Testimony of Dr. Daniel Olsen, Brigham Young University
April 15, 1996
IN THE UNITED STATES DISTRICT COURT
FOR THE EASTERN DISTRICT OF PENNSYLVANIA
- - -
AMERICAN CIVIL LIBERTIES : CIVIL ACTION NO. 96-963-M
UNION, et al :
Plaintiffs :
:
v. : Philadelphia, Pennsylvania
: April 15, 1996
JANET RENO, in her official : 9:31 o'clock a.m.
capacity as ATTORNEY GENERAL :
OF THE UNITED STATES, :
Defendant :
. . . . . . . . . . . . . . . .
AMERICAN LIBRARY ASSOCIATION, : CIVIL ACTION NO. 96-1458
et al :
Plaintiffs :
:
v. : Philadelphia, Pennsylvania
: April 15, 1996
DEPARTMENT OF JUSTICE, et al : 9:31 o'clock a.m.
Defendants :
. . . . . . . . . . . . . . . .
HEARING BEFORE:
THE HONORABLE DOLORES K. SLOVITER,
CHIEF JUDGE, UNITED STATES COURT OF APPEALS
FOR THE THIRD CIRCUIT
THE HONORABLE RONALD L. BUCKWALTER
THE HONORABLE STEWART DALZELL
UNITED STATES DISTRICT JUDGES
- - -
APPEARANCES:
For the Plaintiffs: CHRISTOPHER A. HANSEN, ESQUIRE
MARJORIE HEINS, ESQUIRE
ANN BEESON, ESQUIRE
American Civil Liberties Union
132 West 43rd Street
New York, NY 10036
-and-
STEFAN PRESSER, ESQUIRE
American Civil Liberties Union
123 S. 9th Street, Suite 701
Philadelphia, PA 19107
2
APPEARANCES: (Continued)
For the ALA BRUCE J. ENNIS, JR., ESQUIRE
Plaintiffs: ANN M. KAPPLER, ESQUIRE
JOHN B. MORRIS, JR., ESQUIRE
Jenner and Block
601 13th Street, N.W.
Washington, DC 20005
-and-
MICHAEL TRAYNOR, ESQUIRE
Cooley Goddard Castro Huddleson & Tatum
One Maritime Plaza, 20th Floor
San Francisco, CA 94111-3580
For the Defendant: ANTHONY J. COPPOLINO, ESQUIRE
PATRICIA RUSSOTTO, ESQUIRE
JASON R. BARON, ESQUIRE
THEODORE C. HIRT, ESQUIRE
MARY KUSTEL, ESQUIRE
CRAIG M. BLACKWELL, ESQUIRE
Department of Justice
Federal Programs Branch
901 E. Street, N.W., Room 912
Washington, DC 20530
-and-
MARK KMETZ, ESQUIRE
U.S. Attorney's Office
615 Chestnut Street, Suite 1250
Philadelphia, PA 19106
- - -
Also Present: MICHAEL KUNZ
Clerk of the Court for the
Eastern District of Pennsylvania
- - -
Deputy Clerks: Thomas Clewley
Matthew J. Higgins
Audio Operator: Andrea L. Mack
Transcribed by: Geraldine C. Laws
Grace Williams
Tracey Williams
Laws Transcription Service
(Proceedings recorded by electronic sound recording;
transcript provided by computer-aided transcription service.)
3
1 (The following occurred in open court at 9:31
2 o'clock a.m.:)
3 CLERK OF THE COURT KUNZ: Oyez, oyez, oyez, all
4 persons having any matter to present before the Honorable
5 Dolores K. Sloviter, Chief Judge for the United States Court
6 of Appeals for the Third Circuit; the Honorable Ronald L.
7 Buckwalter and the Honorable Stewart Dalzell, Judges for the
8 United States District Court for the Eastern District of
9 Pennsylvania; may at present appear and they shall be heard.
10 God save the United States and this Honorable Court. Court
11 is now in session, please be seated.
12 JUDGE DALZELL: Good morning, everyone.
13 ALL COUNSEL: Good morning.
14 JUDGE SLOVITER: Good morning. Let's see, we're...
15 I think we have Mr. Olsen on the stand?
16 JUDGE DALZELL: Mr. Olsen? Yes.
17 JUDGE SLOVITER: And he has been previously sworn.
18 (Pause.)
19 MR. BARON: Good morning, your Honors, Jason R.
20 Baron for the Justice Department.
21 DANIEL OLSEN, Defendants' Witness, Previously Sworn,
22 Resumed.
23 REDIRECT EXAMINATION
24 BY MR. BARON:
25 Q Good morning, Dr. Olsen.
4
1 A Good morning.
2 Q You will recall that on Friday Mr. Ennis started off by
3 asking you a series of questions about your technical
4 expertise as it relates to the issues involved in this case,
5 do you recall that?
6 A Yes, I do.
7 Q You received your PhD in 1981 from the University of
8 Pennsylvania right here in Philadelphia, correct?
9 A That is correct.
10 Q And after more than ten years at Brigham Young University
11 you have recently been appointed to be director of the Human
12 Computer Interaction Institute at Carnegie-Mellon University,
13 correct?
14 A That is correct.
15 Q What was your specific background and expertise that has
16 led to this recent appointment?
17 A Carnegie-Mellon has one of the foremost computer science
18 departments in the world, they saw a need to address the
19 needs of people in using computers. I have a long background
20 of software expertise and how people use computers.
21 Q Now, with respect to your general expertise regarding
22 technical matters involving the Internet let me first ask,
23 have you ever created a new application that required
24 software to communicate over the Internet?
25 A Yes. Several years ago we were interested in what are
5
1 called multi-user interfaces, this is where multiple people
2 interact simultaneously across the Internet. To make that
3 work we had to develop a new protocol for communication
4 between those two applications.
5 Q Have you created software which integrates with the
6 Worldwide Web?
7 A Yes. I believe that there are two papers listed in my
8 declaration where what we were interested in doing is
9 enhancing the interactivity of the Worldwide Web. It is
10 currently somewhat restricted in what interactively you can
11 do. So, what we needed to do is take our user interfaces and
12 be able to download them via the Web and have them
13 automatically run on the user's machine, so that we could
14 enhance that ability. So, we built that software and
15 integrated it with existing Web technology, yes.
16 Q Have you also created software that automates E-mail?
17 A Yes. As part of that we were very interested in all of
18 the various other Internet activities that we could integrate
19 our software with. One of the first ones we used was being
20 able to from the user interface automatically generate E-mail
21 of various kinds.
22 Q Would it be fair to say that in creating all of this
23 software one needs a working familiarity with the standards
24 and protocols used over the Internet?
25 A Yes. I and my graduate students spent several months
6
1 studying those protocols and deciding how we could use them
2 in -- as part of our software and what we would have to put
3 in our software to be able to perform the same functions over
4 the Internet.
5 Q Now, let's turn to the specific areas Mr. Ennis touched
6 on regarding your expertise. First, based on your knowledge
7 of Surfwatch and other blocking software programs, and your
8 general knowledge and expertise in the field of computer
9 science, do you believe you are able to form an expert
10 opinion on how Surfwatch purports to function?
11 A Yes.
12 Q Would you please tell the Court what you believe the
13 level of expertise would be to form such an expert opinion?
14 A My opinion is based on two principles; one is called
15 computational complexity, that is, how difficult
16 computationally the problem is and what the nature of that
17 problem is; and the other is general data base and
18 communication technology.
19 Q Would your answer be the same for what technical
20 expertise is necessary to evaluate parental-control software
21 utilized by America Online?
22 A Yes.
23 Q The third area Mr. Ennis mentioned was with respect to
24 direct or third-party verification of credit cards, do you
25 recall that?
7
1 A Yes, I do.
2 Q Dr. Olsen, are you familiar with the computer mechanisms
3 that exist to process a transaction through Mastercard over
4 the telephone?
5 A Yes, we did look in a particular instance with IC verify,
6 we did look at exactly what a program, say, for example, a
7 CGI program from a Web Browser would have to do in order --
8 Q Excuse me, you may need to explain CGI.
9 A CGI is the common gateway interface. One of the
10 particularly nice features of the Worldwide Web and
11 particularly HTTP is that when a user makes a request for a
12 file or a named item the Web server, that is, on the content
13 provider's side does not necessarily have to have a file by
14 that name. What they can do is they can run a program, a CGI
15 program, which will go out and compute a file of the type
16 that the user has requested, this leads to a very powerful
17 mechanism.
18 Q Do you also have expertise in secured protocols that
19 would lend itself to issues involving third-party
20 verification over the Internet?
21 A Yes. One of the problems we had when we were downloading
22 user interface software is that in essence you were
23 downloading an executable program. If you don't know who you
24 downloaded it from then you could download it from a stranger
25 who could then do damaging things to your system. So, we
8
1 spent considerable time looking at existing technology for
2 how to protect ourselves there. We are currently finishing a
3 Master's thesis that points out several holes in the formal
4 technology and ways in which those things could be
5 circumvented by individuals.
6 JUDGE SLOVITER: Did you -- I didn't hear, did you
7 say that you are getting a Master's --
8 THE WITNESS: No, I am supervising a Master's.
9 JUDGE SLOVITER: No, you were supervising. I didn't
10 think -- I thought that was going backwards. Your student is
11 getting a Master's degree?
12 THE WITNESS: That is correct.
13 JUDGE DALZELL: And you're supervising that thesis?
14 THE WITNESS: And I am supervising that, yes.
15 BY MR. BARON:
16 Q You're the thesis advisor, as --
17 A That is correct.
18 Q Okay. Mr. Ennis also brought up your knowledge of PICs;
19 have you read the PICs technical specifications on labels and
20 services found at the PICs Website?
21 A Yes.
22 Q Do you see anything, Dr. Olsen, in those specifications
23 that represent anything other than normal, traditional
24 computer science?
25 A No, I don't see anything in that regard. A label under
9
1 the PICs standard is essentially a record, this is a standard
2 concept in computing. A -- sort of the threshold you would
3 set in a browser are simply a very simplistic numeric
4 mechanism for defining a class of records you want to
5 receive, we teach this at sophomore-level computer science.
6 Q Dr. Olsen, do you believe you can form an expert opinion
7 on how PICs technology can be utilized based on your study of
8 the PICs materials?
