The Proceedings of the 1989 NanoCon Northwest regional
nanotechnology conference, with K. Eric Drexler as Guest of Honor.
Return to NanoCon
Top
Return to NanoCon
Proceedings Table of Contents
Return to Jim's Molecular
Nanotechology Home Page
NANOCON PROCEEDINGS page 7
IV. HYPERTEXT PUBLISHING (continued)
D. Hypermedia Panel Discussion
Introduction by Mike Thomas
So far we have talked about the need for hypertext to support
nanotechnology; we have shown you some of the new tools coming down the
road; we have given you an example of a current application. We'd like to
take this next session and talk about some of the issues involved in building
systems to critically discuss advanced technologies. What are the problems
that have to be solved for hypertext publishing to effectively support nanotechnology?
Before we get started, we're adding a couple of additional people. Greg
Bear, author of Blood Music and other books is going to join us.
I'm Mike Thomas, also of Boeing Computer Services. I've been working with
hypertext and knowledge-based systems for quite a while.
Eric's book talks extensively about using hypertext to start a "grand
conversation", to establish a medium for critical analysis, to describe
the relationships of the various things that will have to happen to bring
nanotechnology to bear.
I have some questions here to provide ammunition for discussion.
- What is hypertext publishing? Some of that should be clear at this
point.
- What types of knowledge are we going to need to support these activities,
and how will it be selected and evolve?
- How to navigate through such complex systems. People tend to get lost.
- What about the filtering problem? How do you get the level of detail
that you want to see?
- How to provide context-sensitive links so that a manager and an engineer
could get information appropriate to their different needs.
- There should be triggers. I would want to know if someone placed a
comment about one of my works.
- What are the techniques that could be used and what are their costs
for publishing, data storage, data transmission.
- Controlling configuration and security access.
- Scale-up: a VERY big problem.
- Although the technology is available, making one of these systems
work for a particular application, as Louis Roberts did, can take several
months to several years.
- Representing knowledge in these systems: unless you put in menus,
nodes, and navigation tools, it is not clear to the user where he is going.
- If one is going to build a system around Engines
of Creation, as as we have started to do here in Seattle with
the PATH project, who is going to do all the work involved?
These are the issues. I will ask each of the panel members to state their
positions and then we will open a dialog.
1. Eric Drexler:
One thought that occurs to me that addresses some of the points that you
raised concerns who will do all of the intellectual work that will be necessary
to convert a blank system into one with a lot of content and structure,
and with guides that help you decide which of the items that you could go
to next are likely to be of interest to you. A key point is that
in the present world an immense amount of intellectual work is wasted.
When someone reads something, he understands it at some level and makes
some judgement, perhaps finding an error. There is no way for those judgements
and insights to be captured for the benefit of later readers. If you have
a system that facilitates someone making a small comment or simply a rating
("This was worth my time."), then some fraction of that intellectual
effort can be saved in a form that is useful to later readers.
2. Louis Roberts:
Our primary involvement with hypertext has been with browsing systems, which
are an easier problem than systems for criticism. The problem
of the reader always will be easier than the problem
of the author. The reader has only to gain the information that
he needs, while the author has to create and organize the information. In
the case of hypertext, organizing the information is especially difficult
in that the entry points have to be made sufficiently obvious that the user
is confident that he is seeing everything that he needs to see. We're not
doing that!
3. Greg Bear:
What Eric said frightens me. I am not so egalitarian that I would like to
have the reactions of everyone who has read one of my books to be attached
to the book! One advantage of the society that we have is that it takes
some considerable effort to have your words immortalized. This tends to
make you more careful of what you say. The legal implications of a society
where everything is written down and can be commented on by whoever reads
it, are interesting. I think you may want to have a formal screen. Practically
speaking, this has to happen. If the organizer of the Ku Klux Klan makes
a comment on an article in the Journal of the American Medical Association,
how do you screen that out? Do you have editors screening what is entered
on the hypertext panels? What sort of status do you have to have to make
a comment? Do you need a panel to screen out libel. If you've been a science
fiction fan long enough, you know what it's like to have everybody in the
world commenting on your work endlessly, without screening. Much as I like
pluralism, there will have to be a screening process in this orgy of information.
Maybe this screening could be done through licensing. We asked how to pay
back contributors. You've all heard of ASCAP. Basically a radio station
pays so much to ASCAP and there is a recording made of each song, and that
is broken down and they pay a royalty to each artist based on the license
fee. Something like this but even more complicated would have to be worked
out for a huge hypermedia library system.
