*****

Spreadsheets and Cyberspace

Marcos Novak
"Liquid Architectures in Cyberspace"

from
Michael Benedikt
Cyberspace: First Steps
1991

Introduction

What is cyberspace? Here is one composite definition:

Cyberspace is a completely spatialized visualization of all
information in global information processing systems, along pathways
provided by present and future communications networks, enabling full
copresence and interaction of multiple users, allowing input and
output from and to the full human sensorium, permitting simulations of
real and virtual realities, remote data collection and control through
telepresence, and total integration and intercommunication with a full
range of intelligent products and

environments in real space.

Cyberspace involves a reversal of the current mode of interaction with
computerized information. At present such information is external to
us. The idea of cyberspace subverts that relation; we are now within
information. In order to do so we ourselves must be reduced to bits,
represented in the system, and in the process become information anew.

Cyberspace offers the opportunity of maximizing the benefits of
separating data, information, and form, a separation made possible by
digital technology. By reducing selves, objects, and processes to the
same underlying groundzero representation as binary streams,
cyberspace permits us to uncover previously invisible relations simply
by modifying the normal mapping from data to representation.

To the composite definition above I add the following: Cyberspace is a
habitat for the imagination. Our interaction with computers so far has
primarily been one of clear, linear thinking. Poetic thinking is of an
entirely different order. To locate the difference in terms related to
computers: poetic thinking is to linear thinking as random access
memory is to sequential access memory. Everything that can be stored
one way can be stored the other; but in the case of sequential storage
the time required for retrieval makes all but the most predictable
strategies for extracting information prohibitively expensive.

Cyberspace is a habitat of the imagination, a habitat for the
imagination. Cyberspace is the place where conscious dreaming meets
subconscious dreaming, a landscape of rational magic, of mystical
reason, the locus and triumph of poetry over poverty, of "itcanbeso"
over "itshouldbeso."

The greater task will not be to impose science on poetry, but to
restore poetry to science.

This chapter is an investigation of the issues that arise when we
consider cyberspace as an inevitable development in the interaction of
humans with computers. To the extent that this development inverts the
present relationship of human to information, placing the human within
the information space, it is an architectural problem; but, beyond
this, cyberspace has an architecture of its own and, furthermore, can
contain architecture. To repeat: cyberspace is architecture;
cyberspace has an architecture; an

d cyberspace contains architecture.

Cyberspace relies on a mix of technologies, some available, some still
imaginary. This chapter will not dwell on technology. Still, one brief
comment is appropriate here. A great number of devices are being
developed and tested that promise to allow us to enter cyberspace with
our bodies. As intriguing as this may sound, it flies in the face of
the most ancient dream of all: magic, or the desire to will the world
into action. Cyberspace will no doubt have physical aspects; the
visceral has genuine power over us. And though one of the major themes
of this essay has to do with the increasing recognition of the
physicality of the mind, I find it unlikely that once inside we will
tolerate such heavy devices for long. Gloves and helmets and suits and
vehicles are all mechanocybernetic inventions that still rely on the
major motor systems of the body, and therefore on coarse motor
coordination, and more importantly, low nerve ending density. The
course of invention has been to follow the course of desire, with its
access to the parts of our bodies that have the most nerve endings.
When we enter cyberspace we will expect to feel the mass of our
bodies, the reluctance of our skeleton; but we will choose to control
with our eyes, fingertips, lips, and tongues, even genitals.

The trajectory of Western thought has been one moving from the
concrete to the abstract, from the body to the mind; recent thought,
however, has been pressing upon us the frailty of that Cartesian
distinction. The mind is a property of the body, and lives and dies
with it. Everywhere we turn we see signs of this recognition, and
cyberspace, in its literal placement of the body in spaces invented
entirely by the mind, is located directly upon this blurring boundary,
this fault.

At the same time as we are becoming convinced of the embodiment of the
mind, we are witnessing the acknowledgment of the inseparability of
the two in another way: the mind affects what we perceive as real.
Objective reality itself seems to be a construct of our mind, and thus
becomes subjective.

