Why Cyberspace?
Randy Farmer -- A survivor of the Lost Patrol
This article originally appeared in the Journal of Computer Game Design last
October, and is positioned as kind of primer. Those of you who received a
copy of this a hackers, hit 'n' now :-).
----------------------------------------------------------------------------
Cyberspace: Getting There From Here.
(c) Copyright 1989 F. Randall Farmer
>Randy was lead C64 programmer on the fabled 'Habitat'(1) project
>for Lucasfilm Games. He has taken a hiatus from building Cyberspace
>while he works on a new telecommunications product at AMIX, a
>subsidiary of Autodesk. He can be reached via Usenet as
>ra...@xanadu.com or through the Journal BBS.
"Golden Vaporware" is a term I've heard Ted Nelson(2) use to
describe such exciting computer projects as Xanadu, Dynabook and
Cyberspace. I understood the term to mean Really-Great-Stuff-That-
People-Are-(Sort-Of)-Working-On-And-Never-Ship. But now, it
seems, "The Age Of Golden Vaporware" has arrived. Ted's Xanadu
project is now well under way and should ship product in '90. There
is a company called Dynabook Technologies that has released "step 2"
of the "20 steps"(3) to Alan Kay's vision of the ultimate laptop. And
Cyberspace has been prototyped by QuantumLink and Lucasfilm.
These projects are likely to succeed not because they are
exciting to us techie-geeks, but because they have solid commercial
and technical foundations. Here I will illustrate a path from current
software and hardware technologies to the Cyberspace that many of
us have dreamed of for almost 20 years.
What is Cyberspace anyway?
"Cyberspace" is a fuzzily defined word that has been in the
press a lot recently(4)(at least in Silicon Valley). Vernor Vinge first
described a vision of what might now be called Cyberspace in his
novella True Names(5). The hero of Vinges's story connected to "The
Other Plane" using EEG electrodes placed on the forehead. The Other
Plane was a place where complicated software systems were
represented by familiar objects. The Max Headroom television show
used similar representations in various episodes. Other fictional
treatments include William Gibson's Neuromancer (6), which has
been turned into a popular graphic adventure, and other
"Cyberpunk" novels. Here the vision of Cyberspace is based around
commercial and governmental network computer access, where
"cowboys" and "wizards" break in and wreak fiscal havoc. In these
works, Cyberspace is a metaphor for abstract concepts.
Autodesk, maker of the best selling AutoCAD system, has
thrown its hat into the Cyberspace ring where it joins VPL's Jaron
Lanier and NASA-AMES. All of these products share similar base
technologies and assumptions: two head-mounted TV monitors for
stereo imaging, a powerful computer, and a 3D input device like
VPL's DataGlove. They are primarily single user systems, and, in the
case of at least Autodesk's product, probably aimed at CAD users who
want to edit drawings in three dimensions. There are many other
applications, but discussion is beyond the scope of this article. The
key to all of these systems is enough computer power to do double-
buffered stereoscopic real-time graphics at a high frame rate (say
20-30 frames/second). This technology is expensive: in the case of
VPL's "virtual reality" system (their special two user version) the
price tag weighs in at $69,925. The "Eyephones" cost $9,400 and the
glove $8,800.(7) These companies envision Cyberspace as a new kind
of user interface.
To me, Cyberspace is a place, not just an interface or a
metaphor. A place where people, regardless of location, hardware, or
purpose can get together in a participatory experience to conduct
business, socialize, or have a good game of SpaceCombat9.6.
Why this vision instead of the others? People. Unlike the
interactions with artificial personalities created for computer
adventure games, Cyberspace interactions could create relationships
that are "greater than the sum of their parts". Here the consequences
of individual actions take on a greater significance because they
affect the world of, and the personal belongings of the other
participants. Because Cyberspace will be so malleable, each
individual can "make a dent"(8) and participate in its ongoing
creation and adaptation.