9 A Yes.
10 Q Dr. Olsen, am I correct that your very own field of
11 research has contributed to the development of PICs
12 technology?
13 A Yes. One of the issues that PICs faced was when they
14 wanted to have lots of rating services, but the parents would
15 only have, say, one browser, how are the parents with their
16 one browser going to set the controls for several rating
17 services. PICs has a very nice solution where they
18 distribute on the Web the information about what the controls
19 are, then the browser can automatically configure the user
20 interface to be able to set those controls. There is a paper
21 listed in my curriculum vitae on language-based
22 specifications that actually pioneered that user-interface
23 technique.
24 Q Dr. Olsen, the fifth and final area Mr. Ennis inquired
25 about concerning your expertise was with respect to issues
10
1 involving libraries, do you recall that?
2 A Yes.
3 Q Have you ever studied the costs involved in creating an
4 on-line digital library?
5 A Yes. The Family History Library in Salt Lake has about
6 20 million rolls of microfilm, currently the masters are
7 locked in a mountain in Little Cottonwood Canyon, this makes
8 them less accessible than people would like. The people that
9 run that library have been working with us for several years
10 as to what it would take to bring those materials and bring
11 them on-line. And we did some cost studies of what it would
12 take to scan them and what it would take to index them, we
13 spent several years working with them on that issue.
14 Q Have you ever studied or done any work on electronic card
15 catalogues?
16 A Yes. Again, in our distributed user interfaces research
17 we were very interested in providing interactive access to
18 other repositories other than Websites. So, we spent some
19 time working with the people at the BYU library,
20 understanding how the library electronically manages their
21 card catalogue. We looked very carefully at the Mark
22 standard, which is the one that libraries use to communicate
23 with each other, essentially that is a property list, which
24 is a standard mechanism in computer science. Yes, we did
25 look at that.
11
1 Q Have you ever done anything with naming methods for
2 electronic library materials?
3 A Yes. Again, with the Family History Library, they have
4 the unique problem of wanting to link together genealogies of
5 -- essentially their goal is the world, if they could
6 accomplish it, there are things about the way things are
7 named in the Worldwide Web that make that difficult. Once
8 you have linked up a genealogy you would not like it to be
9 broken. So, we spent some time developing new naming
10 technology for how to do that, so that once somebody decided
11 a person was related to a particular other person that link
12 would not easily be detached. That approach is actually
13 outlined in the papers we delivered to the plaintiffs.
14 Q One last question: Have you ever written any software
15 code that would enable the posting of library materials?
16 A Yes. In the same project with the Family History
17 Library, they have the largest ideological data base that's
18 on-line in the world, they wanted to make -- so, they wanted
19 to know if it was technically feasible to put that on the
20 Worldwide Web. I spent a couple of weeks, built the software
21 and demonstrated that it was; it's not available because of
22 copyright problems, but we demonstrated the technology.
23 Q Let's turn to the substantive topics on which Mr. Ennis
24 and Mr. Hansen cross-examined you on Friday. Dr. Olsen, you
25 were asked a number of questions about your Minus L18
12
1 proposal, could you succinctly explain to the Court what the
2 central points of that proposal represent?
3 A Most of the proposal is a response to some assertions by
4 Mr. Bradner that it would be exceedingly difficult to -- for
5 content providers to label their materials, most of the
6 proposal is a counter-example. It's a standard approach in
7 computer science that when someone says it can't be done you
8 disprove that by a counter-example, that's the purpose of the
9 L18. In essence the argument is that for content providers
10 to label their materials is technologically quite
11 straightforward. On the other hand, for a third party or for
12 parents to detect those materials without the assistance of
13 the content provider is computationally quite difficult.
14 That really is the central issue of what I stated in the
15 declaration.
16 Q You will recall that you were asked a number of questions
17 about PICs, you recall those?
18 A Yes.
19 Q Would you please explain to the Court the ways in which
20 your Minus L18 proposal is consistent with aspects of the
21 PICs proposal?
22 A The PICs proposal specifically lays out something called
23 self-labeling, I believe is what they termed it, whereby
24 content providers can classify their materials. Because the
25 L18 proposal was meant as a counter-example I did not
13
1 extensively develop it in any way, it was only meant to show
2 that it was possible. It is trivial to take the L18 proposal
3 and embed the same idea inside of PICs labeling. PICs
4 provides a more extensive way to describe the information,
5 but relative to this law they are the same in that regard.
6 Q How does your L18 proposal go further than PICs in terms
7 of enabling communications in a variety of applications over
8 the Internet to be labeled?
9 A PICs is restricted to being able to label something that
10 has a URL. So, for example you could label a news group, you
11 can label a file, you can label a site, et cetera. L18 can
12 do that also, they're pretty much the same in that regard.
13 There are many kinds of communication that do not have
14 specifically a name, for example, an individual E-mail
15 message does not have a name. For that, in my declaration, I
16 said you could take Minus L18 and put it in the subject line
17 and thereby tag it, so even though it doesn't have a name a
18 speaker could identify it. The same thing with a news group.
19 A news item can have a name, that is possible; however, news
20 items are so ephemeral, they live typically for a week or two
21 weeks, most of them, some actually get preserved, but most of
22 them, they appear and then some time later they disappear.
23 So that the task of actually rating individual news postings
24 and naming them in PICs would be very difficult, if not
25 impossible.
14
1 Q There's nothing magical about your having picked Minus
2 L18, correct?
3 A No, any string of characters that didn't have an English
4 meaning would work fine.
5 Q And nothing in your proposal does away with or eliminates
6 the continued use of more sophisticated schemes like PICs,
7 right?
8 A No, it was simply a counter-example, other things are
9 possible and probably should be used.
10 Q What does the term content selection standard mean to
11 you, Dr. Olsen?
12 A In my mind it was a mechanism or a way agreed upon within
13 the community, let's say the Internet community, whereby
14 people would identify particular kinds of content.
15 Q Is your Minus L18 proposal a type of content selection
16 standard?
17 A Yes, it would -- if adopted by a large portion of the
18 community it would be a standard for saying these things are
19 inappropriate for people under 18.
20 Q Now, Mr. Ennis asked you a whole set of questions on
21 whether cooperative technology presently exists which will
22 pick up the Minus L18 tag, do you recall those questions?
23 A Yes.
24 Q How does Surfwatch and other parental-control technology
25 already function to pick up such tags?
15
1 A If, as Mr. Ennis pointed out on Friday, you used XXX it
2 already picks it up. If Surfwatch was to add the string
3 Minus L18 to the data base they already distribute to their
4 customers that would be picked up and it would be recognized,
5 yes.
6 Q Tell the Court about your Netscape proxy server
7 experiment and why it's relevant to the issue of available
8 technology?
9 A That was a specific response to the fact that no software
10 exists that could do blocking on a tag. What we did is we
11 took the Netscape proxy server, we specifically put in it the
12 regular expression that would identify any URL with Minus
13 L18, and then we checked to see if it was blocked, they were.
14 We then said can we block it and require a password, we did.
15 This took us about four hours of work, again, it was a
16 counter-example. It was asserted that this was hard or
17 impossible, it was not.
18 Q Could you also tell the Court about your Eudora
19 experiment?
20 A We were looking at the same issue related to mail
21 programs. Eudora is a client mail program, it's the one I
22 happen to use that receives mail over the Internet. In
23 Eudora it has a filtering technique, as I have described. I
24 took ten minutes, put in the Minus L18 tag to see if it would
25 filter, it did. The minor difficulty with this particular
16
1 experiment is that, unlike the proxy server, what I did could
2 be easily undone. The point, however, being that if Eudora
3 wanted to it would take them an hour to make that code so it
4 couldn't be undone. But the technology for checking is
5 there, we tried it, it works.
6 Q Do you have any reason to believe that the folks at
7 Netscape couldn't repeat your experiments and incorporate
8 them into their existing software?
9 A It would be very easy for them to do.
10 Q The same question for Microsoft, do you have any reason
11 to believe that Mr. Bill Gates couldn't basically do the same
12 thing with respect to Microsoft software and browsers?
13 A I suspect Mr. Gates would have to ask one of his
14 programmers to do it, but it could be done.
15 (Laughter.)
16 Q The same question with respect to America Online, Prodigy
17 and Compuserve, could they do the same thing?
18 A I see nothing that is difficult about this technology,
19 they could do it very easily.
20 Q Now, in response to Mr. Ennis' questions you talked about
21 a notion of, quote, "statistical assurance," could you
22 explain to the Court what you meant by statistical assurance?
23 A Statistical assurance has to do when you're dealing with
24 not an absolute is it or is it not possible, but what is the
25 probability of something. I looked specifically at that
17
1 because my understanding of the law was that a hundred
2 percent was not required under the law. So, statistical
3 assurance says to some extent that, for example, browsers
4 which pick up L18 or browsers that pick up PICs have been
5 deployed, if they have been deployed to, say, 90 percent of
6 all client sites then you are 90 percent assured that if you
7 have labeled in accordance with that that your material will
8 not get through to minors, if they're deployed to 99 percent
9 of the sites then you're 99 percent assured.
10 Q How high would your statistical assurance be if Netscape,
11 Microsoft, America Online, Prodigy and Compuserve fixed their
12 browsers and software to incorporate tags such as the L18
13 tag?
14 A Microsoft -- excuse me, Netscape claims to have 80
15 percent of this market. Therefore, if all of their current
16 customers updated, which they almost always do because new
17 features are added that they want, then that would lead one
18 to believe you have 80 percent coverage. Assuming that
19 Microsoft is reasonably effective at overcoming -- or
20 grabbing off the remaining 20 percent or perhaps snatching
21 away that 80 percent, then your coverage is increased to
22 above 90 percent.
23 Q Now, how would a consensus standard or convention develop
24 around a proposal like Minus L18?
25 A There are several ways that it could be done. It could
18
1 be done very formally through the IETF; it could be done
2 informally through news group communications, a lot of things
3 are done that way; it could be done, as Mr. Vezza has
4 testified, his W3C consortium is very interested in doing
5 exactly this kind of thing, it could be done that way.
6 Someone like Netscape, who dominates a market, they have
7 already shown their willingness and ability to set standards,
8 they would simply say this is the standard we are using and,
9 given their market dominance, that would quickly be adopted.
10 Q Do you think Mr. Bradner and his colleagues at the IETF
11 have the technical expertise to sit down and design such a
12 standard?
