were -- if
60
1 someone wanted me to design a, say, a computer
2 program -- in fact, I don't know. I don't know
3 how the various player programs work, but I
4 assume that if someone asked me to write
5 something that simultaneously decrypted something
6 while it played it, or if someone said to me,
7 "Try to minimize the amount of intermediate
8 storage that a software program that would play a
9 DVD would work," that's a place where I would
10 want to look at the object code, specifically to
11 talk about these pipelining issues that I
12 mentioned before.
13 Q. Now, Mr. Hart asked you about your schedule. The
14 trial starts July 17. Tell me your schedule for
15 the 17th.
16 A. What day of the week is the 17th?
17 Q. A Monday.
18 A. I am planning to be on the West Coast right now
19 on the 17th. What I can say pretty, with pretty
20 much certainty, is that I will be on the East
21 Coast either the week of the 17th or the week of
22 the 26th, and I haven't decided which one yet.
23 Q. And is that consistent with your Hewlett Packard
24 obligations?
61
1 A. Yes.
2 Q. That's what draws you out there?
3 A. That's what draws me out there.
4 Q. Now, Mr. Hart asked you some information about
5 transmission times. Let me show you an affidavit
6 of Michael Shamos.
7 MR. GARBUS: Can we mark this as an
8 exhibit? I only have one copy.
9 By the way, you do not have DeCSS or
10 CSS with you, or the object or source code.
11 MR. HART: I think you asked me that
12 before, and I think I told you no.
13 MR. GARBUS: Okay.
14 (Document marked as Exhibit 5
15 for identification.)
16 MR. HART: May I see what you're
17 showing the witness, please?
18 MR. GARBUS: Let the record indicate
19 I'm showing the witness an affidavit from a
20 Martin Shamos, it's the identical affidavit. The
21 only thing that's different is on the front page
22 in my handwriting is Mr. Abelson's phone number.
23 MR. HART: There's also a handwritten
24 notation on the 6th page of this.
62
1 MR. GARBUS: Yes, there is.
2 MR. HART: You know, without
3 belaboring, this I sort of object to your handing
4 the witness something that somebody else has
5 marked on. That's not a true copy of the
6 exhibit.
7 MR. GARBUS: Okay.
8 MR. HART: If I said "I sort of
9 object," I strenuously object.
10 (Document exhibited to witness.)
11 Q. I show you --
12 A. Is there something I'm not supposed to see here?
13 Q. No, no, no.
14 I show you the affidavit of Michael
15 Shamos, and I ask you, with respect, if you can,
16 with respect to his computations and his
17 analysis, if you can give me your views of that.
18 A. So I've seen this before. You faxed it to me a
19 couple of days ago. That's my fax number that's
20 written on there.
21 I've been thinking about it some. I
22 think the -- I'm certain the facts in there look
23 true. Shamos has a very good reputation. I
24 think he's very well respected. We were both at
63
1 Princeton at the same time. He had a very good
2 reputation at Princeton.
3 There's nothing in here to disbelieve
4 his, except that it's based on -- the
5 implications are based on theoretical capacity.
6 So I notice he talks about transmitting a DVD
7 movie in less than 20 minutes, and I think -- so
8 it looks like I got the test you gave me right, I
9 said about 15. That sounds about right for
10 transferring it.
11 The one thing that he doesn't take
12 account of, though, is actual network capacity as
13 opposed to bandwidth. So if you go back to --
14 you asked me about Napster before, and university
15 banning it.
16 MIT, as I said before, didn't ban it
17 or didn't have a problem with it, but many
18 universities did, and they had a problem because
19 of capacity. The notion here that a lot of
20 people would be transmitting DVDs is going to, is
21 probably going to be impossible because of
22 constraints on network capacity.
23 So, for example, let's suppose I take
24 a DivX compressed DVD. If I use Shamos's numbers
64
1 here -- what does he say? He doesn't say
2 exactly. But assume it's, when I compress it,
3 it's 600 megabytes, so it's roughly a CD. And
4 then he talks about --
5 Q. Excuse me. Can we just --
6 MR. HART: I'd like the witness to
7 complete his testimony rather than you interrupt
8 him.
9 Q. I know that. Can we use as an assumption that a
10 DVD is somewhere between 5 and 8 gigabytes?
11 MR. HART: Objection. You're
12 testifying.
13 THE WITNESS: Wait, wait.
14 MR. HART: The witness was saying
15 something different, and you have interrupted
16 him.