This is not speculation! During Habitat's beta test, several social
institutions sprang up spontaneously: There were marriages and
divorces, a church (complete with a real-world Greek Orthodox
minister), a loose guild of thieves, an elected sheriff (to combat the
thieves), a newspaper (with a rather eccentric editor), and before
long two lawyers hung up their shingle to sort out claims. And this
was with only about 150 people! My vision encompasses tens of
thousands of simultaneous participants.
How can it work?
There are several major problems facing a large-scale
Cyberspace system: Bandwidth, Graphic Resolution, User Interface
Standards, Event Integrity, Data Communications Standards, and
Computer Horsepower.
Bandwidth is the most overrated problem. Everybody thinks
you need to send megabytes of data every minute to each user in
order to make Cyberspace work. This is simply not true. Habitat
works at a mere 300 baud. The keys to keeping bandwidth down are
distributed processing (having the local computer do most of the
display and interface work), object oriented command messages, and
avoiding communication intensive graphics primitives (such as those
associated with NAPLPS, that slower-than-molasses protocol they use
for Prodigy and other Videotex systems).
Graphic Resolution, User Interface and Local Horsepower issues
should all be lumped together. To be viable over time, Cyberspace
must be designed with the "to each according to his abilities"
principle. If a C64 with NTSC TV and joystick-only interface
encounters a 60 megahertz 486 user with Eyephones and DataGlove,
the interaction should be filtered so that each user can have a
satisfying experience (e.g the C64 user doesn't see all the intricate
detail the 486er has put into his attire, and all people walk exactly
the same, but they can still sit, chat, and play a nice game of
Strattagema.)
Event Integrity is a side effect of distributed processing. Simply
put: you can't trust the home computer. Hackers have all the time in
the world to crack the software and make it send messages it
shouldn't. This is akin to the long debated copy protection problem,
but comes with its own unique solution: The Host. Since the host is
the final arbiter of all events, it should be programmed at a very
high level to reject bad messages.
Data Communications Standards are key to the future viability
of Cyberspace. A protocol is required that is efficient and forward
looking, with hooks for future revisions (as bandwith increases) and
data encryption. Anything that is going to be used for business will
need to transmit its data securely for protection against espionage.
Also, the standard needs to address such issues as Email and
gateways to those entrenched text-only systems. I want to read my
Usenet mail printed on a sheet of paper in my Cyberhand!
Since the host coordinates all activity for all users, Host
Horsepower is by far the most critical issue. For the first few years of
operations, Cyberspace will have a single centralized host consisting
of several connected multi-processing computers. In order to keep
overhead to a minimum, bandwidth and database access will need to
be tightly limited. The communications protocol will need to be
super efficient and messages must be kept tiny. Eventually there
will be several independent Cyberspaces, perhaps on LANs or a
larger BBS. Of course, while you are walking around in your office
Cyberspace, you might want to go to other Cyberspaces, so a
distributed host model will need to be designed. By not having a
central host keep track of everything and everyone, this distributed
model will actually decrease the load on each host. This will allow
Cyberspace to take on a few interesting features: 1) Each host can be
a different kind of place, with different 'rules' (e.g. A role-playing
Cyberhost would probably have monsters, combat rules -including
character death- and a no-holds-barred policy about participant
behavior whereas an office Cyberhost would have access protection
and wouldn't allow personal combat of any type), 2) Increased
capacity without loss of access, and 3) Faster growth.
When will it happen?
Rather than pin a date on my Cyberverse, I will outline a
possible implementation path, including all of the developments
publicly announced to date.
Event: Spring 1989 "Habitat" A.K.A. "The Poor Man's Cyberspace"
Ship Low End Prototype:
Third person, 2D, Low resolution, Low bandwidth,
multi-user Cyberspace with joystick/keyboard interface
Purpose: Entertainment
Event: Fall? 1990 "CyberCad"
Ship commercial CAD version of Cyberspace Interface
Eyephones, DataGlove/DataSuit. First person, single-user
3D graphics (wire frame on small computers).