13 A Easily.
14 Q Would you even need a consensus on such a standard, that
15 is, what I'm asking, could individuals who use Minus L18
16 essentially contact Surfwatch and other companies like
17 Netscape and Microsoft and say, here's my site and here's the
18 label, please block the site using a key-word search for
19 Minus L18?
20 A They could easily do that.
21 Q And how do the market forces surrounding the rollout of
22 the PICs-compatible software influence the ability of
23 browsers to pick a Minus L18 tag?
24 A Those market forces are similar to what would be required
25 for Minus L18, the market forces that Mr. Vezza discussed. I
19
1 think it's a very compelling case that the people who produce
2 this software are very interested in providing parental
3 controls. They see an enormous home market, they see the
4 extent that the public feels that these kind of materials are
5 available to their children, that that market has diminished.
6 So, there is a very strong motivation on the part of the
7 people who produce this software to sort of make the problem
8 go away. So, they have already bought on the PICs, which is
9 far more complicated than what's required for L18, they may
10 or may not adopt to L18 or they may just go ahead with PICs,
11 as they have said, either way would accomplish the arguments
12 I have presented.
13 Q Let's turn to third-party schemes; would you please
14 explain to the Court, in your view, how do tagging schemes
15 without the required cooperation of content creators
16 -- strike that. In your view, do tagging schemes without the
17 required cooperation of the content creators have less chance
18 of assuring the screening of inappropriate material?
19 A The simple answer is yes, they do have less of a chance.
20 If I could clarify that a little, one approach for a label
21 bureau, which is what's outlined in the PICs proposal, or
22 someone like Surfwatch or Netnanny or one of the others, one
23 of the approaches they can take is to say that we will go out
24 and we will look at all of the Internet sites, and we will
25 decide which ones need to be blocked and we will put this in
20
1 a data base. Surfwatch takes that data base and distributes
2 it to the individuals, PICs proposes that that data base be
3 kept on some Website somewhere, so that the individual's
4 computer is not encumbered with that. Either way, what it
5 means is it means that some entity has to, number one, find
6 all of the Internet sites, which as Mr. Bradner has testified
7 is difficult even to count them, let alone find them, they
8 would have to find those sites, they would then have to look
9 at the materials and then have to make a judgment. They
10 would also -- this is particularly problematic, because
11 again, as Mr. Bradner testified, the number of sites in the
12 Internet is doubling every nine months. So, this means that
13 for a label bureau to be successful they would have to do as
14 much work in this nine months as they have ever done in the
15 history of their company, that's a significant challenge and
16 I'm not sure how they would ever accomplish that. Now, that
17 assumes that what you're going to do is you're going to find
18 all of the sites that are a problem and that means that when
19 you find such a site you are going to block the entire site.
20 Well, this is problematic and PICs tries to get around this,
21 because there are certain nonprofit sites that may have very
22 specific areas that are sexually-explicit and most of the
23 site is not.
24 You would have a problem with the Surfwatch
25 approach, which is as I understand it sites only. PICs
21
1 allows you, however, to label for a label bureau to tag a
2 specific file or class of files. Well, the problem with that
3 is that now instead of having the problem of finding all of
4 the sites we have to also now find and document all of the
5 possible files on all of those possible sites, and files are
6 growing at an even faster rate than the number of sites,
7 because they're easier to create. So, we essentially -- if
8 we adopt a label bureau approach or a third-party approach we
9 have set for them the task of taking an exponentially growing
10 problem and trying to monitor that problem.
11 What we have also set for them is that they must
12 somehow get income for doing this, this is very expensive.
13 The current approach under Surfwatch, Netnanny and others is
14 to sell a subscription to parents. Well, if the work
15 required to do this is exponentially growing we have a
16 problem with where those costs might go in the long run.
17 Q Are there other problems with key-word searches
18 associated with this kind of scheme?
19 A Yes. Another one of the technologies that parental
20 controls are using is to search for individual words. The
21 problem there is the words, unlike something like L18 or
22 unlike a PICs label, which are specially designed to have
23 computable meaning, key words have English meanings. For
24 example, if you search for the word sex and say I will not
25 allow anything to go through that contains the word sex it is
22
1 possible that you might screen out sites that you would not
2 consider offensive but did mention sex. However, if you
3 search for the word sex you would have problems where sex is
4 embedded in a larger name. For example, many of the
5 directories I looked at had the name Hotsex, well, that's now
6 a different word. Well, we would either collect all of the
7 ways people would assemble adjectives and smash them
8 together, computer scientists regularly eliminate spaces and
9 periods and vowels and other stuff. So, we could say, okay,
10 well, what I'm going to do then is I'm going to search for
11 the three characters s-e-x. Well, that's a possibility too,
12 but now anything related to Essex County is caught and
13 screened and the child cannot look up Essex County, they
14 could not look up Middlesex County. So, what we have done is
15 by using English terms we have been very imprecise about what
16 we want to screen. Another example is one might want to put
17 playboy in as something I want to screen for, I don't want
18 that to go to children. People who are fighting against such
19 restrictions might say, well, I will go to playmate. Well,
20 if I go to playmate I could very easily then screen out a
21 chat room for kids, because it has a different meaning among
22 children than it does among people interested in sexually-
23 explicit material. So, the more words I add to the list, the
24 more things that are appropriate for kids that will be
25 screened out; the less words I add to the list, then the more
23
1 things that I want screened out will come through. The
2 problem is as you're trying to use English in a precise form,
3 for a precise purpose, all of the library people and the
4 information retrieval people have recognized this problem and
5 where possible they always recommend assigning a tag or a
6 classification, like the Library of Congress classification,
7 so that they can get around this imprecise-meaning-of-English
8 problem.
9 Q Now, you were in the courtroom Friday when Mr. Vezza
10 described business models for third-party rating schemes
11 developing, correct?
12 A Yes.
13 Q Do you see specific problems arising with third-party
14 rating services such as Disney or the Boy Scouts actually
15 entering into contracts with sites that wish to label
16 content?
17 A As I remember the interchange in that testimony it was
18 -- Mr. Vezza was asked how would you get around this problem
19 I have described of how do I track all of these sites and how
20 does a third party do all of this labeling. The business
21 model he proposed that would solve that was that labeling
22 bureaus would enter into contracts with the content providers
23 and, based on the force of the content -- contract, the
24 content providers would do the labeling. The problem is is
25 that -- there are two problems: If for example Disney is the
24
1 labeling bureau I see no reason why the purveyors of serious
2 pornography want to enter into a contract with Disney,
3 they're not interested in what Disney thinks of them. There
4 is also a problem with why would Disney want to make the news
5 by having contracted with some pornographer to do the ratings
6 for them, they want to stake their reputation on the fact
7 that they're doing the ratings themselves.
8 So, I agree with Mr. Vezza to the extent that if we
9 can provide some legal force by means of a contract or by
10 means of a law that the content providers can do an excellent
11 job of labeling the stuff, but I see a lot of people who
12 would not voluntarily enter into such a contract.
13 JUDGE DALZELL: But haven't you said repeatedly, as
14 Mr. Vezza said, that the market forces -- I said and he
15 agreed with me that the market forces were enormous to solve
16 the problem that brought about this law, for example, and
17 that brings us together, and you agree with that point, don't
18 you?
19 THE WITNESS: Yes, I do agree with that point.
20 JUDGE DALZELL: Because huge parts of the market are
21 not even getting onto the Net precisely because of the
22 existence of the material that are in these binders.
23 THE WITNESS: That is correct, but those market
24 forces are brought to bear on the people who create browser
25 software, those market forces that we talked about are not
25
1 brought to bear on the people who produce the content. The
2 people who produce the content have no motivation whatsoever
3 to cooperate.
4 JUDGE DALZELL: Oh, sure, that's true. And wouldn't
5 you agree with the fact that we've heard testimony, I don't
6 think it's in dispute, that about 40 percent of the sexually-
7 explicit material is created offshore?
8 THE WITNESS: I couldn't characterize, I haven't
9 looked, frankly.
10 JUDGE DALZELL: Well, assume that's true.
11 THE WITNESS: Okay.
12 JUDGE DALZELL: If 40 percent is controlled offshore
13 or created offshore how in the world is anything that's going
14 to happen in the United States going to affect that?
15 THE WITNESS: To the extent of them providing
16 leadership of how it can be controlled for those countries
17 who are interested in controlling, it would help. To the
18 extent of forcing a country who is not interested, I don't
19 see how it has effect, but I didn't consider that because I
20 didn't know that the CDA would apply to offshore.
21 JUDGE DALZELL: Well, that would be an interesting
22 subject by itself. I'm sorry to interrupt you.
23 BY MR. BARON:
24 Q Well, let me just pick up on Judge Dalzell's question.
25 If U. S. sites were tagged -- if U. S. sites tagged their
26
1 speech using a Minus L18 approach, as you suggested, would it
2 be easier or more difficult for the Surfwatches and other
3 blocking software to search for foreign sites to block?
4 A Okay, if for example the CDA were upheld and people used
5 tagging or some other mechanism to take care of that, that
6 means now that the other 60 percent, using your 40 percent
7 figure, the other 60 percent is now no longer of interest to
8 Surfwatch and Netnanny and they can now concentrate on the 40
9 percent, which is a smaller problem.
10 Q And based on your knowledge and experience of the
11 Internet do you believe that the U. S. sets a leadership role
12 on standard setting that would influence the behavior of
13 foreign speech providers on the Net?
14 A In a technical sense, yes, most of the standards that
15 exist on the Internet were begun in the U. S. Some of them,
16 for example the Worldwide Web, began elsewhere, but generally
17 the U. S. has a leadership role technically.
18 Q We now turn to Mr. Hansen's questions to you Friday. Do
19 you recall being asked hypotheticals about one or more
20 nonprofit groups having the problem of having to rate 14,000
21 pages or files on their Website?
22 A Yes.
23 Q Mr. Hansen used EFF, the Electronic Frontier Foundation,
24 as an example, do you recall that?
25 A Yes, I do.
27
1 Q Do you know if EFF rates their site with PICs?
2 A Not to my knowledge.
3 Q If a parent was blocking any site that was not rated
4 under the PICs methodology would children be able to reach
5 the EFF?