17 Would you mark this, please, Ms.
18 Reporter, because I think that was inappropriate,
19 Marty.
20 A. Let me speak to, actually --
21 Q. Okay. Do it without my reference.
22 A. What Shamos did where -- what he's talking about
23 is pushing around the compressed form, which is
24 what he got from this other person over the
65
1 Internet, in 20 minutes. Let's leave aside what
2 an uncompressed, what he says here,
3 multigigabyte. Let's only talk about the
4 compressed one, the thing he's actually writing
5 to.
6 So he transmitted -- what did he do?
7 He transmitted that from this other person in, he
8 said, 11 hours, which sounds reasonable, given
9 that that person was probably at home, or
10 something. Then he said that if that were going
11 on between students in the university, you could
12 do the same thing in 20 minutes. That's true
13 based on bandwidth.
14 However, if you assume that that's a
15 600 megabyte file, then transmitting a 600
16 megabyte file in 20 minutes is a data rate of
17 something like 4 megabits per second. So before
18 we were talking about 10 or a hundred. So if you
19 assume -- assume you're only using 4 megabits per
20 second.
21 Now, at a place like MIT, you have a
22 hundred megabits per second available for
23 everybody. So what that says is that if 25
24 people were simultaneously doing this, you would
66
1 completely saturate the MIT network. And I can
2 assure you if five people were simultaneously
3 doing this, this would have a pretty big impact
4 on the MIT network.
5 Q. Why would it saturate the network?
6 A. Because -- when you say that the network is a
7 hundred megabits per second, so at MIT that is a
8 shared hundred mega bits for everybody, and that
9 means that if people are using -- if there are 25
10 people, each of whom are transferring at 4
11 megabits per second, then you've used up the
12 entire network capacity for anything at all,
13 electronic mail, messaging, anything.
14 So I think what you find is the same
15 phenomenon that encouraged some universities to
16 close down Napster; you'd have that phenomenon
17 multiplied by ten or a hundred using these
18 numbers. And certainly well before you got to a
19 lot of DivX trading, you would have universities
20 shut this down, quite apart from any issue about
21 liability or legality or anything.
22 The same phenomenon -- he goes on and
23 talks about Internet 2. So my understanding of
24 Internet 2 right now is that there's about 2.4
67
1 gigabits per second. You've got to remember, 2.4
2 gigabits per second shared for the whole country.
3 So that means if you had 6,000 people who were
4 simultaneously pushing one of these DivX files
5 around, you would shut down all of Internet 2 for
6 all university connections for the whole country.
7 So the analogy is, imagine that there
8 was a special kind of phone call that somebody
9 could make, and this phone call took 20 minutes.
10 And if at any given time you had 6,000 people
11 somewhere making, somewhere making this 20-minute
12 phone call, you shut down the entire U.S. phone
13 system. It just wouldn't work.
14 I don't know what the actual numbers
15 can be, but this, but the notion that simply
16 because the theoretical bandwidth is there that
17 you would actually have that kind of trading
18 going on on the Internet is a little hard to
19 believe, certainly in this generation of Internet
20 technology that's available to universities,
21 almost certainly in the next one, which is
22 Internet 2, and you can't quite say after that.
23 Q. When you say this generation, the next
24 generation, how many years are we talking about?
68
1 A. Internet 2 is just starting. And what are people
2 talking about? MIT is just hooking up to it. I
3 would expect -- I don't quite know the roll-out
4 plans. I would expect it's something like two
5 years. I haven't even heard people talking very
6 much about generations after that.
7 Q. Now, when you talk about MIT's capacity, is that
8 larger than most universities, smaller than most?
9 Can you give me a context?
10 A. It's larger than most.
11 Q. Can you tell me something about that?
12 A. It has to do with the whole infrastructure. So,
13 again, you can lay cables -- it has to do with
14 how many routers and how many switches you have
15 and how well you keep them up to date.
16 I don't have any facts on
17 universities, but, again, as I said, Napster
18 wasn't a particular problem for MIT, and I know
19 that it was for lots of other universities. But
20 whether that's a factor of 2 bit or a factor of 4
21 bit or a factor of 10 bit, I don't know.
22 Q. You've talked about transferring music files,
23 Napster, and transferring movie files, let's say
24 through DVD. What is the difference in size of
69
1 the material being transferred and the amount of
2 time that -- the time difference?