Purpose: Commercial CAD
Event: AT&T starts installing nationwide ISDN(9), promises
installation
by 2001 (this should drive bandwidth to 1.5gigabits/second!)
Event: 'DataGlove, the cheap version' arrives.
Event: 9600/19.2k baud modems <$100 dollars
Event: A Cyberspace team defines the first Cyberspace Data and
Communications Standards (an event they will regret later
when this interim hack becomes the defacto standard)
Event: 486/68040 50+ megahertz computers sell for <$1000 dollars.
Event: (about 1995) "Cyberverse 1.0" (U.S.? Japan?)
Ship first multi-purpose Cyberspace universe
Eyephones optional, mouse or glove required
Local data stored on high density media (CDROM?)
9600baud minimum speed, LAN version available.
(host still required)
Event: Computers get 10X faster and 10X more memory/storage.
Ongoing event: AT&T still installing ISDN
Event: (about 2000) Japan completes ISDN installation, selling
ISDN 'modems' cheap!
Event: First 'cracker' group has successfully done measurable
damage to the global Cyberverse.
The back doors are all closed. Or are they?
Event: After cursing the original Cyberdesigners for several years
now, the Data and Comm standards are updated to support.
multi- and distributed-host models. Massive testing required.
Event: First suicide attributed, by the media, to a game played while
in Cyberspace.
Event: First combination television/phone/computer successfully
mass marketed??
Event: (about 2000) "Cyberverse 2.0"
Ship distributed host version
Supports latest user interface hardware (optionally).
Now a host is not required.
Event: (about 2010) AT&T finally completes ISDN. Wall size
videophone reaching suburban markets (10-20 years late?)
Event: Congress takes significant note of the Cyberverse, because
trade unions (and other lobbies) note significant
changes in the distribution of political power.
First Cyberspace lobby arrives in Washington.
Event:10-20% of Americans spend over 4 hours per day "on the
Other Side". First chapter of Mothers Against Cyberspace
forms, claims 'Junior' never goes out to playI
(We gotta think about this stuff too, ya know!)
Event: (the Future) "Cyberverse 3.0" A.K.A. "The Big Payoff"
Distribute version 3.0 via Cybersoft retail outlets in Cyberspace
100,000+baud, allowing life-like personalizations
EEG direct input? Hologram Video? Retinal Projection?
Full Body digitizing? Forced feedback?
Voice input/output? World access via remote-robots?
Farmer, you're full of Cybercrud!(10)
Oh, yeah? A 386/CDROM version of Habitat will soon be
released in Japan. Other foreign companies are also interested in this
technology, and some are talking about multi-machine support,
standards, and the future. My only fear is that the long-term nature
of this project will cause American companies to shy away, and let
yet another new technology be monopolized by other, more forward-
looking nations.
(1)Credit where credit is due: Chip Morningstar was original designer
and leader of the Habitat project. Aric Wilmunder was the other
primary C64 programmer. Janet Hunter did most of the Host work. I
was also chief operator and world builder. Habitat has not yet been
released in its complete form (it will in Japan in '90). For a taste of a
much-scaled-down version, check out Club Caribe on QuantumLink
(you'll need a C64 & software). Call 800-782-2278 for details.
(2) Nelson, Theodor, Standard Public Xanadu Speech. Xanadu is the
name of the original mega-hypertext product.
(3) Khosla, Vinod, of Dynabook Technologies quoted in Bay Area
Computer Currents "Dynabook! Dynamite?" Vol 7,#4 pp 22-24.
(4) Laurel, Brenda, mentions Cyberspace in "New Interfaces for
Interactive Entertainment" in Vol 2, #5 of The Journal of Computer
Game Design.
(5) Vinge, Vernor, "True Names" republished in an anthology "True
Names and Other Dangers" (mandatory reading for would-be
Cyberhacks)
(6) Gibson, William, "Neuromancer" (recommended reading)
7)Nix, Shan, interviews Jaron Lanier of VPL in The San Fransisco
Chronicle, "Welcome to the World of Hyper-Reality" early August
1986, pp B1,B6
(8) ibid
(9)ISDN:Integrated Services Digital Network: instead of converting
digital data to analog (via modem) and back again, digital data is
dropped directly on to phone lines. Maximum throughput of full scale
fiber-optic ISDN is 1 to 6 gigabits/second or more. Several foreign
countries have made ISDN a national priority.