Will anyone be able to publish on the system? That would be very idealistic,
but then you're going to clog up the system with a lot of pretty bad stuff.
Are you going to have publishers and editors coming into the system and
frustrating all of us writers? Suppose an Ayatollah-type becomes a publisher
and says, "If you try to publish your book, I'll kill you." We
are talking about a world-wide system of information here. The only way
to make money and be effective is to be inclusive. I'll give you my personal
view. I am an information junky! I don't care what it's about, I want to
at least look through it. I've got from the Smithsonian Air & Space
Museum three laser disks containing the entire photographic history of the
Space Age. 300,000 photographs, including a picture of Arthur C. Clark when
he was only 25 years old. It's not yet in an easily digested form, but it's
there, like the Encyclopaedia Britannica without a Mac attached
to it. This demand is going to push for more inclusiveness; all information
available through one terminal. Then how are you going to protect it? You
will have proprietary information. How can you put on blocks; how will hackers
get through those blocks?
4. Mark Stiegler:
I wanted to talk about two related questions. One is "How to you handle
scaling?" I have an embarrassingly simple rhetorical answer - "Of
course you build it on top of Xanadu, because scaling is what Xanadu is
all about." The other is how to deal with having potentially millions
of people writing into a shared information pool. How you tell the good
from the bad is in some ways complicated and in others fairly simple. I
predict that 5 years from now there will be a healthy,
competitive market in filtering systems. Different companies specializing
in artificial intelligence and pattern recognition will be building these
high powered tools to be able to filter for you based on your custom preferences.
One of my predictions for the year 2000 is that there will be a great scandal.
The leading filtering program in that market place, it will be discovered,
not only filters information for you, but it also filters out and discards
anything that is a positive review about other filtering programs. I would
go on to predict a great hue and cry for government regulation of the filtering
industry because of this. But fortunately because of the links in the hypermedia
information pool showing the consequences of government regulation in high
technology areas, a very large lobby will be formed to prevent the government
from regulating the industry, but meanwhile the company that produced the
disreputable filtering system will go out of business.
Forgetting about the complicated filters for the moment, how does filtering
work today? Every time that I go into a book store, I am momentarily overcome
by horror at all the information there that I don't have time to read. I
walk past the sections on cooking, racing cars, and sports, because they
aren't interesting to me, and stop at the science fiction section. Inside
the science fiction section there are books by many people. There are certain
people whose books are invisible to me because I know that I hate what they
write. There are others that reach out and grab my eyeballs because they
were written by people who blow my head apart. So we use two very simple
filters: (1) by type of material, (2) by author - a very powerful reputation-based
filtering system. Starting with these two will get you through a lot of
bad years waiting for the sophisticated filtering systems.
As far as the problem of lots of schlock out there, editors and publishers
do a lot of filtering, but still there is lots of schlock out there. Bookstores
contain books by Jeremy Rifkin. The answer to this problem is (as long as
the hardware and software is capable of growing and handling enough material
to include both good and bad works) to let the schlock writers put their
material into the system, and, if no one reads it, they will have to pay
to store it there, and eventually they will get the idea.
5. Mike Thomas:
Eric proposed in Engines
of Creation a novel solution to the filtering problem: perhaps
we put a voting system on these links. We attach little places to the links
where you can say "I've been here and liked what I saw. I tip the author
a nickel or a dime." This voting scheme would allow the work to float
up and down through the knowledge system, and if it broke through a threshold
of popularity, it would catch the attention of individuals who have set
their filters for popular works. People who didn't have a lot of time could
set up criteria that they would like to see information meet before they
read it.
As a practitioner of these technologies, who has been trying to implement
such schemes, I would have to say that a lot of these mechanisms are not
in place. We have overwhelming amounts of knowledge available to us and
these techniques for dealing with it are valuable, but not timely. We need
to automate the links, and have algorithms that can dissect a piece of information
and make appropriate links and talk to other algorithms for us. I don't
see such things coming in the near term. For us to use these technologies
effectively, we need smarter links. I would like to challenge the manufacturers
of these technologies to take a look at some of these issues.
6. Discussion
L. ROBERTS: A filtering system by its very nature is a dangerous beast.
It will never show you anything that doesn't fit. Voting systems are very
dangerous beasts too, depending on what are the criteria for voting. Once
upon a time some people voted in very effective ways that the sun moved
around the earth. That view held for a very long time, but it didn't change
physical reality one bit. Earth still moves around the sun. Filtering systems
are something to be real careful about.