The "reality" that remains seems to be the reality of fiction. This is
the reality of what can be expressed, of how meaning emerges. The
trajectory of thought seems to be from concrete to abstract to
concrete again, but the new concreteness is not that of Truth, but of
embodied fiction.

The difference between embodied fiction and Truth is that we are the
authors of fiction. Fiction is there to serve our purposes, serious or
playful, and to the extent that our purposes change as we change, its
embodiment also changes. Thus, while we reassert the body, we grant it
the freedom to change at whim, to become liquid.

It is in this spirit that the term liquid architecture is offered.
Liquid architecture of cyberspace; liquid architecture in cyberspace.

Cyberspace Navigation, Synthesis, and Rendition

Some initial definitions are necessary before we proceed: cyberspace
navigation refers to the traversal of information spaces; cyberspace
synthesis refers to the reconciliation of different kinds of
information into a coherent image; cyberspace rendition refers to the
production of highquality graphic presentation of that image. These
are separate tasks.

The Hypermedia Navigator

Navigation through cyberspace is achieved by interacting with a
hypermedia navigator, a virtual control device that follows the user
and always remains within arm's reach. It is possible for the user to
circumvent the cyberspace synthesizer and enter and traverse the space
of the navigator, riding the links, as it were.

Every paragraph an idea, every idea an image, every image an index,
indices strung together along dimensions of my choosing, and I travel
through them, sometimes with them, sometimes across them. I produce
new sense, nonsense, and nuisance by combination and variation, and I
follow the scent of a quality through sand dunes of information. Hints
of an attribute attach themselves to my sensors and guide me past the
irrelevant, into the company of the important; or I choose to browse
the unfamiliar and tumble through volumes and volumes of knowledge
still in the making.  Sometimes I linger on a pattern for the sake of
its strangeness, and as it becomes familiar, I grow into another self.
I wonder how much richer the patterns I can recognize can become, and
surprise myself by scanning vaster and vaster regions in times shorter
and shorter. Like a bird of prey my acuity allows me to glide high
above the planes of information, seeking jewels among the grains,
seeking knowledge.

Just as hypertext allows any word in a normal text to explode into
volumes of other words, so a hypergraph allows any point in a graph to
expand to include other graphs, nested and linked to any required
depth. We may, of course, extend this idea to other media to arrive
upon corresponding hypermedia. We can now make some further
distinctions: static and dynamic, passive and active, pure and hybrid.
A static hypermedium is one in which the links are fixed and can only
be changed manually; a dynamic one is one where the links are in some
way variable. While the distinction of static/dynamic applies to the
links of a hypermedium, the distinction of passive/active applies to
the nodes between the links. A passive hypermedium is one in which the
information nodes themselves remain stable though the links may vary
dynamically; in an active hypermedium the information nodes themselves
can change. Pure hypermedia remain within the confines of a single
medium, hybrid ones roam freely. The hypermedia navigator, or
navigator for short, is a virtual device for traversing vast hybrid
hypermedia spaces that have both active links and dynamic nodes.

Sometimes I wander out of my world into the larger spaces. I travel
along, mostly empty. The passages I traverse are not still, however.
Along their boundaries processes sparkle, information flows like water
on a moist wall, schools of data swim around me curiously, and
lattices of fact and fiction tangle and untangle.  The ones I touch
open out into texts and images and places.

Every node in a hypermedium has a dimensionality. Hypertext, for
example, occurs in a onedimensional space, but we can easily envision
hypermedia with higher dimensions. While the dimensionality between
nodes in a text can open to a hologram, a point within the hologram
can open to an animation, a frame in the animation can return to a
text.

Every node in a hypermedium is therefore an information space, a space
of potential information, and the "text" of the node is the actual
information within that space. In an active hypermedium the
information within the information space, as well as the links among
spaces, my change according to internal or external conditions. These
conditions may be usercontrolled or userindependent. The processes
within which these conditions are embedded may be defined explicitly
by the user or may be autonomous.