(10)Nelson, Theodor: Computer Lib/Dream Machines Microsoft Press
Revised Edition 1987 pp 27-29
----------------------------------------------------------------------------
Well, there you have it, my first published article :-). All comments welcome.
Don't pay too much attention to the timeline, it's meant mostly to point
out that there are no real technical barriers to this vision, and the future
should only see the kind of improvments we've seen with computers: Faster,
Bigger, and more Resolution.
Nobody you know
Peter da Silva
> This is not speculation! During Habitat's beta test, several social
> institutions sprang up spontaneously: There were marriages and
> divorces, a church (complete with a real-world Greek Orthodox
> minister), a loose guild of thieves, an elected sheriff (to combat the
> thieves), a newspaper (with a rather eccentric editor), and before
> long two lawyers hung up their shingle to sort out claims. And this
> was with only about 150 people! My vision encompasses tens of
> thousands of simultaneous participants.
A similar phenomenon was observed in PCC's Public Caves system, and its
descendents (Pubcave at Berkeley, and the Caves BBS that I ran for a while
here in Houston... the software is still running as a program on Sugarland
UNIX: 713-438-5018). You don't need fancy graphics to get the consensual
reality you're playing with. In fact, I'd say that often the fancy graphics
get in the way. Which do you prefer: comic books or paperbacks?
> Bandwidth is the most overrated problem. Everybody thinks
> you need to send megabytes of data every minute to each user in
> order to make Cyberspace work. This is simply not true.
This is almost a tautology. Put enough processing power locally and you don't
need a high-speed link. Like X-windows versus NeWS, or in your case like
Habitat versus Teletext.
> Graphic Resolution, User Interface and Local Horsepower issues
> should all be lumped together. To be viable over time, Cyberspace
> must be designed with the "to each according to his abilities"
> principle.
Don't rule out a text-only interface. Some of us prefer text or a text-oriented
GUI like Smalltalk's to cartoon characters.
> Event Integrity is a side effect of distributed processing. Simply
> put: you can't trust the home computer.
Viruses. Crackers. Sure.
> This ... comes with its own unique solution: The Host.
I don't see that as the only solution. Each workstation can do its own event
validation, a-la Kerberos. (using the term workstation loosely, from Atari
800, Apple II, Commodore 64 on up)
> Data Communications Standards are key to the future viability
> of Cyberspace.
TCP/IP and PPP. OSI. ISDN. That's what you're gonna get in the business world.
If you want a single cyberspace it had better talk the language of business.
> Usenet mail printed on a sheet of paper in my Cyberhand!
That's up to your workstation.
> Since the host coordinates all activity for all users, Host
> Horsepower is by far the most critical issue. For the first few years of
> operations, Cyberspace will have a single centralized host consisting
> of several connected multi-processing computers.
You mean like the CDC Plato system? Look how successful that was. No,
distributed processing is the way to go. The "host" should be no more than
a nameserver and possibly a big switch.
> By not having a
> central host keep track of everything and everyone, this distributed
> model will actually decrease the load on each host.
Like I said.
> Cyberhost would probably have monsters, combat rules -including
> character death- and a no-holds-barred policy about participant
> behavior whereas an office Cyberhost would have access protection
> and wouldn't allow personal combat of any type
Again, this is simply an application running on the network. Like NetWar
or MazeWar taken to an extreme. So long as you have the IPC (which is at the
base of it a given) then you'll be able to play games.
> When will it happen?
It's here. You're using it. Just get a better workstation.
> Event: 9600/19.2k baud modems <$100 dollars
Nah. Gimme ISDN. It's already being installed in test markets.