3 MR. HART: Object, form.
4 A. Well, the difference is the size of the files you
5 have to transfer.
6 Q. Can you give me some sense of that difference?
7 A. Well, let's use Shamos' numbers. There are three
8 stages. There's uncompressed video, like you
9 would get out of running DeCSS. So we use
10 Shamos's numbers, he says those are
11 multigigabytes, so, I don't know, say 6
12 gigabytes. Then there's compressed down to DivX,
13 which I gather from his numbers is something like
14 600 megabytes. But, again, I don't know what
15 quality that is. I believe they compress to 600
16 megabytes because people want to write them on
17 CDs, but I don't know what quality compromise you
18 make from going to multigigabytes to 600
19 megabytes. Let's use the 600 megabytes.
20 600 megabytes is about what, is about
21 what fits on an audio CD. So let's assume that's
22 an hour of uncompressed audio music.
23 Now, if we compare that with Napster,
24 Napster, first of all, is transmitting MP3, which
70
1 is about a factor of 10 over, factor of 10
2 compression. So you start with your one CD and
3 you get a factor of 10 for compression. And how
4 long is a song, five minutes? So that's a 20th
5 of an hour. You tell me. I'm thinking about
6 people trading Napster songs. Let's say it's a
7 10-minute song. So that says it's transmitting
8 one DivX, one DivX compressed feature length
9 movie is a hundred times transmitting a Napster
10 song.
11 Q. If it's not compressed?
12 MR. HART: Objection, form.
13 A. If the movie is not compressed.
14 Q. Then what is the --
15 A. God. So we're doing 600 megabytes to -- let's
16 use 6 gigabytes. So it's another factor of 10.
17 Q. So what is that?
18 A. A thousand Napster songs.
19 Q. Now, you said -- you talked about Internet 2,
20 let's say, having a life of two years until
21 something --
22 A. I believe they're rolling it in over two years.
23 But, again, I'm not sure.
24 Q. And tell me how you arrive at that? How do you
71
1 know that?
2 A. This is just impressions from talking to people.
3 Again, MIT has been planning for this for a year,
4 and we don't quite have it yet. And we tend to
5 be closer to the backbone than a lot of other
6 universities.
7 Q. Do you know of any other universities that have
8 gone into Internet 2 yet?
9 A. Probably -- Shamos mentioned CMU. They're
10 certainly very early in putting the technology
11 on. I would expect Stanford, I would expect
12 Cornell, UCLA.
13 Q. Now, with respect to the consumer -- go ahead.
14 A. No, I'm just trying to....
15 Q. When you talk about Internet 2, you're talking
16 about something that the universities are
17 starting to deal with. With respect to the
18 consumer, how long is it before you will have
19 this next generation available?
20 MR. HART: Objection to form.
21 A. I don't --
22 MR. HART: And prior testimony.
23 A. I don't know. I don't know of plans that are
24 much beyond cable modems going to houses.
72
1 There's talk about people trying to, but I know
2 of no plans that are particularly faster than
3 what you get over cable modem.
4 Q. I draw your attention in the Shamos affidavit
5 where he says he uses DeCCS software. What is
6 the significance or lack of significance of him
7 using DeCCS to do his experiment in transmission?
8 MR. HART: Objection, form, prior
9 testimony.
10 A. Well, that's one of the things he doesn't, that
11 doesn't quite come out here. One of the things
12 that you don't know is he went through this whole
13 process, which started with a DVD and ended up
14 with a compressed DivX file. One step in that
15 process is running DeCSS. And then the question
16 is in terms of the overall process that you have
17 to go through, how much of that is DeCSS; and if
18 you didn't have DeCSS, could you effectively get
19 the same thing.
20 I would imagine he could have just
21 taken his -- what did he do -- a video player, he
22 could have just taken the output of whatever he
23 played it on and intercepted the signal and
24 gotten a computer file, and he would have had
73
1 something that is functionally equivalent to
2 having run DeCSS on that thing.
3 And then by the time you take that and
4 you compress it through DivX, I would be actually
5 pretty surprised if you could see the difference
6 between someone starting with a DVD, running
7 DeCSS to get the output, and compressing it,
8 versus starting with the DVD, just grabbing the
9 output of the video player and compressing it and
10 producing the same file. And one of the things
11 -- I wouldn't even know which process would be
12 more difficult.
13 Q. When you say "more difficult," you mean also
14 faster and more efficient?
15 A. Faster and more difficult.
16 Q. You don't know which one would produce a better
17 image?