6 A No.
7 Q Assuming that a tag or label -- that browsers detecting
8 tags or labels were widely available, could the EFF tag their
9 entire site as not for minors?
10 A Sure.
11 Q Minors would still not receive any of the EFF speech
12 whether parents blocked all unrated sites or EFF tagged them
13 -- tagged their entire site as inappropriate for minors,
14 correct?
15 A Either way minors would not receive it.
16 Q If the EFF tagged their whole site would children still
17 have access to the entire remainder of the Internet?
18 A Yes, they would.
19 Q If EFF tagged their whole site and the parents left their
20 browser open to all sites not rated as inappropriate would
21 the child have access to more of the Internet?
22 A Yes, they would.
23 Q You also discussed some transition costs on Friday; how
24 would you go about identifying text materials that are
25 sexually-explicit under a very short time frame as Mr. Hansen
28
1 proposed?
2 A Yes, Mr. Hansen I believe posed a very specific case of a
3 very short transition time. In terms of the six fifteen
4 transition time, nothing can be done in software by six
5 fifteen. If we impose a more reasonable transition time of a
6 month or two, you have to consider textual materials and you
7 have to consider images, I guess the first one I'd look up is
8 textual materials.
9 Q I asked you about textual materials.
10 A I'm sorry, textual materials. If I had the
11 responsibility the first task I would take is mark the entire
12 site as not accessible to minors as a temporary measure, this
13 doesn't prohibit adults looking at it or anything else. The
14 next step would be to identify some words that I would like
15 to search for. In this case I can adopt quite a broad
16 screen, because what I'm really hunting for in this first
17 step is to find out of all of my materials those materials
18 that I know I can turn loose to kids and it won't be a
19 problem.
20 So, one approach would be to pick a collection of
21 words that we think are sexually-explicit or otherwise
22 inappropriate, I would go to the thesaurus and expand that
23 collection with all of their synonyms that I could find. If
24 I was feeling really conservative I could expand it again,
25 but let's assume I didn't. I now have a set of search words.
29
1 Most large sites of that size already have a whole-word
2 index, but let's assume they didn't, because if they do have
3 a whole-word index then finding all of the documents that
4 contain those words is very easy, but assuming they don't one
5 could write a Purl (ph.) script that could run overnight and
6 examine all of their documents and flag them. Now what I
7 would do is I would take all of the ones that weren't found
8 that way and un-tag them, these are inappropriate for minors.
9 So, in a matter of less than a week I've now taken most of my
10 site and re-exposed it to minors. Now I have a few files
11 left and those files someone would have to manually look at
12 and make a determination. At no point have any of these been
13 denied to adults.
14 Q And how would you go about identifying sexually-explicit
15 images?
16 A Images are another challenge, because sort of the key-
17 word search technique doesn't help here, I don't know of a
18 technique that will identify them. After Mr. Hansen's
19 questions I sat down with my calculator and made the
20 following assumption: Assuming all 1400 are images, which
21 they are not, but assuming they were --
22 Q I believe Mr. Hansen said 14,000.
23 A 14,000, excuse me. Assuming all 14,000 are images and
24 assuming I could hire someone or take someone on my staff and
25 say I want you to look at each one of these, all I want to
30
1 know is is there a sexual organ or act or an excretory organ
2 or act in this picture, that's all I want to know, in which
3 case, assuming I gave this person a full 15 seconds to make
4 this determination, and that's a long time to look at a
5 picture and say is there a sexual act or organ here, then it
6 would take such a person about two weeks to have made this
7 judgment on every single -- all 14,000 images. Now, my
8 projection would be that most of those images are not. I
9 mean, certain sites this would not be true, but those sites
10 already know that they have this problem. So, most of those
11 images would be not, in two weeks I have taken the vast
12 majority of my images and said there is no problem here, and
13 I have opened them up to minors. Now I take the last group
14 and I say how badly do I want to tell minors or show minors
15 these pictures. Well, if I think really badly then I might
16 have to make a more careful analysis and I probably have to
17 involve my lawyers, but we very quickly screen, taking the
18 vast majority of our material, and made it available to
19 minors in a couple of weeks.
20 Q And what about ongoing costs in terms of looking at text
21 and images?
22 A Okay, this is a matter of the judgment calls being made
23 at the time the material is produced. In the case of text
24 materials, if a site was particularly concerned about their
25 text materials they could very easily put a piece of software
31
1 in place that would periodically check for all new pages and
2 check to see if they had any of the words that we ought to
3 look at, that's one way. Again, what I'm looking at here is
4 not a precise screen, but a screen by which manual
5 intervention can be required. What we're trying to do is
6 take the vast majority of the stuff and simply not look, we
7 know it doesn't have a problem. So, your initial screen
8 simply tells you that small amount of material that you
9 actually need someone to look at and make a human judgment.
10 In the case of my images, again, if I simply publish
11 to the staff, if it contains a sexual or excretory organ or
12 act please let us know, so we can check it before you
13 actually put it up or mark it as not available to minors,
14 take the conservative approach, in which case we may not
15 care.
16 Q Mr. Hansen asked you about Adultcheck and your knowledge
17 of whether third-party registration services such as
18 Adultcheck cater to pornographic sites, do you recall that?
19 A Yes.
20 Q In your opinion, Dr. Olsen, is there any technical
21 problem with other types of third-party registration services
22 catering to non-pornographic sites arising on the Internet?
23 A Yes. If Mr. Hansen's clients didn't like that particular
24 neighborhood they could form their own version of Adultcheck
25 and have a nicer neighborhood.
32
1 Q Your testimony is that there is not a technical problem?
2 A There is not a technical problem, they could do exactly
3 what Adultcheck did.
4 Q Just on a stray topic here, the Court inquired how many
5 teenagers you knew who could reinstall an operating system,
6 do you recall that question?
7 A Yes, I do.
8 Q You said you knew four or five kids in every high school
9 that could do that and you characterized this number as a,
10 quote, "small majority," do you remember that testimony?
11 A Yes.
12 Q Did you mean small minority?
13 A Yes, I did. I'm sorry.
14 Q Your proposals are not limited to just Minus L18,
15 correct?
16 A That is correct.
17 Q Your declaration sets out a number of options, right?
18 A That is correct. One of the things I intended to do in
19 the declaration is provide a broad menu of possibilities.
20 Q And you have not ruled out using registrations of URLs
21 with specific directories as one way to reduce the
22 availability of sexually-explicit speech to minors on the
23 Net, correct?
24 A Yes, if there was some directory or Surfwatch or Netnanny
25 I could very easily notify them and save them the labor of
33
1 having to search my material, and that would create a barrier
2 to minors, yes.
3 Q Now, you've been working on these problems just for the
4 past few weeks, correct, Dr. Olsen?
5 A That is correct.
6 Q In your view, is there sufficient creativity in the
7 Internet community to come up with a variety of other
8 solutions to the issues of screening based on available
9 technology?
10 A To the extent that I took a couple of weeks, generated
11 several technically-feasible solutions, there are a lot of
12 people who have done a lot of work on the Internet, I have
13 every reason to believe that they could generate many more
14 than I have thought about.
15 Q In your view, Dr. Olsen, are the technical issues
16 involving and insuring compliance with the CDA as difficult
17 to solve as other issues facing on the Internet, like
18 electronic commerce security issues?
19 A No, no. We're not talking about an issue here of
20 entering into a contractual agreement to pay for a service,
21 which is one of the things people really want to do on the
22 Internet, we're not even talking about the difficulties of
23 actually getting all of the Internet to talk to each other,
24 that was a far more difficult problem. If we have the
25 cooperation of the content providers in labeling their
34
1 materials, this is a trivial task, without their cooperation
2 it's a very difficult task.
3 Q One last question, Dr. Olsen: Do you believe that the
4 use of the Minus L18 tagging scheme you propose would have
5 any adverse effect on the growth or use of the Internet?
6 A No. For the people who are not producing sexually-
7 explicit materials, which constitute most of the content
8 providers, they frankly don't care, and they would not be
9 involved and they would not have any obligation of any kind
10 and could happily go their way.
11 Q Your proposals would not have an adverse effect on the
12 Net as a whole?
13 A Absolutely not.
14 MR. BARON: I have no more questions, your Honor.
15 JUDGE SLOVITER: Thank you.
16 JUDGE DALZELL: Recross?
17 (Pause.)
18 MR. ENNIS: Bruce Ennis, your Honors, for the ALA
19 plaintiffs.
20 RECROSS-EXAMINATION
21 BY MR. ENNIS:
22 Q Dr. Olsen, you began by saying that it is simple for a
23 speaker to add a four-character string to speech for which
24 such a tag is appropriate, correct?
25 A That is correct.
35
1 Q But the speaker will add such a string after the judgment
2 has been made that that string is appropriate for that
3 particular speech?
4 A They could do that or they could add it conservatively if
5 they thought there was a doubt and defer making that speech
6 available to minors to some later time when a determination
7 had been made.
8 Q But, in any event, before the simple act of typing in
9 those four characters there's going to have to be a human
10 judgment about whether typing in those four characters is
11 appropriate, correct?
12 A At any time you want to block material from minors
13 somebody will have to make a human judgment, that is correct.
14 Q Is it correct that someone has to make a human judgment
15 to block material from minors if you're using the PICs system
16 and the PICs browser is set to reject all speech that's not
17 tagged or labeled, then you don't require a human judgment,
18 do you?
19 A Other than the judgment to lock up most of the Internet
20 and put the kids in a ghetto of their own.
21 Q Well, that's a parental judgment?
22 A Other than that judgment your content providers would not
23 have to do anything.
24 Q But with respect to any system that does require the
25 speech to be tagged or labeled, your proposal, that kind of
36
1 system would require human judgment as to what to tag?
2 A That is correct.
3 Q And I assume you agree that there is no -- I think you
4 just testified there is no technology that can make that
5 judgment with respect to images?
6 A That's correct.
7 Q That's not possible to automate, a human being is going
8 to have to look at the image and decide whether the image is
9 appropriate or inappropriate for minors?
10 A That is correct with a minor caveat. It is conceivable
11 that image processing technology could improve to do that,
12 but using today's technology it is impossible.
13 Q All my questions are assuming using today's technology,
14 today that is not possible?
15 A That is not possible.
16 Q And in fact even with respect to words there can be
17 patently-offensive descriptions that don't use one of the
18 seven dirty words, correct?