E. ELLIOTT: First, the problem you have with the voting system is, how many
people are going to vote for astrology. In a democracy, astrology is fact
and large areas of physics are false.
Second, the Ayatollah poses a brute force problem, but Jeremy Rifkin is
a far subtler problem, which is the spread of memes that have to be overcome.
Rifkin will use certain philosophical groups, like New Age groups to get
his concepts across, but he will also attempt [to make] policy through lawsuits
and lobbying Congress. How do you prevent the Jeremy Rifkins from setting
up a maxi-filter for every thing that this whole conference is about?
L. ROBERTS: Isn't this a case where universal enfranchisement doesn't seem
to work very well. Just because you have a vote doesn't mean that you can
use it very effectively.
G. BEAR: If your vote has power. We're talking about a system where the
vote is simply recorded, NOT a system where people at the lower end have
the power to remove something from the system. THAT you can never allow
in any world library system. I wish there were some way to remove Jeremy
Rifkin; there isn't - too bad.
Actually, it's the bullet you don't hear that gets you. Jeremy Rifkin
is not a dangerous man; he's a noisy man.
The question that I have is "What is the criterion for removing information
from the system?" Are you postulating a system that is so huge that
throughout all eternity the companies running the system aren't going to
feel the need to get the dross out and clear up space for more valuable
material. Five million people voted against this book; they all wanted their
money back. It didn't sell very well. The book was Moby Dick by Herman
Melville. Such a loss!
E. DREXLER: Two words. Market forces. So long as someone pays for the storage,
it stays on. If no one does, it goes away.
G. BEAR: What happens then. Do they keep them on hard copy? There are books
in the Library of Congress that haven't been checked out for 200 years.
Yet they're there. Suddenly someone wants to look up the registry for a
small New England town to find out whether or not the small pox epidemic
in that town could connect to a recent outbreak of small pox in opening
up grave yards 200 years later.
E. DREXLER: I think that a historical society would pay for it.
G. BEAR: I wouldn't count on it.
M. STIEGLER: Disk space is getting cheaper, and in particular, tape archiving
is pretty cheap.
G. BEAR: Libraries are not paid for by the grace of market forces. They
are paid for by enforced billing of citizens.
AUDIENCE: There are books 500 years old stored under the stairs of libraries
400 years ago because no one read them in 100 years. You're saying, when
you talk about voting, that what we think today is going to be appropriate
500 years down the line.
E. DREXLER: I think that there is a straw man here: a system in which the
majority somehow decides what is on the system and what is not on the system.
Keeping things on disk is dirt cheap. All it takes is one
person to keep something around, to say that we'll keep this old data around
because it costs us essentially nothing to keep it around. I think that
in practice that is how it will happen, and if you want to get some government
support for this to happen, that is fine. This will be cheaper than keeping
books.
M. STIEGLER: There are some interesting complexities on how a market system
could work. Since this is all computer-mediated, you can make the costing
mechanism work in a number of ways. As far as these historical societies
are concerned, what is their pay-off other than their own deep belief in
the value of the material? The guy who picks up the tab on a couple of gigabytes
of material will be able to sell access to it to anyone who wants to look
at it 10 years later. There will be a pluralistic society of charging and
pay back mechanisms as well.
AUDIENCE: I've always thought of hypertext as a passive database, but once
you have voting, you almost have the same sort of thing as in a neural net,
where you're emphasizing some paths and de-emphasizing others. It wouldn't
be difficult to have some sort of software that "activates" the
hypertext net and make it more of an active "brain thing."
G. BEAR: A variation of what you are talking about would be a society of
people whose function would be to look for worthwhile, but hitherto ignored
material, and if these hadn't been advertised by the library for some time,
the people would be paid as if they had created these things, but they would
not have control over the materials. Dover Books is an example -- they are
all public domain.
J. HAWKINS: I believe there is another straw man here. We already have market
forces, editors, and books that just rot in the library. The real issue
is not whether there will be a filtering system on a hypermedia knowledge
base, but rather, what will be the differences between this filtering system
and present filtering systems, and what will be the consequences of those
differences.
D. GAGLIANO: Filtering and voting are useful only if they add value. One
of the key features of a system like that is that it stops you from having
to read things that you might not be interested in. I suspect that there
will be people who make a good living by finding the information that has
been left by the wayside.