18 A. I would -- again, I would venture to guess that
19 you couldn't tell the difference. But someone
20 ought to --
21 MR. HART: Wait. Again, I'd like the
22 witness to finish.
23 A. I would venture to guess you couldn't tell the
24 difference, but someone ought to try that and
74
1 see.
2 Q. Do you have any sense of what you lose when you
3 do a compression of the kind that has been
4 described in the loss of image? Have you ever
5 seen anything with that degree of compression?
6 MR. HART: Object, form.
7 A. No, I don't know. There's also, there's the
8 sound compression that you have to worry about,
9 too.
10 One of the things I did notice is that
11 there's this step in here where he talked about
12 having to fix the desynchronization. For all I
13 know, if you grab the video from the output
14 recorder, you wouldn't even have to do that step.
15 But, again, I've never done either of those.
16 Q. Tell me why you think you might not have to do
17 that step?
18 A. Because you're getting the output synchronized
19 from the recorder. If it comes out as a MP3
20 file, then you just have it synchronized.
21 Q. In a way, DeCSS is less user friendly than this
22 other process?
23 MR. HART: Objection, form, prior
24 testimony.
75
1 A. Well, I'm not sure that either of them is less --
2 I'm not sure I would characterize either of them
3 as user friendly. It's not clear that if I'm in
4 the business of wanting to copy videos to send
5 them over the network to a friend somewhere, it's
6 not clear to me having DeCSS makes a big
7 difference in whether or not I can do that.
8 MR. GARBUS: I have no further
9 questions.
10 MR. HART: I'm going to have a few,
11 but I would like to take five minutes before I
12 do.
13 (Brief recess.)
14 MR. GARBUS: I'd like to ask the
15 witness a few more questions that came to me
16 during the break.
17 MR. HART: While you were talking to
18 the witness off the record?
19 MR. GARBUS: Oh. Sure.
20 BY MR. GARBUS:
21 Q. You had indicated before that your site linked to
22 the Touretzky site; is that correct?
23 A. Yes.
24 Q. And you indicated that the Touretzky site
76
1 contains various material. To your knowledge,
2 does it contain source codes and object codes for
3 DeCSS?
4 MR. HART: Object, asked and answered,
5 prior testimony.
6 A. I believe it had links to a whole bunch of
7 different descriptions of DeCSS. I think some of
8 them -- I think one was source code. I think one
9 was a picture. I think one was a T-shirt. I
10 didn't look in detail. I don't know whether it
11 had object code or not.
12 Q. Is it your perception that you were doing
13 something illegal in linking to a site that might
14 have object or source code --
15 MR. HART: Object.
16 Q. -- of DeCSS?
17 MR. HART: Same objections, prior
18 testimony.
19 A. No, of course -- it's certainly not illegal.
20 Q. Do you know what the World Wide Web Consortium
21 was or is?
22 A. Yes. I was part of setting up the World Wide Web
23 Consortium at MIT.
24 Q. And what is it?
77
1 A. It's an organization whose mission is, quote, "to
2 bring the Web to its full potential," and it does
3 this by advocating different kinds of standards
4 and interoperability standards.
5 Q. Is the consortium just a private organization or
6 is the government involved in it?
7 A. It's a group of mostly members who are private
8 companies. I don't remember whether there are
9 any government organizations who are members, but
10 there may be. And it's sponsored by a couple of
11 universities, MIT being one of them.
12 Q. Who are some of the private companies that are
13 involved, if you know?
14 A. Microsoft, Hewlett Packard. I believe the Direct
15 Marketing Association is a member. I forget. I
16 think there are about 400 members.
17 Q. And have they expressed any views on linking?
18 MR. HART: Object, the obvious
19 reasons.
20 A. The World Wide Web Consortium takes the position
21 that linking is a reference, and it's fundamental
22 to the architecture of the Web that people link
23 and that you have these references, and that's
24 how the Web is glued together.
78
1 Q. Does the name Timothy Berners-Lee mean anything
2 to you?
3 A. Tim Berners-Lee, he's the chairman of the Web
4 Consortium and the inventor of the World Wide
5 Web.
6 Q. And the author of a book called Weaving the Web?
7 A. Yes.
8 Q. Do you know on the Web you can link to
9 information about making bombs?
10 A. I believe you can. You can discuss that it
11 exists.
12 Q. And you can link to sites that give you the
13 formula for seran gas?