19 A Absolutely, that's I believe stated in my declaration.
20 Q So, it's not enough just to search for key words that
21 would in some way be themselves dirty or offensive for
22 minors, correct?
23 A Correct.
24 Q There could be a combination of otherwise inoffensive
25 words which in combination produces a patently-offensive
37
1 depiction, correct?
2 A Correct. What I have just described to you in the
3 testimony earlier has to do with how you could use technology
4 to make a pre-screen, this is based on the assumption that
5 the content provider is completely ignorant of what the
6 materials are that are being posted, most content providers
7 are not that ignorant.
8 Q Well, you're making a distinction between the content
9 creator and the content provider in assuming that the content
10 provider is also the content creator?
11 A I haven't made a clear distinction, I'm more interested
12 in the creator than the actual provider.
13 Q Well, let's make a clear distinction. Let's suppose
14 you're a library and you're putting 2,500 magazines on-line,
15 every time the new issue comes out it goes on-line, you're
16 not creating that content, you're not the editor, you didn't
17 write it, you didn't look at it, you just put it on-line,
18 right? Now --
19 A That's correct.
20 Q -- there's a difference there, the library is not going
21 to know in advance the content of all those 2,500 magazines?
22 A That depends, that depends on what they expect of whoever
23 provided the content to them.
24 Q Well, let's suppose Vanity Fair, which is one of the
25 magazines the declarations indicate is on-line today, is the
38
1 library going to know in advance the content of each issue of
2 Vanity Fair?
3 A No, but the library may enter into an agreement or
4 libraries in general could enter into agreement to ask
5 magazines to identify the materials.
6 Q All right. Suppose a magazine --
7 A They could, for example, this is what libraries already
8 do with most books, most books come with a Library of
9 Congress categorization.
10 Q Then someone at the library is going to have to be
11 responsible for getting that information from Vanity Fair and
12 all of the other 2,500 magazines --
13 A Or somebody at Vanity Fair.
14 Q Well, somebody at the library has to receive it and then
15 make a human judgment about how to tag or label that before
16 putting it on line, correct?
17 A This presumes Vanity Fair did not make the judgment
18 beforehand, yes.
19 Q Well, Vanity Fair doesn't make the judgment, Vanity Fair
20 says, I'm going to have a cover image of Demi Moore and she's
21 partially nude, somebody at the library has to make a
22 judgment about whether that's offensive or not, correct?
23 A The library would either make that judgment or they would
24 make the judgment that we're just not interested in
25 distributing Vanity Fair to minors --
39
1 Q All right.
2 A -- and tag it anyway.
3 Q Now, let's suppose the library runs a key-word search and
4 suppose all of these magazines have the data base and
5 technology and it's there, which I think is a big assumption,
6 assume that's so and you find that somewhere in these -- each
7 of these 2500 magazines there is one word or two words or
8 three words that might be considered offensive, does the
9 librarian then have to go to the magazine and look at those
10 words in context, see how many of them there are?
11 A You're asking me to make a judgment as to how the library
12 would decide whether or not they had met the CDA, I'm not
13 prepared to make that evaluation.
14 Q Well, suppose each of the 2500 magazines has one of the
15 dirty words in it, the seven dirty words, you just
16 automatically classify them all as inappropriate for minors
17 or do you make a human judgment?
18 A You could classify the particular issue or the particular
19 article.
20 Q Which way would you classify it, inappropriate for
21 minors?
22 A Inappropriate for minors, that's a possibility, but
23 that's a judgment, what you're asking for is do I know how
24 libraries would go about classifying materials and the answer
25 is, no, I don't.
40
1 JUDGE DALZELL: In other words, if in one issue -- I
2 want to get this right -- if let's say The Economist is on-
3 line and if in one issue the word fuck appears that under
4 your proposal the whole issue would be blocked?
5 THE WITNESS: Not necessarily, not necessarily, it
6 depends on --
7 JUDGE DALZELL: Well, wouldn't the library have to
8 do exactly what Mr. Ennis just said, it would have to go
9 through all of the content and tag -- I realize your system
10 would allow you to tag the word fuck so that it wouldn't be
11 accessed, but the Carnegie Library would have to do that if
12 The Economist didn't. And the reason I give The Economist is
13 because it's based in the United Kingdom.
14 THE WITNESS: Somebody would have to -- to have to
15 make this screen. As far as -- I would like to oppose a
16 little bit the characterization that the entire issue would
17 have to be screened or even the entire magazine would have to
18 be labeled. The granularity is quite flexible as to how
19 deeply you wanted to actually do your labeling, but you are
20 correct, somebody would have to make this judgment.
21 BY MR. ENNIS:
22 Q Is it quite flexible if you, the librarian, risk going to
23 jail for two years if you make the wrong judgment and you put
24 on-line material that is found to be patently offensive for a
25 minor?
41
1 A Going to jail is a legal opinion.
2 MR. BARON: Objection. This line of questioning
3 presumes that this witness is answering legal conclusions,
4 he's a lawyer, he knows what the CDA's legal import is, none
5 of that is true, he's just a technical expert.
6 JUDGE SLOVITER: Well, his testimony went pretty far
7 in terms of the implications of his proposal, so I think that
8 we ought to let them cross-examine him.
9 BY MR. ENNIS:
10 Q Dr. Olsen, I believe you testified that your proposal
11 would require the cooperation of entities other than the
12 speaker and that without that cooperation it would be a very
13 difficult task to protect minors from inappropriate material,
14 correct?
15 A If there isn't some filter at some place along the
16 communication chain the tag alone is not sufficient, that is
17 correct.
18 Q And you mentioned that the entities that might have to
19 cooperate would include the people who create browsers, such
20 as the Netscape Navigator, correct?
21 A That is correct.
22 Q And the on-line service providers, correct?
23 A On-line -- not necessarily. To the extent that the on-
24 line service provider provides the browser --
25 Q All right.
42
1 A -- but -- or the on-line service provider, I believe in
2 my declaration there's the discussion of the provider could
3 if they wanted do this by means of a proxy server.
4 Q Or even the end-user blocking software, like Surfwatch,
5 Cyberpatrol, Netnanny --
6 A Absolutely.
7 Q -- they could change their software to recognize your L18
8 tag, correct?
9 A Or the PICs tags.
10 Q Or PICs, all right. But the central point is your
11 tagging proposal does require the cooperation of entities
12 down the communication pipeline?
13 A It is effective to the extent that down the pipeline
14 screens have been deployed, that is correct.
15 Q Do you know whether the Communications Decency Act
16 requires any cooperation or any such effort by any entity
17 down the communication pipeline?
18 A I am not aware of any such requirement.
19 Q Have you read the Act?
20 A I have read the pieces you showed me.
21 Q And that's all?
22 A I believe I read a couple of other pieces that Mr. Baron
23 showed me.
24 Q Did you read the conference report?
25 A No.
43
1 Q Are you aware that Congress made a considered decision to
2 impose no requirements on entities down the communications
3 chain whatsoever?
4 A That is my understanding.
5 Q So, your proposal would be directly contrary to the
6 policy choice Congress has already made?
7 A That's not what I said. What I said is that market
8 forces, which we have already had testimony on and which I
9 believe in, would provide the impetus and the legal impetus
10 in my mind is not required, but I do not know of a legal
11 impetus for that cooperation, no.
12 Q All right. Now, you testified that if all of these
13 entities down the line did cooperate, say, voluntarily that
14 you might then have a perhaps 90 percent statistical feeling
15 of security that if you're the speaker and patently-offensive
16 material would not reach minors, correct?
17 A Correct.
18 Q Of course that means patently-offensive material would be
19 reaching ten percent of the people it shouldn't reach?
20 A Possibly.
21 Q And if you are at risk of criminal prosecution if your
22 material reaches ten percent of the population that might be
23 a concern for you?
24 A You're asking me I believe for a legal judgment as to
25 what effective means and I can't make that --
44
1 Q Well, suppose you're the speaker, would you consider that
2 to be effective enough that you would feel comfortable in
3 putting your speech on-line?
4 A If I was a speaker I would consult my lawyer as to
5 whether or not I was meeting the CDA.
6 Q Now, your 90 percent statistical significance figure
7 assumes, does it not, that all of the speech has been
8 properly tagged and labeled according to your proposal?
9 A Correct.
10 Q And that assumes that all of the speech that originates
11 abroad by foreign speakers has been tagged and properly
12 tagged?
13 A I think I have stated previously that we have not done
14 anything relative to foreign speakers and we would have to
15 rely upon Surfwatch or Netnanny technology for foreign
16 speakers.
17 Q Well, then if we assume that some of the speakers are
18 foreign speakers and they're not tagging at all would your 90
19 percent go down considerably?
20 A The 90 percent -- I think you're fallacious here. The 90
21 percent is if I am a U. S. speaker how much can I depend will
22 actually get through to minors, that doesn't say anything
23 about how much minors -- how much potential sexually-explicit
24 material has come to a minor, that's a different question.
25 Q I understand that, I'm asking you a different question.
45
1 I'm asking you to assume all of the communication entities
2 downline change their browsers, change their end-user
3 software, change everything so that your L18 proposal could
4 technologically be implemented, if no foreign speaker tags or
5 labels their speech will that speech be kept away from 90
6 percent of minors in America?
7 A That speech, no, but I didn't remember that there's
8 anything in the CDA that involved those speakers.
9 Q Now, let's just talk about domestic speakers for a
10 minute. Your proposal assumes that domestic speakers, all
11 domestic speakers will tag, correct?
12 A Mm-hmm.
13 Q And that they will tag responsibly?
14 A Mm-hmm.
15 JUDGE SLOVITER: Was that a yes? I'm sorry.
16 THE WITNESS: Yes, I'm sorry.
17 BY MR. ENNIS:
18 Q I assume it's of course possible that there are some
19 speakers out there who will willfully violate the law and not
20 tag or tag inappropriately, correct?
21 A That's a good presumption.
22 Q And I assume there are a larger number of speakers out
23 there who will tag, but they will not exactly know what's
24 patently offensive or not and will make the wrong judgment
25 and say my speech is appropriate for minors when later it's
46
1 turned out some community thinks it's inappropriate, that's
2 possible too?
3 A That's possible.
4 Q All of that speech, whether willful violations of the Act
5 or inadvertent violations of the Act, that will reach minors
6 in America, correct?