My point in space is given by my navigator. Its forms can vary but the
idea remains the same: I control a point in an ndimensional space, say
a cube. I assign meaning to each of the axes, and to any rotational
parameters, material parameters, shape parameters, color and
transparency parameters, and so on, that describe the "reality" of my
icon. By moving my icon in this abstract space I alter the cyberspace
I occupy. My navigator follows me at all times, and my position within
it is fixed while I move with

in the cyberspace I have defined. Should I decide to search through a
slightly different "reality" all I need to do is reach out for my
navigator and alter a parameter. Otherwise, for more drastic
navigation, I can alter a dimension, or even the number of dimensions.
Finally, I may choose an entirely different coordinate system. In
every case my deck is responsible for synthesizing the requested
information in a new cyberspace.

If a point in one cyberspace is an entry into another cyberspace, then
a new navigator is spontaneously created and pushed onto my stack of
navigators. I can now maneuver in the new system without losing my
place in the old one. Since the stack of navigators constitutes a
pathway, I can be reached by anyone who encounters one of my icons,
opens a channel, and sends me a message.

There are no hallways in cyberspace, only chambers, small or vast.

Chambers are represented as nodes within my navigator. The topology of
their connections is established by the settings and nesting of the
coordinate systems of cyberspaces within my navigator. Chambers allow
different users to share the same background, as well as encounter and
interact with the same objects.

Motion 

There are two kinds of motion: motion within my navigator, and motion
within the cyberspace pointed to by my navigator.  If my place in my
navigator is fixed, the reality I experience is also stable, though I
can move in it and interact with the entities that inhabit it. If,
however, I concurrently change places within my navigator, then the
reality I experience is no longer stable, but fluctuating. As my
location within the navigator changes, new perhaps distant, realities
are brought forth to replace the old ones.  Entities, landmarks, and
landscapes appear and disappear, time and space become discontinuous,
and the increment of my motion changes from one scale to another to
fit the current reality.

Others 

Initially, a user has a personally configured cyberspace, and
maneuvers through it by manipulating the navigator. If this cyberspace
is configured to accept signals from other users, or simply Others,
that is, if communications ports have been defined and enabled, then
an appropriate indication is made within the cyberspace that another
user is hailing. At that point the users must agree on a manner of
interaction, ranging from simple text to full interpresence. For full
cyberspace interaction the coordinate systems of all interacting
users' navigators must be configured along identical orderings.
Partial coincidence will result in hiding of information.

This world of mine has ports: through them I gather and give. This
world has windows: through them I can see and be seen. Through ports
other worlds are accessible; through windows other worlds are simply
visible. I can open and close my ports and windows, naturally, and I
can also have curtains and filters that only permit some information
to enter or escape. Sometimes the information itself controls which
apertures are open, how much, and how.

Within my navigator traces of light move from node to node, indicating
the presence of others in the chambers of cyberspace; outside my
navigator, in the chamber I am currently occupying I encounter some of
them directly.

Interaction with others depends on the degree to which they share
information coordinate axes and orderings. If users wish to coexist
fully in the same virtual space, they must set their navigators to the
same settings. Communications established between users sharing
information axes partially will result in "ghosting," that is,
inexplicable and unpredictable "appearances" and "disappearances" as
aspects of the cyberspace of one user engage and disengage aspects of
the cyberspace of the other. In this re

spect cyberspaces are "consensual," since any complete exchange
requires a sharing of settings of the participants' navigators.

Using my deck, I enter the cyberspace. At first the world is dark, but
not because of an absence of light, but because I have not requested
an environment yet. I request my default environment, my personal
database. From it I choose my homebase, or workbase, or playbase. I am
in my personal cyberspace, and I am not yet in contact with others.
This is my palace, and it is fortified. Only guests can visit my
"fortress of solitude, " and in here I can be Superman to the Clark
Kent of my realspace self. Sometimes I organize my information around
my armchair and navigate through it at a glance, extracting what I
need by effortless exercises of will; other times, for the sake of
exercise or play, I scatter it around my globe and fly across immense
distances to recover minute recollections with the most strenuous
"physical" effort. Sometimes I use a single surrogate, other times I
divide into a legion.