G. BEAR: They're called critics.
R. BURTON: Optical tape exists today and in a few years there will be systems
to store terabytes on a 12 inch reel. You don't gain more memory by taking
something off because you throw out the entire tape.
AUDIENCE: A Terabyte is about a mega-book. Not very much! How many people
are going to write a million books?
G. BEAR: Three years ago I got a 5 megabyte hard disk for my computer, and
I thought, "Gee, I'll never fill this up." Now I have 40M and
I say "I only have 6M left; I'd better throw something out." If
I had a terabyte, I'd want to down load the entire Encyclopaedia Britannica
and attach hypertext to it.
R. BURTON: It doesn't seem that people realize how big a terabyte is.
M. STIEGLER: Some of us are aware of how small a terabyte is.
G. FJERMEDAL: What Rick is saying is that the storage technology will keep
up with us so that there will be no need to flush out old books.
G. BEAR: But say that you have a search engine that can search a terabyte
of information per minute. A hundred years from now, you have a mega-terabyte
and you put out a general search order...
E. DREXLER: With a trillion processors per cubic centimeter...
G. BEAR: My worry is that our hunger for information will equal the speed
of advance of the technology.
E. DREXLER: If you have the space, you will use it. Empty space isn't worth
anything. You might as well put something in there. On the other hand, if
your yard stick is "Can we store what people have written?", then
if the cost of storage is less than the cost of people writing at the minimum
wage, then in some sense you can afford to store whatever people write.
But if you start digitizing all of the world's video, then you fill up the
terabyte.
G. BEAR: Then you send out the nanomachines to record all bacterial DNA
sequences to trace the evolution of these forms around the world. They send
back their information, and you have tera-terabytes.
AUDIENCE: If links to [a piece of work] are not hit within some time window,
then you store it on tape. Then you could pay someone to drag it out.
AUDIENCE: How are we going to convince the masses that information and knowledge
have value.
M. STIEGLER: Make it fun! A hypermedia information pool is a lot more enjoyable
and convenient to use [than conventional methods of storing information].
One example concerns the debates that have raged over the centuries on the
merits of Marxism vs. free market systems. Think about how much fun it would
be to interact in a hypermedia shopping plaza with other people in the plaza
in a simulation where, on the first day the plaza is a Marxist system and
you get to learn all about what it's like to live in a Marxist country.
On the following day you learn the free enterprise system on the same computer.
It's going to be a much more powerful learning environment, and the fun
will draw a lot more people in.
V. McINTYRE: An issue that we have skirted is the issue of privacy.
AUDIENCE: It's great to talk about putting these masses of information into
a hypermedia system, but how are you going to do it? Are you going to set
up in the system an AI expert system to create these hypertext links as
the material comes in, or is someone actually going to have to put these
buttons on each interesting word?
G. BEAR: Engineers are such spoil sports.
L. ROBERTS: That's the difference between the Guide® engine and the
HyperCard® engine. Guide doesn't have an underlying programmatic tool
so you have to not only pick up all the beads, but lace them together yourself.
Engines with programmatic tools let you automate that process.
M. STIEGLER: I predict that there will be a market not only for
different filter systems, but also for automatic linkers.
Part of it is making it very easy for a human to install a link himself
when he adds a new piece of information.
G. BEAR: Let's address Vonda's issue for a moment. A very important issue
is that when we talk about all the capabilities, we are ignoring the legal
and ethical considerations that come with the technology. Eric did a very
good job in his book of considering many of these things, but there are
some that we are just not going to recognize. I wrote an article many years
ago entitled "1984", in which I talk about the fact that a government
that is surrounded by millions of people with computer systems in their
homes is a government that cannot keep anything very secret for long. The
knife cuts both ways. You're going to have the problem of government intrusion
into your privacy, and you're going to have the problem of your intrusion
into the government's privacy. You have millions of hackers out there, and
they gather up all the information from published sources, and build an
atom bomb. The government says "You can't do that!" They say "Then
you get out of my file! I'll trade you."
AUDIENCE: We are talking about something that will be formed from the center
out, like Washington, DC, was designed. What I suspect is that instead,
due to dollar and man-month considerations, it will happen like Tokyo did.
A variety of businesses will put this together, as Xanadu has done, and
the structure will be defined by the business environment that it grows
in. It will gradually be hooked together element by element, with various
committees desperately trying to set up standards. Government will be about
three years behind, as usual.
AUDIENCE: One other addition to Vonda's comment, what it will basically
come down to, regardless of what rules you have, is that the more you participate,
the less privacy you're going to have. I don't think that there is any way
around that.