7 A So will slander and fraud, yes, it's the same thing.
8 Q But it would not, would it, if you were using the PICs
9 technology set to the default to tag -- to reject all un-
10 tagged speech and to reject all tagged speech that has not
11 been approved by a third-party rater, none of that would
12 reach minors in America?
13 A That is correct.
14 Q Even foreign postings?
15 A That is correct.
16 Q You indicated you doubted that Disney would be willing or
17 happy to label pornographic sites as pornographic?
18 A That's not what I said.
19 Q Well, perhaps I misunderstood you. You don't doubt, do
20 you, that there are several, plenty of groups in America
21 today who would be happy to rate pornographic sites as
22 pornographic?
23 A Yes, I believe that to be true. What I did say, to
24 clarify though, was I doubt that many of them would
25 necessarily want to enter into contracts whereby purveyors of
47
1 pornographic material would do the rating in their behalf,
2 that was my testimony.
3 Q And I believe you concluded your redirect testimony by
4 saying that the ACLU or other speakers who wanted to set up
5 their own verification systems could set up their own
6 neighborhoods, their own systems, they could set up their own
7 Website, correct, do that?
8 A Yes.
9 Q And is it fair to say that creating a Website costs
10 anywhere between $1500 and $10,000, depending on how
11 elaborate you want to be or need to be?
12 A That is fair.
13 Q And is it fair to say that maintaining, operating a
14 Website costs anywhere from $20 a month to thousands of
15 dollars a month, depending on how much traffic you have and
16 what you want to do?
17 A That's fair.
18 Q And is it fair to say that if the ACLU or one of these
19 groups wanted to set up their own Website to do this
20 verification they would also need software to be involved in
21 the verification process?
22 A That is correct.
23 Q And that would cost something too?
24 A That is correct.
25 Q And that also needs people who would be involved in
48
1 managing the actual act of verification?
2 A They would have to have people involved in managing the
3 software, I believe that there is evidence that they could
4 automate the act of verification.
5 Q Now, suppose you are not the ACLU, you're an individual
6 speaker and you simply want to go home some night and you're
7 disturbed by some event, and you want to go on-line and say
8 something which you think is appropriate for adults, but
9 might be patently offensive for minors, it's not practical or
10 realistic to think you can on the spot go out and set up your
11 own Website and all of this mechanism to screen access, is
12 it?
13 A Nobody goes out and sets up their own Website -- or I
14 should say nobody but very few of the people, the speakers
15 we're talking about. They almost all of them, the vast
16 majority have some service where they go to. I presume that
17 such a person, if the ACLU were to set up such services, they
18 could go to the ACLU as a place to put their speech. So,
19 nobody actually sets up their own Website with the exception
20 of universities and such entities.
21 MR. ENNIS: No further questions.
22 JUDGE SLOVITER: Thank you.
23 JUDGE DALZELL: Mr. Hansen?
24 JUDGE SLOVITER: Mr. Hansen?
25 MR. HANSEN: I have no other cross.
49
1 (Pause.)
2 JUDGE SLOVITER: We'll take ten minutes before the
3 Court...
4 (Court in recess; 10:35 to 10:50 o'clock a.m.)
5 THE COURT CLERK: Court is now in session.
6 JUDGE SLOVITER: Thank you. Given the complexity of
7 this case and its quick wind-down and all of the technical
8 matters that we have to absorb and the different views of the
9 different witnesses, we needed a few minutes, although I'm
10 not sure I wouldn't have liked the weekend.
11 (Laughter.)
12 JUDGE SLOVITER: Judge Buckwalter will begin.
13 Thank you, Mr. Olsen.
14 JUDGE BUCKWALTER: Mr. Olsen, if the creator of the
15 material doesn't buy into your system, I think you testified
16 it really creates a big problem. Maybe they weren't your
17 words, but if the creator of the material doesn't buy into
18 your system, what did you say, it would make it very
19 difficult to --
20 THE WITNESS: Oh, yes. It's not so much a matter of
21 buying into my system but buying into the notion of if the
22 creator doesn't make an effort to electronically identify --
23 JUDGE BUCKWALTER: Okay, that's what I meant.
24 THE WITNESS: Then what happens is for parents to
25 protect their children, they have to hire somebody like
50
1 Surfwatch to go and hunt down all of this material on their
2 behalf and that's a problem because of the explosive growth
3 of the material.
4 JUDGE BUCKWALTER: Does it follow from that that
5 therefore the PIC system as proposed by the plaintiffs makes
6 more sense?
7 THE WITNESS: No.
8 JUDGE BUCKWALTER: Why not?
9 THE WITNESS: In the PIC system there are -- there
10 are multiple ways that the PIC system has been proposed,
11 there are multiple techniques in their proposal. The self-
12 labeling technique in their proposal is for the purpose of
13 this case the same as what I've described. It's a different
14 way of tagging but it still relies upon the content provided
15 to do the tagging. So to that extent we are -- we are on the
16 same wavelength there.
17 To the extent that they rely upon a label bureau,
18 that means the label bureau must look at all the sites, look
19 at all the documents, must hunt down all of the material that
20 might be offensive and that is a challenge. It's actually a
21 greater challenge than the judgment required if a content
22 provider just simply said this is the nature of my speech.
23 JUDGE BUCKWALTER: You also stated that -- that this
24 service and news groups and chat rooms, that there is no
25 technology for the speaker to ensure that only adults are
51
1 listening. I think you said you know of no possible way,
2 only statistically or something to that effect. I'm sure
3 I've butchered up what you said, but what did you mean by
4 that? If you could --
5 THE WITNESS: What I meant is that if the speaker in
6 any of those forums --
7 JUDGE BUCKWALTER: Right.
8 THE WITNESS: -- were to label L18 PICs, whatever,
9 PICs is a little hard in some of those forums, but if they
10 were to label them then to the extent that browsers were
11 screening, let's say, 80, 90 percent, then they have a 90
12 percent assurity that it isn't reaching minors.
13 That's the statistical, you know, how much can I
14 depend on. To believe that 100 percent of all browsers are
15 screening is naive, but if we presume the marketing claims of
16 Netscape and Microsoft to presume that in short order 90
17 percent are screening is a reasonable assumption.
18 I shouldn't, if I could clarify that, not are
19 screening but could screen. Nobody is screening today, with
20 the exception of Netscape, Surfwatch, Net Nanny, et cetera.
21 JUDGE BUCKWALTER: And all your declarations dealing
22 with determining which applicants are really adults have
23 nothing to do with the feasibility of that from an economic
24 standpoint, only that it's possible to in some way verify
25 that?
52
1 THE WITNESS: I only addressed is it technically
2 possible and is it technically difficult. Those issues I did
3 address. As to what the exact costs are, I suppose I could
4 find out but I'm not prepared to testify today.
5 JUDGE BUCKWALTER: Well, that's all right. It's
6 technically possible, is it technically difficult?
7 THE WITNESS: For which one?
8 JUDGE BUCKWALTER: For that adult verification, to
9 very whether or not the person is an adult?
10 THE WITNESS: Uhm --
11 JUDGE BUCKWALTER: By that I mean the user.
12 THE WITNESS: I guess -- I guess an easier thing to
13 do is sort of describe in layman's terms what the technology
14 is. What would have to happen there, there are two ways that
15 that could occur. Let's presume the first way in which the
16 person or the organization is providing the Webserver decides
17 to take this responsibility.
18 The technology would be when a person first came to
19 this site and the owner of the site doesn't know who they
20 are, the owner of the site could request a credit card, use
21 something like IC verified to make a charge on that card,
22 determine that it is a valid card, and then could issue them
23 a password. And there's lots of stuff in the declaration
24 about how user databases and passwords or no user end
25 database could be used.
53
1 But essentially they would issue an access code, if
2 you will. From that time on standard Webserver technology
3 would do the verification for you. It would ask for the
4 password whenever somebody came in.
5 So in terms of asking for passwords, this is not a
6 problem, standard software that does that. In terms of doing
7 the verification, you would have to write some CDI code that
8 actually goes out and calls IC verifier, whoever else you
9 decide is your provider of that.
10 I don't believe that's a difficult code to write,
11 but it would take some time.
12 Does that help clarify?
13 JUDGE BUCKWALTER: Yeah, that does to a certain
14 extent, it does. Thank you.
15 You may proceed, I don't have any other.
16 JUDGE DALZELL: Yeah, I have a few questions for you
17 of starting with chat rooms or news groups. Let's assume a
18 chat group is talking about the CDA and its students are
19 talking about the CDA, students varying in age from 13 to 18.
20 And in the course of the chat an 18 year old, exasperated by
21 his or her view of the law, types in "Fuck the CDA." Is it
22 your proposal that before he types in "Fuck the CDA" he
23 should tag that minus L18?
24 THE WITNESS: Yes.
25 JUDGE DALZELL: I beg your pardon?
54
1 THE WITNESS: Yes.
2 JUDGE DALZELL: Okay, so that is the -- so that
3 anybody even in that context must tag, that's your -- the way
4 it works?
5 THE WITNESS: If they want to identify -- if they
6 want to identify their speech that is one way they can do it,
7 yes.
8 JUDGE DALZELL: And protect themselves because,
9 after all, that is one of the seven dirty words.
10 THE WITNESS: Okay.
11 JUDGE DALZELL: Okay. We've had a lot of testimony
12 in this case about caching. Do you know what caching is?
13 THE WITNESS: Yes, I do.
14 JUDGE DALZELL: And since it's also agreed, I think
15 it's not in dispute, that upwards of 40 percent of the
16 sexually explicit content comes from offshore, at least we've
17 heard testimony about that. I don't think that's an issue
18 but very significant percentage is offshore, but that it can
19 get, it does get cached on this side.
20 Is it feasible in your judgment for the entity that
21 is doing the caching to tag one, let us say the people in
22 Amsterdam won't tag?
23 THE WITNESS: Okay. Let me -- if I could
24 characterize a legal presumption for you and I may be wrong
25 on the legal aspects of this, but I'm under the assumption,
55
1 which may be wrong, that the transmitters of the information
2 are not liable under this act. My presumption is that only
3 the creators or the people who served the information are
4 liable.
5 The caching you describe is a standard computer
6 science technique in communications and the caching, to
7 computer scientists, will be considered part of the
8 communication mechanism.