I sense the presence of others. I see the traces of passage, the
flares of trajectories of other searches. Those who share my interests
visit the spaces around me often enough for me to recognize the
signature of their search sequences, the outlines of their icons. I
open channels and request communication. They blossom into identities
that flow in liquid metamorphosis. Layers of armor are dropped to
reveal more intimate selves; otherwise, more and more colorful and
terrifying personifications are built up in defense; but true danger
is gray.  The world opens and others flood in. Now there is congestion
and noise, interference, but also excitement, risk, and challenge. I
travel with the constellation of my possessions, and barter and trade
information. I can scan the horizon and avoid what is busy, enjoy what
is free.



Underlying Considerations

Minimal Restriction and Maximal Binding

The key metaphor for cyberspace is "being there," where both the
"being" and the "there" are usercontrolled variables, and the primary
principle is that of minimal restriction, that is, that it is not only
desirable, but necessary to impose as few restrictions as possible on
the definition of cyberspace, this in order to allow both ease of
implementation and richness of experience. In addition, maximal
binding implies in cyberspace anything can be combined with anything
and made to "adhere," and that it is the responsibility of the user to
discern what the implications of the combination are for any given
circumstance. Of course, defaults are given to get things started, but
the full wealth of opportunity will only be harvested by those willing
and able to customize their universe. Cyberspace is thus a userdriven,
selforganizing system.

Multiple Representations

Cyberspace is an invented world; as a world it requires "physics,"
"subjects" and "objects," "processes," a full ecology. But since it is
an invented world, an embodied fiction, one built on a fundamental
representation of our own devising, it permits us to redirect data
streams into different representations: selves become multiple,
physics become variable, cognition becomes extensible. The boundaries
between subject and object are conventional and utilitarian; at any
given time the data representing a user may be combined with the data
representing an object to produce . .  . what?

Digital technology has brought a dissociation between data,
information, form, and appearance. Form is now governed by
representation, data is a binary stream, and information is pattern
perceived in the data after the data has been seen through the
expectations of a representation scheme or code. A stream of bits,
initially formless, is given form by a representation scheme, and
information emerges through the interaction of the data with the
representation; different representations allow different corre

lations to become apparent within the same body of data. Appearance is
a late aftereffect, simply a consequence of many sunken layers of
patterns acting upon patterns, some patterns acting as data, some as
codes. This leads to an interesting question: what is the information
conveyed by the representation that goes beyond that which is in the
data itself? If a body of data seen one way conveys different
information than the same body of data seen another way, what is the
additional information provided by one form that is not provided by
the other? Clearly, the answer is pattern, that is, perceived
structure. And if different representations provide different
perceived information, how do we choose representations? Not only do
different representations provide different information, but in the
comparison between representations new information may become
apparent. We can thus distinguish two kinds of emergent information:
intrarepresentational and interrepresentational.

I substitute the characters on this and the next page of this text
with grey scale values; two images emerge, pleasingly rhythmical. The
gray tone of the letter e stands out, forming snakes along the pages;
I apply spline curves to the snakes, and, in another space, my text
itself is changingQwhat will it say? Now I combine the two pages, and
convert the result into a landscape, using the grey values to
represent height. What snakes were left after the combination of the
two pages become Serpent Mounds. The Others who were reading my text
with me a while ago are now flying over this landscape with me, but
only I can command it to change.  Today.

Attribute-Objects

There are no objects in cyberspace, only collections of attributes
given names by travelers, and thus assembled for temporary use, only
to be automatically dismantled again when their usefulness is over,
unless they are used again within a short timespan. Thus useful or
valued objects remain, while others simply decay. These collections of
attributes are assembled around nameless nodes in information spaces.
Travelers and processes can add, subtract, or modify collections of
attributes. By specifying a set of attributes, travelers define
temporary attributeobjects, and by ordering each attribute in any
desirable way they create their own dimensions for navigation. Three
such dimensions result in a threedimensional space, and motion along
this space results in transfigurations of the environment that enable
a holographic browsing. X, y, z, roll, pitch, yaw, color, material,
size, all the parameters that define my point in space are indices
into dimensions of attribute space. My motion makes my environment
melt from one image into another, and my navigation becomes a
knowledge dance.