G. BEAR: I'm not sure. Big Brother participates in everybody's business;
nobody participates in his. Someone might find away around the hackers intruding
on the government's privacy.
AUDIENCE: Big Brother isn't a monolithic whole so that people could peer
into what the various bureaucrats...
G. BEAR: That may be the saving grace. However, if instead of human lackeys,
they are totally dedicated machines, that don't leak to the media or have
alcohol problems...
AUDIENCE: The first hypermedia use that I can think of is Winston Smith
[in "1984"] who modifies records for Big Brother. If this system
is going to be centralized...
G. BEAR: It must be holographic; otherwise it won't work because eventually
someone would crash the whole thing. It won't be like the central library
in "The City and the Stars"; like Eric tells us, your wallet will
contain the entire knowledge of human kind.
E. DREXLER: Decentralization, redundancy, and digital signatures should
go a long way toward making sure that information stays around and is not
corrupted.
G. BEAR: Perhaps we should start getting laws that mandate no centralized
information storage system that is not accessible to private citizens.
B. WEBB: We discussed a lot of issues. The key is plurality and market forces.
Many different things will be done in each area; the system will somehow
find its way regardless of what we do by a Darwinian selection process.
What I haven't heard being discussed, and which interests me enormously,
is beyond the issues of the mechanics of the system that we've been discussing.
What impact will the existence 5 to 10 years from now of a hypertext system
of the type that we have been discussing have on the variety of human institutions:
the education system, the political process.
G. BEAR: Teachers will be among the most highly paid entertainers in the
world. You will have home schooling. You will have mandatory socialization
programs where you take your children out of the house once a week and put
them together so they can talk. Otherwise they'll be glued to the computer.
M. STIEGLER: When I was talking about the Bush-Dukakis debate, someone commented
that "They'll never allow it to happen." I think that eventually
they'll let it happen, but the government will be one of the last organizations
that will be driven to use a hypermedia information pool. The organizations
to adopt it most quickly will be those working in a competitive market place,
places like Boeing.
G. BEAR: An exception within government will be the IRS. Only the agencies
that have a mandate to get the job done, otherwise government falls, will
be using the system. The Congress could plug around for years and things
don't fall apart, but the IRS and CIA need to do their jobs as efficiently
as possible.
AUDIENCE: Who pays for what and who receives the benefits? Everyone inputs,
but the people who hold the copyrights receive the benefits. Do only a few
get the rewards from a large collaborative enterprise?
G. BEAR: Copyright is not a guarantee of earnings. A magazine can copyright
my story but, by contract, can not sell the story elsewhere without paying
me.
M. STIEGLER: The Xanadu position on this is that we will support the ability
to create "inclusions", wherein you include in your document material
from someone else's document. Thus as the user reads your stuff, he is lead
to the other writer's stuff, and the royalties are appropriately divided.
AUDIENCE: From computer science we know that it is not only the information,
but the indices to the information that takes up the volume. Arthur Clark
wrote a famous short story where in the far future the library for the entire
galaxy occupied a star system and unfortunately they lost the actual information,
which, thanks to advanced information compression technology, was smaller
than a grain of sand.
AUDIENCE: The value of redundancy. Always back up.
M. STIEGLER: I don't think the indexing problem will be that great with
Xanadu. I am confident that no one knows today what percentage of a hypermedia
information pool is going to be dedicated to linking information. We are
focusing on making links very light weight. How many light weight links
does it take to equal the amount of data that you're using to store a 10-minute
video clip? It's not clear to me what the ratios will be.
G. BEAR: We've been talking about text search engines because our technology
doesn't permit us to handle thinks like visual or music search engines.
We don't want to have to describe a picture in text to search for it.
L. ROBERTS: That's exactly the problem we had. It took us four man-months
to build the basic system, but now someone else will face a two man-year
effort filling the systems with images and index material. There will likely
be misplaced and/or mis-labeled material.
G. FJERMEDAL: I'd like to recommend two science-fiction books. One is Marc
Stiegler's David's Sling, which classically illustrates how hypertext
would be used. I know that a lot of software engineers at Apple consider
it a bible. The other is Greg Bear's Eon, in which a library is discovered
inside a hollow asteroid. This library is the ultimate knowledge navigation
tool. You sit in a chair and knowledge flows into your mind, and you follow
links by thinking.
S. SALKOVICS: I see two competing benefits in the design of hypertext architecture:
knowledge processing and link adaptability. Can you see the time when the
basic hypertext architecture will stop evolving?