> Event: 486/68040 50+ megahertz computers sell for <$1000 dollars.
Taking a RISC aren't you?
> Event: First 'cracker' group has successfully done measurable
> damage to the global Cyberverse.
Internet Worm, 1988.
> Event: After cursing the original Cyberdesigners for several years
> now, the Data and Comm standards are updated to support.
> multi- and distributed-host models. Massive testing required.
Go tell that to MIT and Project Athena.
> Event: Congress takes significant note of the Cyberverse, because
> trade unions (and other lobbies) note significant
> changes in the distribution of political power.
Computer crime laws, Internet Worm trials, Intel versus NEC, Apple
versus everyone...
> Event:10-20% of Americans spend over 4 hours per day "on the
> Other Side".
That's me!
Habitat is a nice user interface to Public Caves, but you've been scooped
by over 10 years.
--
Peter "Have you hugged your wolf today" da Silva <pe...@sugar.hackercorp.com>
`-_-'
'U` "Really, a video game is nothing more than a Skinner box."
-- Peter Merel <pe...@basser.oz>
Randy Farmer -- A survivor of the Lost Patrol
(Peter da Silva) writes lots of interesting stuff in reply to
my Article...
>Don't rule out a text-only interface. Some of us prefer text or
>a text-oriented GUI like Smalltalk's to cartoon characters.
Exactly my point! But I disagree that 'text is better' as a rule. But
if that's what you like, fine. On a lark, we designed a text-only interface to
Habitat. It would work fine. (We could agree all day!)
>
>TCP/IP and PPP. OSI. ISDN. That's what you're gonna get in the business world.
>If you want a single cyberspace it had better talk the language of business.
>
That is what there are gateways for. "What protocol" is a non-issue (but I bet
something like ISDN will EVENTUALLY be the carrier).
>Peter "Have you hugged your wolf today" da Silva <pe...@sugar.hackercorp.com>
Have we met? Your name is familiar.
I am GLAD to hear we were 'scooped' by ten years. It just helps validate
14 years in multi-user environments reasearch ;-D.
BTW, I agre that the various networks are all proto-cyberspaces
of thier own... I just want them connected, and the representations
standarized.
--------------------------------------------------
Someone posted a message that this group is for discussion of the work going
on at NASA and Autodesk. Is that True? Eyephones and DataGlove only? If
so, this group is mis-named... It SHOULD be alt.VPL.is.the.only.way!
John Walker put it this way at Hackers 5.0: ''Assume the technology
(eyphones, dataglove) were "free" today: What would you use it for?''
That's what I think this group is for. I think Habitat (AKA Club Caribe)
and Public Caves can provide some interesting data points on this issue.
--------------------------------------------------
See you on the Other Side
Randy
Ken Chin-Purcell
> k...@uc.msc.umn.edu (Ken Chin-Purcell) writes:
> > A PostScript derivitive would not be the way to go. The idea is to
> > communicate object desciptions, not images.
>
> Eh? I don't know all that much about PostScript, but I do no that it
> doesn't involve the communications of images. It describes lines, arcs,
> shapes. By images I'm guessing you mean bitmaps [....]
I was thinking of PostScript in terms of lines, arcs, etc.
By suggesting IGES or PDES over PostScript, I was pointing out a
problem engineers have been dealing with for quite a while:
A description of how a part looks is not a total description.
Suppose I hand you, in Cyberspace, a coffee mug full of hot coffee.
In data terms, I need to pass to your processing world a
description of an object that has weight, topology, surface
texture, and thermal properties. Worse, it has a hot liquid
with a free surface. And if it's good coffee, you would want
me to transmit the smell and taste too :').
We need to think beyond graphics.