In order for it to be possible to direct data into alternative
representations it is helpful to have an open and extensible highlevel
manner of structuring that information. Objectoriented programming has
shown the usefulness of interacting with objects that respond to
"messages" with "methods." We can envision a world of
attributeobjects, objects assembled as collections of attributes,
where each attribute is itself an attributeobject, to some limit.
Messages and methods are now attached not to objects, but to
attributes of objects, in such a way that if a series of objects share
an attribute, they can be expected to behave in a certain way.  This
implies a connection of attributes to affordances. For example, any
attributeobject that has a weightattribute, and whose weightattribute
is greater than some minimum value, can at any time be used as a
paperweight.

Furthermore, if these objects are not placed in any permanent system
of categorization, that is, if as few assumptions are made about what
these objects are, beyond their collection of attributes, then it
becomes possible to envision cyberspaces created along the parameters
of users' needs. Attributeobjects can be gathered and sorted by
attribute or combination of attributes, and these sorted collections
can then be mapped onto coordinate axes. An information space can thus
be formed, and motion through that space can imply "browsing."
Consider for example a point in Cartesian space, whose coordinates are
indices along the three axes that define the space, and whose
rotational parameters and material and other attributes are indices
along additional dimensions. Let us say that along the xaxis we have
some ordering of "objects that are blue," sorted by degrees of
blueness. Moving the point along the "blueness" axis, we scan these
objects as if browsing through the shelves of a library.

Other attributes can be mapped along the other axes. Moving the point
along those dimensions, or any kind of diagonal motion will alter the
reading of the cyberspace correspondingly. The point thus becomes a
control point in one of infinitely many potential cyberspaces, a
navigation tool through actual or inferred information.
Attributeobjects are locations in attributespace, and can themselves
open to reveal other spaces or information.

Attributeobjects are saved as code strings containing genotype
information. Selected kinds of attribute objects can evolve
autonomously by random mutation and cumulative change processes
(Dawkins 1986), the main survival criteria consisting, first, of
operation within available resources, and second, of interaction with
other attributeobjects. It will not be possible to assign all
pertinent attributes to all objects, or to foretell what attributes a
user may request for a particular journey through a cyberspace. It is
therefore necessary to provide the ability to infer new attributes
from those that are defined. This inference implies intelligence and
expertise. Each user must therefore develop inference rules and a
knowledge base with which to scan the environment and extract from it
those objects that are pertinent to the task at hand.

Cellularity and Distributedness

As far as possible, the computational requirements of cyberspace need
to be distributed among its users. By necessity, cyberspace itself
must be reduced to a digital communications standard that allows users
to exchange information in the most compact form possible, with
cyberspace synthesizers and renderers taking on most of the work. The
topology of perceived interconnected cyberspaces need not have any
direct connection to that of the array of support computers, since the
cyberspaces are perceived, not actual spaces. Like a cellular
telephone system, a neutral grid of computers will permit access and
communication within an apparent hierarchy of cyberspaces.

So far, most discussions of cyberspace and virtual reality have
focused on the current computational paradigm of Von Neumann
computing. The emerging paradigm of parallel distributed processing,
though still not developed enough to be an immediate contender, offers
an alternative view into cyberspace. In a neural net simulation,
information about objects, relations, laws, and other components of a
world, is not encoded in an explicit way. Rather, it is encoded
implicitly, as weightings on connections between simple computational
cells. Reality is an emergent property of the cell, a particular
statistical setting of interconnection strengths that changes, ever so
slightly, every time it is used, and manages to learn that which we
are not yet able to articulate by observing the patterns in our data
or our usage.

Mathematically, parallel distributed processing can be described in
terms of surfaces with local and global maxima and minima, and
convergence is given as a traversal along the surface to a state of
minimum "computational energy." The actual contour of the surface
depends on what the network has learned. In other words, in neural
networks information is already directly spatialized, according to the
networks' "learning. " Thus, an information landscape emerges through
usage. The implications of this for cyberspace is that the
"unimaginable complexity" implicit in the idea of moving through
interlinked information spaces, global networks, and so on, can be
"learned" by a neural cyberspace component within a hybrid system.
Subsequent to this learning and the resultant spatialized
visualization of it as a landscape, we can use our own navigational
skills to find the local extrema we are interested in.