G. BEAR: No.
AUDIENCE: To get a taste of what links can do, look at the videos of James
Burke's "Connections", the PBS series. That is an excellent combination
of entertainment and knowledge and an indication of where hypertext will
lead us.
E. DREXLER: That reminds me that the people who produced and directed the
"Cosmos" and "The Ascent of Man" series are actively
pursuing a one hour show on nanotechnology.
E. ELLIOTT: A lot of problems are going to be solved by having multiple,
competing hypertext systems, rather than a centralized system that could
be vulnerable to information loss.
G. BEAR: Here's a frightening thought. What if the government decides to
portion these systems out like cable networks to get things started?
M. STIEGLER: I think that hypertext will be very good for free enterprise.
I hope that is discussed tomorrow. But there is a very real danger here
for freedom of speech. Governments inherently do not like freedom of speech.
Every time the medium is changed, the government will grasp at the technical
uniqueness of the new medium to try to limit freedom of speech. Perhaps
this business with the Ayatollah will serve to remind people how precious
is freedom of speech.
J. CRAMER: I have a comment and a question. The comment is a very simple
model for what happens when the density of links gets larger than the database
itself: a checker-board. Place all the things that contain references along
one axis and everything that might be referenced along the other axis. The
number of references thus goes up with the square of the number of entries.
If the link is only 10-4 the size of the entries, then as soon
as you get 104 entries, the link volume becomes the same size
as the entry base, and as the entry-base gets larger than that number, the
size of the link-base grows geometrically larger than the entry-base. Thus
you will have a really serious problem.
The question is, will hypertext systems allow contributions from people
who may not wish to be identified? For example, so that government employees
could leak information to the public. On the other hand, there is the problem
of "poison-pen" letters.
G. FJERMEDAL: Everything that you ever said could be held against you. You
could be hounded years later for taking the communist or free-market side
in Marc Stiegler's hypothetical simulation of the free-market/communist
society debate.
E. DREXLER: Addressing John Cramer's comment: you can argue that a given
document can only reference a limited number of other entries, so that will
put an upper bound on the number of entries. That puts an upper bound on
the average number of links per entry in the system. This issue really needs
an item-centered rather than a matrix-centered perspective.
I would like to live in a world in which it is possible to play a couple
of different roles under different names. I need to play a very conservative
role in making technical arguments; however, there are other areas in which
there would be very interesting ideas to toss out, but which would be inconsistent
with the body of material that I am putting together on nanotechnology under
the name "K. Eric Drexler." That also applies to government whistle-blowers,
etc. This raises questions about people who publish libelous material. One
possible answer is that pseudonyms are allowed unless there is a court order
to strip the veil.
G. BEAR: I don't think that would work. If you're going to squeal on the
government, you would probably want to do it outside the hypermedia system.
AUDIENCE: With all the inter-connections in hypermedia, aren't you really
talking about a neural net, in which those connections that aren't used
are eventually lost? Suppose I read and make connections that nobody else
has that might be useful to someone. It would be whether something is used...
G. BEAR: I have this image of a connection-making organism within the hypermedia.
The links exist temporarily while it extracts the information; it then dissolves
the links and takes the information back to its master. In a very sophisticated
system in the future, you may not need any indexing because of these organism-systems.
In Eon I describe one based on a mouse psychology - it enjoys searching
for facts for its master.
M. THOMAS: I'd like now to focus on how hypertext might help the nanotechnology
effort.
E. DREXLER: In diverse interdisciplinary areas, where you're trying to pull
together and evaluate information from various sources, and then have others
evaluate your results and arguments, hypertext capabilities will be of amazing
value.
M. STIEGLER: One of the earliest groups to adopt hypertext systems will
probably be university researchers. Twenty workers scattered across the
country in a particular sub-specialty, in which journals take a year to
publish results, will be looking for ways to quickly and effectively make
their points and counter-points. Numerous of these small groups may want
to link together, especially if they are all using Xanadu.
E. DREXLER: That particular picture answers some of the concerns as to how
we are ever going to get this vast system started, and get everyone to use
it. The answer is that if 20 of the right people use it, you're
ahead. You will have a small information pool that is worth other
people becoming involved in.
G. BEAR: The early stages of this with CD-ROM have already affected my life.
I've got Microsoft Bookshelf right by my computer. I look up a lot more
words.