9 For example, it doesn't' work for telephones, but we
10 would consider the cache site as much like your phone switch
11 and whoever operates the phone switch isn't responsible for
12 what you say.
13 So, yes, it is being stored, but it's being stored
14 as an optimization of the communication process. Now, I may
15 be wrong in my legal assumption, but my legal assumption
16 would be that they're not -- the people who have caches are
17 not even considered under the law.
18 JUDGE DALZELL: Well, but I think you've assumed
19 away my question. What I'm trying to get at is the technical
20 feasibility of the cacher --
21 THE WITNESS: Mm-hmm.
22 JUDGE DALZELL: -- okay, the cacher doing the
23 tagging.
24 THE WITNESS: That would be very difficult, that
25 would be very difficult because generally the cacher has no
56
1 idea of what it is. All they've done is a stream of bits
2 came across that had a particular name.
3 JUDGE DALZELL: And what if the lawyers for the
4 cacher said you know, you may be wrong in your interpretation
5 of the statute, you may be deemed the re-publisher? What
6 then?
7 THE WITNESS: If I was them I'd hire a lawyer and
8 fight that, but --
9 JUDGE DALZELL: Well, okay, well --
10 THE WITNESS: -- it's a legal, it's a legal
11 presumption that, you know, if they did that, then obviously
12 I'm going to have to do something, I think.
13 JUDGE DALZELL: What would you do? That's what I'm
14 trying to get at.
15 THE WITNESS: If I was a cacher?
16 JUDGE DALZELL: Yes. Let's assume you had
17 definitive ruling that caching is a republication of the
18 improper content?
19 THE WITNESS: I'd turn the cacher off.
20 JUDGE DALZELL: Oh, you wouldn't tag?
21 THE WITNESS: No.
22 JUDGE DALZELL: Because?
23 THE WITNESS: Because it's too much bother. I'd
24 turn the cacher off, everything slows down and, I mean, this
25 is not a nice thing to have happen on the Internet but if --
57
1 JUDGE DALZELL: Well, why isn't it a nice thing to
2 have on the Internet?
3 THE WITNESS: Everything slows down.
4 JUDGE DALZELL: Anything that slows it down is not a
5 nice thing?
6 THE WITNESS: Not in my view.
7 JUDGE DALZELL: Okay. Well, I think a lot of
8 witnesses would agree with you on that.
9 All right, I want to talk about standards now. You
10 in your declaration speak warmly of -- I think the word you
11 used was the "nurturing" of the Government for the Internet,
12 okay?
13 THE WITNESS: Mm-hmm.
14 JUDGE DALZELL: Okay. Am I not correct though that
15 the Government did not establish a single standard that
16 creates the actual operation of the Internet as we use it
17 today?
18 THE WITNESS: Uhm, let me characterize what I think
19 you mean by or at least what I think I mean by establish a
20 standard. If you mean establish a standard to mean somebody
21 like the FCC created a regulation which everybody has to sort
22 of conform to, no.
23 JUDGE DALZELL: No, I mean IP-4, I mean IP-4.
24 THE WITNESS: If you mean -- well, if you mean, so,
25 no, there is no legislation that I'm aware of that says this
58
1 is what IP-4 is. On the other hand, the nurturing that I
2 talked about is that the creation of IP-4 and the money for
3 the researchers that did much of the early work, well, most
4 of it came from Government money.
5 JUDGE DALZELL: Well, that's to open that because
6 that was -- that's all the history because that was in the
7 defense business, right, the defense research.
8 THE WITNESS: Right.
9 JUDGE DALZELL: And that's all you were talking
10 about, right?
11 THE WITNESS: That's right.
12 JUDGE DALZELL: Because can you -- can you tell me,
13 can you identify for me any Government official who
14 participated in writing IP-4?
15 THE WITNESS: No, you'd have to ask Mr. Bradner,
16 he's more expert in that area.
17 JUDGE DALZELL: He couldn't think of anybody either.
18 But what I'm getting at is the standards therefore, of course
19 you say in your declaration that the Internet has to have
20 standards to operate, no one disagrees with that, the
21 question is whether it has governmental standards, that is to
22 say governmentally imposed standards. Wouldn't you agree
23 that's the real question here?
24 THE WITNESS: Not completely, not completely in the
25 sense that let's take for example the creation of the
59
1 Arpanet. In the creation of the Arpanet the military said we
2 need a way to communicate that is resilient to adverse things
3 like nuclear war, I think Mr. Bradner stated, so they
4 established a goal that says, you know, we will pay you money
5 if you will create for us a communication medium that has the
6 following attributes.
7 JUDGE DALZELL: Right, and they created the packet
8 switching.
9 THE WITNESS: Right. The CDA has established a goal
10 but has not dictated any technology standards nor have I
11 proposed they do so. They have only established a goal and
12 left it to the community to come up with technical solutions
13 that achieve that goal.
14 JUDGE DALZELL: I thought you've testified to us
15 that a governmentally imposed standard, let's assume either
16 we or the Supreme Court of the United States whose going to
17 review what we do says Dr. Olsen is right, we've had a
18 Damascus Road experience here, minus L18 is the answer. That
19 will then have the force of law, will it not? And so anybody
20 who doesn't tag with minus L18 doesn't have the safe harbor
21 that Mr. Coppolino says they have, isn't that right?
22 THE WITNESS: To the extent that the Supreme Court
23 said L18 is the answer, that would be true. To the extent
24 that the Supreme Court said that tagging is the answer, that
25 would not be true because we could use PICs, we could use
60
1 L18, we could use XXX, we could invent something else
2 provided it met the goal.
3 JUDGE DALZELL: My point is once the Supreme Court
4 or any court definitively holds that a method is a safe
5 harbor, it's then a legally imposed standard, is it not? Not
6 a scientifically imposed standard.
7 THE WITNESS: I would have to beg the question on
8 that because I just don't know enough about how the Supreme
9 Court enforces anything.
10 JUDGE DALZELL: Others enforce it; you're looking at
11 them.
12 (Laughter.)
13 THE WITNESS: I believe President Jackson objected,
14 but that was a long time ago.
15 JUDGE DALZELL: That's a long time ago.
16 (Laughter.)
17 JUDGE DALZELL: I think that's all I have.
18 JUDGE SLOVITER: Dr. Olsen, would you first satisfy
19 my curiosity? At the very beginning of today's questioning
20 you were asked, you mentioned that you built some software.
21 THE WITNESS: Yes.
22 JUDGE SLOVITER: What did -- I wanted to know what
23 the next question was -- what did your software do? What
24 kind of software were you talking about in that colloquy with
25 counsel?
61
1 THE WITNESS: Okay, they're a variety, there are
2 actually a variety of pieces of software that were in that
3 and we were focused on my work that has related to
4 interacting over the Internet.
5 Perhaps the first one I could characterize for you
6 is let us suppose that you and I were both chip designers,
7 microchip designers.
8 JUDGE SLOVITER: It's not likely from this
9 standpoint, but go ahead.
10 (Laughter.)
11 THE WITNESS: It's as good as some of the other
12 hypotheticals we've had, so let's suppose that we are and we,
13 you are here in Philadelphia and I am in Utah and we would
14 like--
15 JUDGE SLOVITER: More likely.
16 THE WITNESS: More likely. And we would want to
17 collaborate on this design. What we would like to do is we
18 would like to have you bring up the design on your machine
19 and I would bring it up on my machine, and when I made a
20 change, a message goes over the Internet and you see what I
21 just did. When you make a change, the reverse thing happens.
22 So we developed a protocol and the user interface
23 architecture to make that work.
24 JUDGE SLOVITER: I see.
25 THE WITNESS: That was one of the pieces. There are
62
1 variations of other kinds of --
2 JUDGE SLOVITER: Is that available now?
3 THE WITNESS: There are things that will do that.
4 They don't necessarily use our technology, it was a prototype
5 that hasn't been widely adopted yet, but there are other
6 software that does that similar idea.
7 JUDGE SLOVITER: All right. Let me go into the
8 final series of questions.
9 Do you think it is likely that there will be -- that
10 they will ultimately develop, if -- if the idea of tagging
11 catches on, is it likely that there would ultimately develop
12 two separate tagging systems, i.e., minus L18 and PICs as a
13 practical matter?
14 THE WITNESS: If the technical people were left to
15 themselves that would probably happen. I believe it will not
16 happen for the following reasons: to the extent that the
17 technical people start to argue with each other about and not
18 come to a consensus, they're far more afraid of the FCC than
19 they are of each other so they would probably come to a
20 consensus very quickly because they just don't want anybody
21 else messing in their part.
22 JUDGE SLOVITER: I'm afraid I don't understand what
23 the FCC has to do with this --
24 THE WITNESS: Oh --
25 JUDGE SLOVITER: -- nor do I understand -- yes.
63
1 THE WITNESS: Internet people very much don't like
2 outside people telling them what to do, so they would rather
3 come to a consensus than be told what they have to do.
4 JUDGE SLOVITER: And who would make this decision
5 then, would it be the market that would make the decision as
6 to whether you go with PICs or with minus L18?
7 THE WITNESS: There are a couple of ways that they
8 can do. There already is a large amount of market momentum
9 behind PICs. Once you get Netscape, Microsoft and Apple
10 committed, most everybody else becomes irrelevant. They
11 will, those three organizations are big enough that even if
12 one of them said this was the standard, it would push the
13 market almost immediately.
14 So let's say it's PICs. But another thing that
15 might happen is having read this transcript, Netscape might
16 decide well, hey, we could screen for minus L18, too, that's
17 easy, it only takes us 20 minutes to put the code in, and
18 they might just do -- put multiple screens in simply because
19 isn't hard and because they just don't want to quarrel about
20 the problem, they just do both and be happy.
21 JUDGE SLOVITER: So then minus L18 is more
22 theoretical and since you're -- I know that's a conclusion
23 but since you are the creator, I feel as though a fair
24 question to ask you, was created as a hypothetical, as a
25 technical possibility that shows what would be available
64
1 rather than as a proffer of something that could be put in
2 relatively promptly to comply with the statute, and that is a
3 question.
4 THE WITNESS: That is -- that is correct. I would
5 not characterize myself as having enormous influence on the
6 market, Mr. Gates would.
7 JUDGE DALZELL: This is your 15 minutes of fame.
8 (Laughter.)