- Ken
Christopher Hassell
#
# Club Caribe is certainly weird, and technologically impressive GIVEN
# it's run on Comm 64's -- but its operation raises another obstacle:
# have you ever observed a Club Caribe discussion? Talk about pearls
# before swine. If this is where cyberspace is taking us, I'm not sure
# I want to go!
# --
# BOB JACOBSON, Associate Director, Human Interface Technology Laboratory
# University of Washington (FU-20), Seattle, WA 98195 USA
# (206) 543-5075 (voice), 543-5380 (fax) * blue...@well.uucp
# >> "We can do virtually anything." <<
### C>H> ### { uunet!rutgers!sunybcs , ncar , nbires } !boulder!tramp!hassell
Peter da Silva
> Someone posted a message that this group is for discussion of the work going
> on at NASA and Autodesk. Is that True? Eyephones and DataGlove only? If
> so, this group is mis-named... It SHOULD be alt.VPL.is.the.only.way!
That was me. And that was a bit strong. What I meant to say is that this group
isn't for literary, technical, or political discussions of William Gibson's
books. About all he's contributed to the *real* virtual reality is a name:
Cyberspace.
> John Walker put it this way at Hackers 5.0: ''Assume the technology
> (eyphones, dataglove) were "free" today: What would you use it for?''
Me? A virtual windowing environment in the middle of Dream Park.
--
Peter "Have you hugged your wolf today" da Silva <pe...@sugar.hackercorp.com>
GurgleKat
> In article <30...@hub.UUCP>, pe...@cavevax.ucsb.edu (GurgleKat) writes:
>> k...@uc.msc.umn.edu (Ken Chin-Purcell) writes:
>> > A PostScript derivitive would not be the way to go. The idea is to
>> > communicate object desciptions, not images.
>> Eh? I don't know all that much about PostScript, but I do know that it
>> doesn't involve the communications of images. It describes lines, arcs,
>> shapes. By images I'm guessing you mean bitmaps [....]
> Suppose I hand you, in Cyberspace, a coffee mug full of hot coffee.
> [...]
> We need to think beyond graphics.
Agreed. But I don't think we should be arguning about PostScript so much as
the basic concept behind a reailty-description language. It's true PostScript
knows nothing about physics. But "something like PostScript" was the original
phrase I used, I think. Perhaps PostScript II would start to know something
about physics. The name of it isn't really important. It's the basic concept
that I'm getting at, and I think we agree on that. A reality description
language would not transmit 3D bitmaps, however that might be accomplished,
and thus removes some of the processing responsibility from the image
originator. That's the real point here.
--
Pete Gontier : pe...@cavevax.ucsb.edu; outgoing .UUCP cause me grief
Editor, Macker : Online Macintosh Programming Journal; mail for subscription
Hire this kid : Mac, DOS, C, Pascal, asm, excellent communication skills
"This was it. This was what he was, who he was, his being. He forgot to eat.
Sometimes he'd resent having to leave the deck to use the toilet..."
-- William Gibson, _Neuromancer_
Jim Hutchison
>A description of how a part looks is not a total description.
>
>Suppose I hand you, in Cyberspace, a coffee mug full of hot coffee.
>In data terms, I need to pass to your processing world a
>description of an object that has weight, topology, surface
>texture, and thermal properties. Worse, it has a hot liquid
>with a free surface. And if it's good coffee, you would want
>me to transmit the smell and taste too :').
Indeed, you need to communicate to my computer/space-generator
all the data about this cup of coffee that it can generate, that
I care about. I may not have odor-generators, so I may not
want the aroma. I may not have heaters, I may not want the
temperature. I may not have the resolution to examine the cup
in detail, or the interest in doing so. You may need only send
simple information such as: blue cup of brown liquid coffee.
When my machine gets it, it can then add all the detail I want
from my databases on blue, cups, brown, coffee, and liquids.
Optionally I might want to accept your definitions, but as
there is not guarantee of hardware/software compatibility
or good taste...
Admittedly it's just a thought which seems to agree with the
notion that you can't transmit all the details in the time
required (immediately).
>We need to think beyond graphics.
Yes, indeed beyond data even.
--
/* Jim Hutchison {dcdwest,ucbvax}!ucsd!celerity!hutch */
/* Disclaimer: I am not an official spokesman for FPS computing */