AUDIENCE: In David's Sling there is a very good discussion of a computer-augmented,
hypertext-assisted debate. Mark makes concrete what Eric talks about in
the abstract in Engines of Creation.
M. STIEGLER: I wrote David's Sling before I ever encountered the
term "hypertext" much less Eric Drexler, and so I don't use that
term...
G. BEAR: I don't say "nanotechnology" in Blood Music either...
M. STIEGLER: The "decision duel" to which he is referring is obsolete.
I have much better ideas now after Xanadu.
AUDIENCE: The concern has been expressed that university researchers are
often releasing their results at press conferences rather than through the
standard peer-review process. With hypertext and with computer access for
laymen who might be interested in science, what about the danger of science
becoming more of a entertainment business, even one where results are "right"
because they are popular.
E. DREXLER: Somewhat facetiously, in the future, there may be complaints
because some scientists are publishing in journals first, rather than putting
their results through the hypertext review process. Seriously, I believe,
that even with an entertainment component included, there will be a serious
scientific literature that is of higher quality than what we see in the
journals today.
G. BEAR: Should there be any rules about hate literature, deliberate lies
and distortions?
AUDIENCE: I would like to comment on voting and filtering systems. I expect
that all of that will become obsolete as soon as we have the first generation
of kids that are brought up on hypertext systems. An analogy would be going
into a book sale and having what you are interested in jump out at you from
a huge line of books, seemingly without any great attention on your part.
L. ROBERTS: A hypertext system should allow you to find something that you
didn't even know existed.
G. BEAR: You must have a random browse capability so that you can have access
to the system without any filters, even your own.
G. FJERMEDAL: Picture a kid going through knowledge the way they get glued
to a video pin-ball game!
AUDIENCE: The next generation will be more interested in video than in text.
G. BEAR: No. You're talking only about the "lumpenproletariat."
The "movers and shakers" are all readers. Eventually something
will replace text. I try to hint at that in Eon.
M. STIEGLER: One of the advantages that text has today is that it is cheap.
Part of the answer is that we will make it much easier to do other media
so that you won't need $20M to put together a film.
E. DREXLER: I would argue that text will last as long as speech.
G. BEAR: There will be something that we can barely imagine right now that
will incorporate all the best features of speech, text, video, and it will
be a fluid medium so that you can instantly project an image of what you
mean instead of having to describe it. Think of a system that connects with
your occipital lobe and projects that as part of a textual communication.
AUDIENCE: If personal interest can be served by someone accessing my stuff,
I will be interested in promoting that. Methodologies like advertising may
determine a vast amount of the traffic here.
AUDIENCE: A lot if this is oriented from the author's viewpoint presenting
his data. What about an active reading assistant that, for example, would
realize that it's 2 AM, I'm very tired, so it modifies the flow of information
appropriately, repeating things as necessary, etc.
AUDIENCE: I read awhile ago that at some threshold of information volume,
it becomes cheaper to extract the structure of the information being discussed
and store that than to store the volume itself. Thus an automatic reading
assistant could be a data structure where you store the structure of the
information and automatically generate text based on the characteristics
and needs of the user. A step beyond automatic linking would be automatic
text generation.
L. ROBERTS: A system where I could say that I wanted to read Moby Dick,
but as if Adam Smith had written it, and instead of a whale...
G. BEAR: Like looking at a CD and knowing whether it is music or CD-ROM
by the "structure" of the refraction.
E. DREXLER: Imagine a hypertext system being used globally. How might machine
translation evolve in such a system? You might not be able to read Japanese,
but you might find a translation made from the Japanese side of the language
barrier. The interesting parts have parallel structure in both languages.
In crucial parts, it is important that the translation be very good. But
it is also true that the crucial points will be be annotated, discussed
in depth, etc., so that the crucial points will come to be perfectly translated
as a result of this debate. This will either bring the world together better,
or lead to some really bloody fights.
G. BEAR: Why not, with advanced nanotechnology, have interfacing that allows
you to learn groups of languages really quickly.
AUDIENCE: I want to attempt to link together nanotechnology and hypertext.
The suggestion that with enough information, you store only the structure
rather than the information itself, means that at that point you start having
a creative memory. Memory as a creative process is a new way of looking
at memory, as advanced by Crick and others. The trick for people on the
forefront of this notion is to remember that there is no such thing as storage
memory. It's like you're a mathematician -- you're always proving these
theorems again for improving. It's sort of this theory of memory that is
going on. In the same way, if we actually get a hypertext system, we'll
have a Gaia-like consciousness emerging. This is pretty far out, but not
so much farther out, really, than nanotechnology is and in the same sense
that your type-II nanotechnology, which I think is just as inevitable as
type I nanotechnology is, for the same reasons you outlined. If we have
any way of coping with it, it may very well be through this "Gaia":
creating a virtually conscious "over" self to take that on.