9 JUDGE SLOVITER: No, I think he's going to have more
10 than 15 minutes.
11 So that under your proposal, you're suggestion is
12 that somebody -- and we'll get into the "somebody" in a few
13 minutes -- would label everything, every content that is
14 potentially within the statute, is that -- is that right?
15 THE WITNESS: Correct.
16 JUDGE SLOVITER: Falls within the statute.
17 THE WITNESS: I think you've sort of characterized
18 the conservative labeling strategy that someone might take.
19 JUDGE SLOVITER: Well, I thought that that was your
20 answer to the questions from Mr. Ennis mostly.
21 THE WITNESS: Right.
22 JUDGE SLOVITER: Okay. But that covers, does it
23 not, only material of concern to parents and otherwise
24 dealing with checks in some way and rather than with other
25 problems that might concern parents more?
65
1 THE WITNESS: Sure.
2 JUDGE SLOVITER: Such as with violence, for example,
3 is that right?
4 THE WITNESS: That is correct. The only reason I --
5 the only reason I narrowed what I responded to in that area
6 is I was responding specifically to the CDA.
7 JUDGE SLOVITER: Well, that's right, and that's what
8 your minus L18.
9 Now, in your proposal is there any way in which
10 there could be a distinction between the type of content,
11 like for example there are some parents who are -- would be
12 concerned about material that they consider sacrilegious, for
13 example, although it would not fall within the CDA, correct?
14 THE WITNESS: That's correct.
15 JUDGE SLOVITER: And some as I mentioned with
16 violence.
17 THE WITNESS: Yes.
18 JUDGE SLOVITER: And some might not be concerned at
19 all with sex because they think their kids just couldn't care
20 less, and having seen Mr. Coppolino's book, you know, they
21 might realize that they just see one picture and turn it off,
22 there would be no way under your proposal if it were to be
23 the successful winner in the marketplace, let us say, to
24 advise the parents or the adults of the type of material that
25 the labeler thought was minus L18, is that right?
66
1 THE WITNESS: Yes, that's correct.
2 JUDGE SLOVITER: At least as the proposal stands
3 now.
4 THE WITNESS: That is correct and there is a valid
5 statement which it seems to me is outside the discussion of
6 the CDA, but there is a valid interest on a lot of parents to
7 get information on more than just what CDA is concerned with,
8 as you characterized, and PICs is actually assigned to
9 address that larger -- that larger context.
10 To the extent that the content providers do the
11 labeling, I'm quite happy with PICs.
12 JUDGE SLOVITER: Now, well, what is it then about
13 PICs that you're not happy about? I mean why did you then,
14 since this idea of PICs has gone out into the sort of
15 literature, literature that you rather than I are more
16 familiar with, what was it then that inspired you to research
17 into an alternate system such as minus L18?
18 THE WITNESS: Two things. Two things, one of them
19 was assertions in declaration and in testimony by Mr. Bradner
20 that certain kinds of materials could not be tagged.
21 JUDGE SLOVITER: Was that -- I'd have to go back
22 because his testimony was sort of long, very long and very
23 technical.
24 THE WITNESS: Yes.
25 JUDGE SLOVITER: And very difficult to understand.
67
1 Was it technologically could not be tagged or -- I forget.
2 I'd have to --
3 JUDGE DALZELL: He said it was very difficult, as I
4 recall.
5 THE WITNESS: There are two -- there are two points
6 that he made.
7 JUDGE SLOVITER: Yeah.
8 THE WITNESS: One of them was that if I have data
9 which has some proscribed format like a JIF (ph.) image, I
10 believe we talked about on Friday, then putting the tag
11 inside the data, you couldn't do that. And he's right.
12 That's why, you know, I looked and said he's right, is there
13 another way? And another way would be to tag the name.
14 JUDGE SLOVITER: Hmm.
15 THE WITNESS: Another way also would be to use a
16 database on the side which is one of the architectures
17 proposed in PICs that contain the tag and that would work,
18 too.
19 JUDGE SLOVITER: Would the technical possibilities
20 that your declaration sets forth as to where the tag could be
21 put, I think you said three places, the server, et cetera.
22 THE WITNESS: That's where it could be filtered, not
23 where it could be put.
24 JUDGE SLOVITER: Filtered, would that work as well
25 for PICs as for minus L18?
68
1 THE WITNESS: Yes. One of the reasons I picked
2 minus L18 is because I knew I could configure the Netscope
3 proxy server and the Netscape Web server today to do that.
4 JUDGE SLOVITER: Mm-hmm.
5 THE WITNESS: But since Netscape is so excited about
6 PICs, I assume that they'll have a product out very soon that
7 would do exactly the same thing with relative to PICs labels.
8 JUDGE SLOVITER: So that that reason, at least, and
9 I don't want to put words in your mouth but I want to
10 question, so that reason for developing minus L18 could be
11 solved with PICs, could it not?
12 THE WITNESS: Yes, I guess --
13 JUDGE SLOVITER: You said there were two reasons
14 then for developing minus L18. What was the other?
15 THE WITNESS: The other one had related to a
16 different argument that Mr. Bradner made that was related to
17 the filtering we just talked about. I think he said
18 something about he knew of no software that could be
19 configured to do the filtering and again I put together a
20 counter example.
21 JUDGE SLOVITER: I see. So that with your computer
22 genius, and I mean that just as I say it, you know, true
23 computer genius, would the -- and with what you have shown
24 through your declaration and your testimony and your
25 experience in your work that will take place in Carnegie-
69
1 Mellon, would the problems that you saw with PICs be
2 technologically soluble?
3 THE WITNESS: My number one complaint with PICs has
4 to do with the concept of a label bureau, okay? And there, I
5 have two complaints there.
6 JUDGE SLOVITER: But that's -- excuse me -- that's
7 not a technical problem --
8 THE WITNESS: Well, no, it isn't.
9 JUDGE SLOVITER: Oh, go ahead.
10 THE WITNESS: Yeah, it is a technical problem and
11 there are two. One of them is the problems of building a
12 database that keeps track of an exponentially growing set of
13 things you're trying to monitor, that's my --
14 JUDGE SLOVITER: Oh, uh-huh.
15 THE WITNESS: The other problem is that the way PICs
16 would have -- the way a label bureau would be implemented, it
17 actually does slow down the way a user would get access to
18 their material.
19 And if I could, without getting into acronyms --
20 JUDGE SLOVITER: Yes, but you're going to.
21 JUDGE DALZELL: Oh, go ahead, everybody else has.
22 (Laughter.)
23 THE WITNESS: I know, I read their testimony. I've
24 tried very hard to stay away from them.
25 JUDGE SLOVITER: And the Court's impatience.
70
1 (Laughter.)
2 THE WITNESS: If I'm sitting at the browser much as
3 Mr. Schmidt showed us on Friday and I click on --
4 JUDGE SLOVITER: You mean Mr. -- when Mr. Schmidt
5 showed us this --
6 THE WITNESS: Right, when he led you through the
7 Worldwide Web.
8 JUDGE SLOVITER: On the -- on the screen?
9 THE WITNESS: Yes.
10 JUDGE SLOVITER: Yeah.
11 THE WITNESS: Yes, that whole demo. When I click on
12 something, what has to happen right now is that name comes
13 into the software, we identify from that name something
14 called the domain which effectively identifies the computer I
15 want to talk to.
16 I open a connection to that computer like making a
17 phone call and I sent it the rest of my request. Their
18 server software will then try to fill that request and will
19 send me the information back. So I've got an over and a
20 back.
21 If the messages are small, my dominant cost is the
22 amount of time it goes over and back, and that really is a
23 problem. Now, with the label bureau, if I want to have a
24 label bureau interposed, then the following thing has to
25 occur: my software gets the URL and before it goes and
71
1 contacts with the information, it first has to go over and
2 back to the label bureau saying is this clear and give me the
3 label.
4 JUDGE SLOVITER: Mm-hmm.
5 THE WITNESS: Then I decide if I like the label and
6 then I go again over and back to actually get the
7 information.
8 JUDGE SLOVITER: Is the "I" in this hypothetical the
9 parent?
10 THE WITNESS: The parent, the software that the
11 parent has installed, the software is actually doing this.
12 JUDGE SLOVITER: So, in other words, it's going to
13 take a little more time. So that if Judge Dalzell, who is
14 the only one among us who has any minor children at home, so
15 if Judge Dalzell were to go on the Internet, would it slow
16 down his or his children's receipt of the information? Is
17 that what -- I'm trying to find out --
18 THE WITNESS: Yes, yes.
19 JUDGE SLOVITER: Okay. And --
20 THE WITNESS: What it means that instead of an over
21 and back, I now have two of those and those are the most
22 expensive part of the transaction.
23 It's also complicated by the fact that if this is a
24 popular label bureau, lots of people are trying to do an over
25 and back to that label bureau. In the PIC spec this is
72
1 termed a hot spot because that potentially becomes a
2 bottleneck.
3 They have proposed some solutions to that which
4 essentially allow me to have multiple label bureaus, but they
5 haven't tested them yet. I --
6 JUDGE SLOVITER: Well, a lot of this all hasn't been
7 testing --
8 THE WITNESS: I'm actually, I mean in a technical
9 judgment I think the scheme they have proposed to alleviate
10 that hot spot problem is probably reasonable and I would not
11 contest them on that.
12 But we still have this over and back, over and back
13 twice as much for every access.
14 JUDGE SLOVITER: So it adds cost or it adds an extra
15 minute to Judge Dalzell's children when they want to reach
16 the material? I'm just trying to find out what we're talking
17 about in practicality.
18 THE WITNESS: If it inhibits the flow, if I'm
19 allowed to say that?
20 JUDGE SLOVITER: Yeah.
21 JUDGE DALZELL: Everybody else did.
22 THE WITNESS: Everybody else did. I mean it
23 actually makes everything much more sluggish than it would
24 normally be.
25 JUDGE SLOVITER: But by sluggish in computer
73
1 language we mean that they get it in a minute or two later
2 than they would have otherwise had to get it?
3 THE WITNESS: Well, if you ever sat down and used
4 the Worldwide Web.
5 JUDGE SLOVITER: Oh, I tried.
6 THE WITNESS: Yes, well, if you ever used it, I mean
7 a minute is absolutely unreasonable, I mean just -- this is a
8 getting into the user interfa