Intelligence

Visualizations of information spaces can be understood in many ways:
static or dynamic, as in photography and cinema; passive or
(inter)active, as in television versus interactive computer graphics;
direct or hyper, as in normal text versus hypertext; and now, with
cyberspace, normal or intelligent. These sets of categories are
orthogonal to one another; they are, in essence, attributes of the
space itself. Some or all may exist at the onset of a session; new
ones may be added; old ones may be deleted or modified.

We need a provisional definition of "intelligence" in order to
proceed. Intelligence, as used here, refers to the capability of a
subject, object, space, time sequence, or process, to detect, respond
to, and modify its own or any other pattern. Patterns can be nested
within patterns and may be invisible if the data within which they are
found is represented in a manner that does not correspond to the
patternrecognition capabilities of another entity. Higher
"intelligence" can detect and operate upon patter

ns more deeply nested, while simpler "intelligence" is restricted to
surface patterns. A knowledge base, consisting of collections of
"patternrecognizer" attributeobjects, specifies what constitutes
pattern for a cyberspace entity, and under what conditions, and in
what ways, such a pattern may be modified at any given time. Pattern
recognizers need be very general, based perhaps on simple but general
methods for detecting unexpected correlations in data streams. Once
again, the capacity of neural nets to act as correlation mechanisms
for information that is only implicitly connected makes them suitable
for many such tasks. While training networks is timeconsuming, the
performance of hardware implementations is virtually real time
(Anderson and Rosenthal 1988).

Of course, not all "intelligence" will be operating at such a
fundamental level. Different attributeobjects will use different
methods of encoding responsiveness using procedural, or declarative,
processes, as well as other present and future artificial intelligence
techniques.

Subject, object, space, time, and process form the basic elements of
cyberspace. These, and any others that are added subsequently, need to
be considered not only continuously, but also discontinuously. Not
only may a subject detect a pattern of light and darkness and call it
"day/ night," but the patternobject "day/night" may itself detect the
recurring presence of certain subjects or objects at times called
"Monday," and may therefore adjust itself, or them, as required by the
nature of its intelligence.



Overlap, Transparency, and Identity

In physical space two objects cannot occupy the same space at the same
time. In cyberspace such a restriction is not strictly necessary. The
identity of objects does not have to be manifested physically; it can
be hidden in a small difference in an attribute that is not displayed.
The motivation for this is twofold: first, to allow a poetic merging
of objects into evocative composites, and second, to keep the
implementation of cyberspaces as simple as possible.

More precisely, the standard that implements cyberspace communications
need not be assigned the overhead of deciding when two objects meet
precisely, for example. It may be sufficient to place them so that
their bounding boxes touch. It is then the common task of the decks
that are viewing the two objects to resolve the situation according to
their capabilities and the representations chosen.

What seems most likely to occur is that entities will behave as
"ghosts, " passing through each other freely and interacting on the
basis of a list of operations vaguely reminiscent of their realworld
analogues. As we become acclimated to this new ecosystem, our
nostalgic desire for the vestiges of realworld physics will for the
most part disappear. We will adjust to cyberspace far more easily than
cyberspace will adjust to us.

This blending is possible because in the computer each entity is
indicated by a tokenvalue pair. Surface appearances are results of
particular values assigned to tokens, and even if two entities have
identical attributes, they can be distinguished by tracing the final,
absolute identity of their tokens, ultimately the physical address
allocated to them at the time of their creation. It is, nevertheless,
possible to envision a situation where, through appropriate
subsumption rules, if two entities have identical attributes they can
be made to collapse into one, but this seems to limit us to a reality
that is surely familiar but not necessarily appropriate to cyberspace.