14 A. I believe you can.
15 Q. Do you know what seran gas is?
16 A. I guess it's a poison gas, very dangerous poison
17 gas.
18 Q. And is it your view, as someone who is a member
19 of the world wide consortium, that linking should
20 be permitted to things like DeCSS or bomb making
21 or the creation of seran gas?
22 MR. HART: I object.
23 Q. And can you break it down with respect to any one
24 of those three?
79
1 MR. HART: I object, and also on
2 competence grounds. Are you testifying on behalf
3 of the world wide consortium?
4 MR. GARBUS: No, he's not.
5 MR. HART: I maintain my objection.
6 MR. GARBUS: Go ahead.
7 A. In my view, telling people that information is
8 available in general should not be prohibited. I
9 think that by -- you're not implying advocacy or
10 responsibility or anything, you're just telling
11 someone that something is available somewhere.
12 Q. Now, when was the World Wide Web formed?
13 A. The Web itself was invented, I believe, in 1990,
14 I think. And it came to MIT in 1995, 1994. I
15 believe that's when the consortium was set up.
16 Q. And the consortium exists today?
17 A. Yes.
18 Q. And how many academics are part of that
19 consortium or members of the consortium?
20 A. Individual academics?
21 Q. Yes.
22 A. Formally, organizations join. Most of the
23 organizations are companies. But many people who
24 are informally associated with it and their
80
1 conferences, generally, are in large part
2 academics who are writing things about Web
3 standards and Web technology.
4 Q. And have you done any writing in that area?
5 A. Not in any of the Web Consortium journals.
6 Q. Elsewhere?
7 A. Mostly been informally involved with working with
8 people in the Consortium design standards.
9 Q. Do you, in any of your classes, or, to your
10 knowledge, do any of the MIT professors teach
11 object code or how to read object code in their
12 classes?
13 A. Oh, I -- teach how to read -- there certainly are
14 many classes in which you read object code.
15 Q. Is that something that's taught also, how to read
16 it?
17 A. Yeah, how to read it, how to write it. I would
18 have said that up until ten years ago, it was
19 completely standard to program, to have courses
20 in programming what's called ASSEMBLY code.
21 Q. Is it your experience that students now come into
22 MIT, many of them already knowing how to write
23 object code?
24 A. Many? Yeah, I think "many" is correct. Again,
81
1 it depends on the particular machine.
2 Q. How long have you been teaching at MIT?
3 A. Since 1973.
4 Q. And what is your status there?
5 A. Full professor.
6 Q. How long have you been a full professor?
7 A. Good question. I don't remember. 1983, perhaps.
8 1984.
9 Q. Where were you educated at?
10 A. I have a graduate degree from MIT and
11 undergraduate degree from Princeton.
12 Q. And the curriculum vitae that has been marked as
13 Exhibit 1, is there anything you would add to it
14 or is there anything more recent or anything more
15 relevant to today's discussion that you can think
16 of? If not....
17 A. You might want to add the -- I mentioned the
18 joint course I teach with Harvard Law School, but
19 other than that....
20 MR. GARBUS: Go ahead, Mr. Hart. I'm
21 through.
22 MR. HART: You're really done? I
23 maintain my objection to all of that.
24 MR. GARBUS: Unless I have another
82
1 thought.
2 MR. HART: I do have a few questions
3 for you.
4 REDIRECT EXAMINATION
5 BY MR. HART:
6 Q. You mentioned earlier that one of your areas of
7 consulting for Hewlett Packard was -- correct me
8 if I'm wrong -- to develop systems that were
9 password protected; is that right?
10 A. I'm not sure that's the right question.
11 Q. What is the right question?
12 A. You mean different kinds of authorization
13 mechanisms for having people get at things. Some
14 of them are passwords, some of them -- there's
15 work being done in smart keys, there's work being
16 done in biometrics.
17 Q. I didn't mean to confine it to passwords per se.
18 What's the reason for having systems
19 that require an authorization before a person can
20 access the information that's on the system?
21 A. The deep reason is that you don't want to give
22 everybody access to everything you have. The
23 proximate reason is that as organizations connect
24 more to the World Wide Web, they're opening up
83
1 their internal systems to a degree that was just,
2 no one contemplated before. So as soon as you
3 have people potentially being able to access
4 anything on any of your computers, you want to
5 restrict access to all sorts of people within and
6 without the organization.
7 Q. Gotcha. But, for example, there is a utility in
8 having such systems for the purpose of exchanging
9 scientific and research information; is that
10 correct?