G. BENFORD: Or possibly there will be a tremendous generation gap, because
those of us who are old won't be able to adapt.
G. BEAR: A good reason for death.
G. BENFORD: Speak for yourself.
G. BEAR: Of course, self-interest will remove us from the sphere of death,
but think of it. In the process of limiting and specializing, and if you
restructure the brain to allow the ability to handle these future possibilities,
then, are you in fact, retaining the personality? Perhaps not. You may only
retain memories. Your personality changes every four or five years anyway
to a small extent, and the memories are altered or in some cases lost. This
technology compels us to redefine what we mean by the human spirit, because
we are getting smart enough to realize that what we thought we meant is
not what we really mean. This is discouraging.
AUDIENCE: What are the immediate future implications of hypertext for the
people here? People who want to contribute to nanotechnology. I am thinking
particularly of the PATH project, which has put Engines
of Creation on disk and made available to anyone who wants it.
M. STIEGLER: Eric once made a wild assertion to me that of course you couldn't
do Engines of Creation other than in text. This in discussion of
the text versus pictures idea. And I said, "Eric, I can do 'EOC' in
pictures." And he said, "Hmmm, we will have to talk about that
some day." So who knows what we might see in five years.
M. THOMAS: I would like to bring this discussion a little bit closer. I
think that everyone here is aware of what hypertext is and that it is indeed
going to be important to the process of technological development. It's
going to be important to society in many ways. But right now we are considering
building a large model to support the nanotechnology discussion. How large
a model are we looking at? How much information is there to deal with? What
kind of time frames? How is this justified?
E. DREXLER: I think right now what is needed is a very fuzzy and flexible
concept. There is a certain minimum amount of information that you have
to have to make it worthwhile having a hypertext structure for accessing
it and exploring it. I don't know what that size is; it depends on the information
domain. I suspect the body of things that has already been written on nanotechnology
could be compressed and worked around and made into something nice, and
things in that direction are being worked on. But it expands from there
in more or less unbounded fashion. I would like to have not only the a large
number of chemical abstracts available in a hypertext medium, but all the
articles that are referenced as well. That would make my research a lot
easier and I am sure make a lot of other people's research in other areas
easier as well. So, I say the more the better, but a little bit is still
valuable.
G. BEAR: It will save my life and my marriage. I buy at the rate of about
five shelves worth of books per week.
AUDIENCE. A more individualized question. Every person in this room is certainly
among a unique portion of the population in that their interest level is
sufficiently high that they would be motivated to spend a major portion
of a weekend to try and learn more about this technology and simply be involved.
I don't know if the Foresight Institute
has given out the address yet , or stated how to stay involved, but after
this weekend is through I feel like I am going to want to keep contributing;
to be involved, and see if I can add to this growing volume of information
and partake in that. I was wondering what are the steps that you would recommend?
[Note: Foresight Institute,
PO Box 61058
Palo Alto, CA 94306 USA
tel 415.324.2490
fax 415.324.2497
Send questions to inform@foresight.org.
]
M. THOMAS: One of the panels tomorrow will concern the critical path of
nanotechnology, and will include a discussion of where the Seattle group
is going with this hypertext model and the kinds of things they are doing
and who you can contact and where you can go. They have been getting together
on a regular basis and would like to see more people show up at those meetings.
Yes, there is a lot of hard work to do, and yes, you are welcome to come
and do some of it. I think we have beat this to death, but I think it was
very useful. I think its a good introduction to what we are going to do
tomorrow with the social issues panel.
Next page of the NanoCon Proceedings
Previous page of the NanoCon Proceedings
Return to NanoCon Proceedings Table
of Contents
Return to Jim's Molecular Nanotechology Home Page
Return to Web Site Design and Authoring
Services
Comments and correspondence? Send email to Jim Lewis:
nanojbl@halcyon.com
NanoCon Proceedings © Copyright 1989, by NANOCON
This page is part of Jim's Molecular Nanotechnology Web, copyright ©1996
James B. Lewis Enterprises. All rights reserved.
Last updated 26June96.
The URL of this document is: http://www.halcyon.com/nanojbl/NanoConProc/nanocon7.html