Animism and Empathy

In cyberspace animism is not only possible, it is implicit in the
requirement that all objects have a degree of selfdetermination, or
are controlled by an Other. Thus a measure of empathy is required in
order to comprehend the behavior of the entities one encounters. To
the extent that any object may act as a front for a real person, its
motives will have to be considered.

Furthermore, it is not simply the animism of objects that provides
information, but the animism of the space itself. More specifically,
any entity in cyberspace, including space itself, and "cybertime," can
provide several levels of information: information about something
else, information about itself, information about the observer,
information about the surrounding environment, and global information.
Beyond this physical atomism, any entity can provide information about
structures of information that it is allowed to have knowledge of.
Instead of atomism, organism.  Within a region of cyberspace, time
itself may pulse, now passing faster, now slower.

Lorca's duende is therefore not only a poetic description of an
attitude toward the construction of cyberspace, it is also a tangible
reality in the sense that every entity in cyberspace, and cyberspace
itself, is somehow animate. The Platonic difficulties of this can
perhaps be made most clear in the following example. Societies of
mutual interest emerge spontaneously in cyberspace, as entities meet
and interact. Friendships and associations of surrogates that only
meet in virtual worlds are inevitable and have already been observed
in online networks as well as in early implementations of consensual
worlds. Suppose now that two entities meet in cyberspace and choose to
produce offspring. Assuming that behind these two entities are humans
who are in a very real sense the "souls" of these entities, we are now
faced with the question of where to find a "soul" for the offspring.
Will this be a friend of the humans in the real world? Will it be an
applicant on a waiting list to cyberspace? Will it be an artificial
intelligence? This problem may of course be avoided by forbidding such
frivolities from taking place in the first place, but the issue has
nevertheless been raised, and it bears a striking resemblance to
Plato's idea that the souls of children exist prior to their births,
and that love and friendship in the physical world were made possible
by the memory and recognition of similar relations among souls in a
metaphysical space prior to "birth."

A grand paradox is in operation here: even as we are finally
abandoning the Cartesian notion of a division of mind and body, we are
embarking on an adventure of creating a world that is the precise
embodiment of that division. For, it is quite clear that our reality
outside cyberspace is the metaphysical plane of cyberspace, that to
the body in cyberspace we are the mind, the preexisting soul. By a
strange reversal of our cultural expectations, however, it is the body
in cyberspace that is immortal, while the animating soul, housed in a
body outside cyberspace, faces mortality.

As we move farther from metaphysics to metafictions, more paradoxes
become visible. Cyberspace is a dream of escape from a mortal plane
even as it is an acknowledgment of that plane. In a comparison between
reality and cyberspace, we can make the following observation: what
nature is to us and to our creations, some technological ground is to
cyberspace; what we call "nature" is simply the technological and
theoretical field that we operate in.

Consider the problem of a cellular automaton sufficiently complex to
exhibit not only the characteristics of life, but even a degree of
selfconsciousness and intelligence, and a scientific and ontological
curiosity. What could such a creature be able to discover about its
"reality"? The automaton could initially observe its own behavior and
the behavior of other inhabitants of the cellular universe within
which it finds itself; it could observe personal and social patterns;
but, like the creatures in Flatland, it would have no knowledge of
higher dimensions. It would be unable to discern the cellular grid
itself, the field, within which it existed; but even if it managed to
infer that its universes consisted of an array of cells whose values
were assigned according to certain laws, it would not be able to
comprehend the technology by which it was "displayed," nor the
intellectual and physical structures behind that technology, and
certainly not the "motives," if any existed, behind the existence of
any of the above, not for any theological or mystical reason, but
simply because those aspects of its existence were orthogonal to its
experience and because the grain of that reality was finer than its
own.

Similarly, from the perspective of an entity within cyberspace, the
laws of physics of cyberspace are layered; the uppermost, and most
visible, are the laws implicit in the software of cyberspace and the
interactions thus permitted; at a second level are those laws that
constitute the conceptual structure of cyberspace; next, more deeply
hidden, are those laws that pertain to the hardware that cyberspace
operates on; finally, are the laws that pertain to the physics within
which the hardware is operating.

Far, far removed, are the motives and preconceptions of those who set
up the cyberspace itself.