11 A. Having --
12 Q. I said such systems, namely systems that require
13 an authorization code to enter.
14 A. There's not a special one for scientific
15 information.
16 Q. I'm sorry, it was a general question.
17 A. Generally scientific and technical information,
18 well, from a university would go out free, but of
19 course in an industry, I mean, much of scientific
20 and technical information would be confidential.
21 Q. But research and scientific information can be
22 shared through such networks requiring an
23 authorization code, whether or not there's a
24 charge made for access; isn't that correct?
84
1 A. Oh, I wasn't thinking about consumer stuff at
2 all. A more typical arrangement would be two
3 companies doing joint research. They'll set up a
4 system that some group of people in company A can
5 share information with some group of people in
6 company B, and that's a very important mode of
7 collaboration because they might be working on
8 something that's confidential to the two of them.
9 Q. And would that include activities like reverse
10 engineering?
11 A. Sure it could.
12 Q. And is there any reason why that model would not
13 apply to any other kind of reverse engineering or
14 sharing of information in an engineering context,
15 namely authorization protected --
16 A. Wait a minute, I lost --
17 MR. GARBUS: I don't understand the
18 question.
19 A. I lost where you're saying reverse engineering.
20 Q. Let me just try to simplify it. You've talked
21 about the sharing of information over sort of a
22 private net through authorization and I'm --
23 A. Right.
24 Q. -- asking you whether that model has utility for
85
1 the sharing of scientific or engineering
2 information outside of strictly commercial
3 applications?
4 A. Yeah. Yes, certainly.
5 Q. Okay. Now, during the course of the testimony
6 that you gave Mr. Garbus, you referred a couple
7 of times to DeCSS player or recorder. I wanted
8 to make sure your testimony was clear in that
9 regard.
10 A. If I did, I shouldn't have said that, I should
11 have said DVD recorder or player.
12 Q. Okay. You also mentioned hops in the course of
13 your testimony.
14 A. Oh, okay.
15 Q. Can you tell me what an average string or trace
16 of hops would be for a normal Internet
17 communication? Is there an average?
18 A. I don't know if there's an average. Certainly 10
19 is pretty common. I wouldn't know what an
20 average is.
21 Q. The more hops there are, the slower the transfer
22 rate or the potential risk that the transfer rate
23 is going to be slower?
24 A. In general.
86
1 Q. And do you, the user, at one end of that chain
2 have any ability to control the number of hops by
3 how you route the signal?
4 A. No, almost never, other than in sort of very
5 experimental settings by people doing network
6 research. But the design of the Internet does
7 that routing adaptively on purpose. So there are
8 a couple of experiments that people are working
9 on that allow you to control that, but an average
10 user has no access to that.
11 Q. Do you?
12 A. No. I mean, for example, the people who do
13 network research at the laboratory for computer
14 science where I work, for the experimental
15 networks they're working on, are able to control
16 that.
17 Q. Gotcha. To improve network efficiency?
18 A. Right.
19 Q. And to reduce the number of hops?
20 A. Right. That's one.
21 Q. Now, you were talking with Mr. Garbus about the
22 consumption of bandwidth, if you will, when
23 multiple users are sharing files of different
24 sizes, right?
87
1 A. That's right.
2 Q. Is there a difference, to your understanding,
3 between a shared or hub-type network and a
4 switched network; and if so, what is that
5 difference?
6 MR. GARBUS: That's already been asked
7 and answered, but go ahead.
8 A. I don't -- I don't quite know how to characterize
9 that difference. I'm not quite sure what you
10 mean.
11 The particular estimates I was giving
12 were quite independent of that because they were
13 simply talking about what's the aggregate around
14 MIT as opposed to -- I think what you might mean
15 is there some master switch in which everything
16 at MIT has gone --
17 Q. No, no. Let me try and help you.
18 We were talking about bandwidth and
19 effective transfer rate as a function of
20 bandwidth, right?
21 A. Right.
22 Q. And you postulated that given the size of files,
23 even DivX files, that multiple users sending
24 multiple DivX files would consume available
88
1 bandwidth at a certain point, making the delivery
2 of movies over the Internet, at least now,
3 impractical; is that a fair --
4 A. Right. What I said --
5 Q. -- statement of what you said in your direct
6 testimony a moment ago?
7 A. Yes.
8 MR. GARBUS: Let the witness answer.
9 MR. HART: I'm sorry, Mr. Garbus?
10 MR. GARBUS: Go ahead, he's answered.
11 Q. And my question is whether that testimony was
12 based on something called a shared network or a
13 hub network rather than a switched network; and
14 in that context, if there is a distinction
15 between those things, I would like you to tell me
16 what your understanding is of that distinction?
17 A. I don't know. I don't think the estimates I was
18 making depend on that.
19 Q. In any way whatsoever?
20 A. I think someone can make many more detailed
21 estimates that would reflect on that, but I was
22 saying an overall generalization.
23 Q. Right. Again, not going to the particular
24 numbers in any case, number of minutes, but going
89
1 to the proposition that multiple users sharing
2 multiple files simultaneously would consume
3 bandwidth?
4 A. Yes.
5 Q. Going to that proposition and that proposition
6 only, my question is, is there -- do you have any
7 understanding of whether there's a difference in
8 regard to the use of shared networks and switched
9 networks as it affects consumption of bandwidth?
10 MR. GARBUS: He's already answered.
11 Go ahead.
12 A. Okay. I think I understand. This is something I
13 know about only vaguely.
14 Q. Only?
15 A. Only vaguely.
16 Q. Vehically?
17 A. Vaguely. Vaguely.
18 Q. I'm sorry.
19 A. I think at the level that I was talking about, it
20 doesn't matter, because I was talking about the
21 network backbone at MIT, and so -- how do I say
22 this? I don't know that you can set up a
23 switched connection at MIT, but assuming you
24 could, you're still drawing bandwidth from that
90
1 backbone. So it doesn't matter whether you're
2 sharing it or switching it at that level.
3 I know at MIT we have a hundred
4 megabits, and we can presumably choose to
5 allocate some of that by switching networks,
6 which I'm almost sure we do not, but it doesn't
7 make any difference for the aggregate numbers I
8 was talking about.
9 Q. Is it your understanding that a switched network
10 enables multiple users to use the same amount of
11 bandwidth at the same time without consuming the
12 amount of bandwidth that's available to other
13 users on the same network?
14 A. I don't think that's quite the right way to
15 characterize it. It has to do with the
16 granularity at which you're sharing things. At
17 some point there's only so much bandwidth. If I
18 switch among users rapidly enough, it may seem to
19 you that you're not sharing, but at the levels
20 where you start saturating the network, that
21 becomes absolutely apparent that the sharing is
22 going on.
23 Q. Does the presence of multiple servers have any
24 impact upon your views in that regard?
91
1 A. Be a little more precise. If I'm trying to get a
2 file and it exists at multiple sources, or simply
3 there's a lot of servers?
4 Q. Either one or both.
5 A. Well, it will have an impact, but the overall
6 estimates remain the same.
7 There are tricks you can do with
8 multiple servers to essentially cut down
9 bottlenecks and server delays. But, again, the
10 kind of estimates I was talking about was
11 independent of that. Simply talking about how
12 much the network itself can handle.
13 I'll certainly grant you that if you
14 generalize from MIT to the national network,
15 there's a lot more complexity, but I don't think
16 it changes the fundamental estimates.
17 Q. Now, is the type of network that you were
18 describing what one might call a hub-type
19 network?
20 A. I don't think so.
21 Q. What's the term for the type of network you were
22 describing?
23 A. At MIT, I believe MIT has what's called a ring.
24 So there's a central thing that lots of things in
92
1 the campus are connected to, and they all fit in
2 the ring.
3 Q. Are you aware if the Napster service depends upon
4 multiple servers?
5 A. The Napster service itself is just a directory.
6 I'm testifying beyond what I know, but my
7 assumption is they simply have a server
8 someplace. I don't know if they have multiple
9 servers or not.
10 Q. Now, the -- what's available to MIT students in
11 the dorm rooms, did you say it's 10 meg --
12 A. I believe --
13 Q. -- Ethernet?
14 A. I believe there's a mixture of 10 and a hundred.
15 Q. When you were talking about the value of DeCSS
16 versus not using DeCSS and you mentioned taking
17 the signal at the output of a player?
18 A. Right. I would imagine you can do that.
19 Q. Are you talking about the analog output or
20 digital output or what?
21 A. Well, certainly you could do it with analog
22 output. I would expect you could do it with
23 digital output. But, again, I haven't done it.
24 Q. Uh-huh. Is there any difference in your
93
1 professional experience between the signal
2 quality of a digital signal versus an analog
3 signal, particularly when it's being replicated?