Here cyberspace leads us to question our own attempts to comprehend
what is "real." Not only are we limited by the grain we are built
upon, but even more so by the possibility of the existence of
processes that are orthogonal to our experience. For, it is evident,
such processes would not be bound to have local causes and local
effects in our realm, though they might in their own frame of
reference. The effects of processes orthogonal to our experience could
be nonlocal at any given time. No connection need exist between one
event and another, as far as we were concerned, and yet distant
phenomena could be tightly bound by laws beyond our comprehension.
This possibility parallels strongly the possibility of "quantum
nonlocality" explored by physicists (Davies 1989).

It is sufficient for our purposes here to note how nature can be seen
as the impenetrable field about which and within which we construct
our hopeful fictions. We can then draw the parallel between natural
and artificial fields. Already we have seen how artificial realities
raise questions about our reality. The artifacts we make within our
reality have equally much to say about what we build in artificial
worlds, how, to what purpose, and for what reasons.

>From Poetics to Architecture

We have examined various aspects of cyberspace from a viewpoint that
stresses the power of poetic language over ordinary, reductive
language. Poetic language is language in the process of making and is
best studied by close examination of poetic artifacts, or, better yet,
by making poetic artifacts.

The transition from real space to cyberspace, from prose to poetry,
from fact to fiction, from static to dynamic, from passive to active,
from the fixed in all its forms to the fluid in its everchanging
countenance, is best understood by examining that human effort that
combines science and art, the worldly and the spiritual, the
contingent and the permanent: architecture.

Even as cyberspace represents the acceptance of the body in the realm
of the mind, it attempts to escape the mortal plane by allowing
everything to be converted to a common currency of exchange.
Architecture, especially visionary architecture, the architecture of
the excess of possibility, represents the manifestation of the mind in
the realm of the body, but it also attempts to escape the confines of
a limiting reality. The story of both these efforts is illuminating,
and in both directions. Cyberspace, as a world of our creation, makes
us contemplate the possibility that the reality we exist in is already
a sort of "cyberspace," and the difficulties we would have in
understanding what is real if such were the case. Architecture, in its
strategies for dealing with a constraining reality suggests ways in
which the limitations of a fictional reality may be surmounted.

Architecture, most fundamentally, is the art of space. There are three
fundamental requirements for the perception of space: reference,
delimitation, and modulation. If any one is absent, space is
indistinguishable from nonspace, being from nothingness. This, of
course, is the fundamental observation of categorical relativity. This
suggests that cyberspace does not exist until a distance can be
perceived between subject and boundary, that is to say, until it is
delimited and modulated.

A space modulated so as to allow a subject to observe it but not to
inhabit it is usually called sculpture. A space modulated in a way
that allows a subject to enter and inhabit it is called architecture.
Clearly, these categories overlap a great deal: architecture is
sculptural, and sculpture can be inhabited.

We can now draw an association between sculpture and the manner in
which we are accustomed to interacting with computers. The interface
is a modulated information space that remains external to us, though
we may create elaborate spatial visualizations of its inner structure
in our minds. Cyberspace, on the other hand, is intrinsically about a
space that we enter. To the extent that this space is wholly
artificial, even if it occasionally looks "natural, " it is a
modulated space, an architectural space. But more than asserting that
there is architecture within cyberspace, it is more appropriate to say
that cyberspace cannot exist without architecture, cyberspace is
architecture, albeit of a new kind, itself long dreamed of.


*****

This file and others may also be viewed by telnet to our guest
account.

Type "telnet gopher.hs.jhu.edu" and type "guest" at the login prompt.
Type "gopher" to start the Gopher Client, and change directory to

This document may be copied freely, provided it is reproduced complete
and includes this bibliographic footer.  Excerptions and quotations
from this document must give a full citation including the document's
Title, its Author, and this source: the Department of the History of
Science, Johns Hopkins University.

All rights are reserved exclusively by the Author.

*****

Editors @ *Scientists on Disk*

447 Gilman Hall Johns Hopkins University Baltimore, Maryland 21218

-> editors@servo.hs.jhu.edu