4 A. Well, the answer is that if you make multiple --
5 digital copies generally are perfect, whereas
6 analog copies, you'll get distortion through
7 multiple copy generations.
8 But remember, there's another -- if
9 this is in reference to what I was talking about
10 making with Shamos's thing, remember that you
11 still have to go through a step of compression,
12 and you are eventually producing a digital copy.
13 If I take -- let's assume I make an
14 output, an analog output coming out of the
15 recorder. So I take that analog output, I
16 digitize it, I compress it, and that whatever
17 degeneration I've gotten from that process, after
18 that I can ship those things around and they can
19 be replicated perfectly.
20 So then you have to ask yourself, in
21 the overall degeneration between, of which most
22 will come from compression, I would guess, how
23 much extra quality loss you would get from going
24 from the recorder initially to analog as opposed
94
1 to doing a perfect copy from decryption, and I
2 was guessing that that wouldn't be noticeable.
3 But, again, I don't know.
4 Q. Gotcha. You testified earlier in response to Mr.
5 Garbus's questions that you regarded notice to
6 the proprietor of a system that's been hacked or
7 breached as the polite thing to do, among other
8 things?
9 A. Right.
10 Q. And you said that it was indicia of good faith?
11 A. (Witness nods.)
12 Q. You have to verbalize your response.
13 A. Yes, that's what I said.
14 Q. Does that mean that the absence of notice to the
15 proprietor would be, in your estimation, indicia
16 of bad faith?
17 A. Not necessarily. It would just be the absence of
18 indicia of good faith.
19 Again, sometimes -- not in your
20 particular hypothetical -- you might know that
21 things would be ignored.
22 Q. What do you mean?
23 MR. GARBUS: It seems to me to speak
24 for itself.
95
1 MR. HART: Well, I'm sure it does, but
2 I want to know what the witness means by that.
3 A. I know an example, and I would rather not give
4 names, of someone who discovered that because of
5 the security system of a new computer, you could
6 remotely turn on the microphone in that computer
7 and just hear everything that was happening in
8 the person's office. And the CEO of this company
9 was someone who usually didn't listen to any
10 kinds of suggestions, and I know the person who
11 got the CEO's attention by handing him a tape
12 recording of everything he said in the office for
13 the past three days. That's an example of, I
14 think, really informing someone that you've had
15 an effect.
16 Q. That was a more dramatic way of informing the CEO
17 in your example, right?
18 A. Right. This wasn't a computer on the market, it
19 was just something they were planning.
20 Q. Now, am I correct in understanding that it's up
21 to you whether or not you will be available for
22 the trial in this case?
23 A. It's more up to Hewlett Packard. It depends on
24 what kinds of meetings get scheduled, then
96
1 depending on what types of meetings, it's more or
2 less difficult for me to be able to appear.
3 MR. HART: I don't think I have
4 anything further. Thank you again.
5 MR. GARBUS: Thank you.
6 (Whereupon at 1:38 p.m., the
7 deposition was adjourned.)
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
97
1 C E R T I F I C A T E
2 I, HAROLD ABELSON, do hereby certify that I have
read the foregoing transcript of my testimony
3 given on July 5, 2000, and I further certify that
said transcript is a true and accurate record of
4 said testimony (with the exception of the
corrections listed below):
5
Page Line Correction
6
7
8
9
10
11
12
13
14
15
16 Dated at , this
day of , 2000.
17
18 HAROLD ABELSON
19
20 SIGNED UNDER THE PAINS AND PENALTIES OF PERJURY
21
22
23
24
98
1 C E R T I F I C A T E
2
3 COMMONWEALTH OF MASSACHUSETTS
4 SUFFOLK, SS.
5 I, Loretta Hennessey, Registered Merit
6 Reporter and Notary Public in and for the
7 Commonwealth of Massachusetts, do hereby certify:
8 That HAROLD ABELSON, the witness whose
9 testimony is hereinbefore set forth, was duly
10 sworn by me and that such testimony is a true and
11 accurate record of my stenotype notes taken in
12 the foregoing matter, to the best of my
13 knowledge, skill and ability.
14 IN WITNESS WHEREOF, I have hereunto set
15 my hand and Notarial Seal this 6th day of July,
16 2000.
17
18
19
20 Loretta Hennessey, RMR
21 Notary Public
22
23 My Commission Expires: 6/10/05
24