Articles

THE COMPUTABLE CITY

Paper By

Michael Batty

Centre for Advanced Spatial Analysis,
University College London, WC1E 6BT, UK

ABSTRACT

By the year 2050, everything around us will be some form of computer. Already, we are seeing a massive convergence of communications and computers through various forms of media. Computerized highways are in prospect and smart buildings are almost upon us. As planners we are accus-tomed to using computers to advance our science and art but it would appear that the city itself is turning into a constellation of computers. The implications of this for city planning are enormous. New data sources emerging in real time, and software to understand many elements of the working of cities such as simulation games and GIS are now widespread. The juxtaposition of media that a generation ago we would have been regarded as unthinkable is generating entirely new opportunities for understanding and planning cities. This paper raises these issues through a travelogue across the Internet. Ideas for what is becoming possible in our domain are illustrated from that latest of networking triumphs, the World Wide Web, from which we draw examples of cities in situ, in vitrio, in the abstract, in real time, and in cyberspace.

A CONTINUING REVOLUTION

This past Christmas, my wife bought a mid-range Macintosh computer from a supermarket in England for word-processing. The model on offer contained a digital television tuner and a CD player which we thought might be useful especially as I envisaged that I might learn how to make QuickTime movies if I could snatch a bit of spare time. When we got the machine home and switched it on, we experimented with the multimedia and soon discovered that not only could you record as much TV as your hard disc could contain, play it back, edit it, but you could freeze frames and immediately print them, you could load them into drawing packages and you could cannibalize them for other purposes. Even more remarkable was the fact that the tuner was also able to access the two passive information systems which have been available for almost 15 years in Britain - Ceefax and Oracle which the BBC and ITV channels run respectively. Our regular TV in the corner of the room could do none of these things and thus quite unwittingly we were already in possession of the ultimate electronic box which represents the convergence of the computer, the television, the fax and every other electronic media which the pundits are still telling us will only happen by the end of the century !

Three years ago at Sun Microsystems in Mountain View I had seen workstations with TV windows running alongside other software but I had not expected to see them so soon on a low-cost Macintosh. You cannot yet buy one of these machines in the United States and despite my best efforts, I cannot find out why. I suspect it is something to do with the way television is regulated in each country for it clear that the technology is there and in fact has been for some time. In Berkeley recently, Pravin Varaiya who is the Director of the PATH program which is developing automated highways, also told me that they had attached a couple of cheap still cameras to their Macintoshes and that software was now available on the Internet which enabled them to turn their machines into videophones. In short, it is possible to have real time video hookups on the Internet across the world essentially providing free voice and video mail. TV, fax, videophone, computer all in one, forming the ultimate communications device.

What is so remarkable about all these events and many others which are besieging us daily is not that they are technologically surprising but that the speed at which they are being developed is so astonishing (Negroponte, 1995). Computers, after all, as the pioneers such as Turing, von Neumann and others told us 50 years ago, are universal machines, but we still find this hard to absorb. Technologies which lead to sweeping changes within society usually result from innovations whose impact is largely unforeseen at the time of their creation (Hapgood, 1993). Computers provide the very best example of this phenomenon. In the immediate aftermath of their simultaneous invention in a number of places during World War 2, the conventional view amongst the industry's founders was that no more than a handful would ever be required, and these would be used solely for scientific purposes. 30 years later just prior to the development of the personal computer, IBM, Hewlett Packard and DEC - the leaders of the mainframe and minicomputer industries - all went on record as doubting that the microprocessor could ever become the basis for devices as ubiquitous as the television and the telephone. This was less than 20 years ago.

The history of computing is full of such examples. Even the greatest visionaries such as von Neumann (Macrae, 1992) who predicted that computers would revolutionize science, and Vannevar Bush (1945) who foresaw the development of multimedia, failed to see the profound impact these technologies would have on material production as well as social behavior. Not only was the path to miniaturization barely foreseen, but perhaps more important, the convergence of computers and telecommunica- tions which now represents the cutting edge of the revolution, was largely unanticipated. At present, there is much speculation that interactive television, multimedia, and networking will change behavior in ways we cannot envisage, yet the really radical notion that all our material infrastruc- ture from communications to shelter is turning into computers still seems far fetched. But within 50 years, everything around us will be some form of computer and the ways we will access this and use it to interact with each other will be through software. By the mid-twenty-first century, cars will be computers, buildings will be computers, entire cities will be computers, all wreaking profound changes on the form and functioning of our environ- ment and the ways we will seek to understand and change it. The interaction of computers across networks reflected in SunÕs slogan ÒThe computer is the network, the network is the computerÓ, is not simply leading to an order of magnitude change in computer power, as if that is not enough. It is leading to a deep qualitative change in what computers are being used for. The telephone and the television provided new media for active and passive interaction but the computer is generating a qualitative change in computational interaction which is considerably more powerful than any interactive media developed so far. The dominant paradigm for this is, of course, the network (Kelly, 1994) which is enabling many new functions and forms of behavior to be developed. Apart from the obvious possibilities of remote computing, access to information services such as data, libraries, software for computing, text processing and design etc. over the net is enabling entirely new ways of working as well as new forms of work. Such opportunities generate different forms of interaction which both substitute for and complement one another in a myriad of unanticipated ways, thus leading to new forms of social behavior. This, in effect, is but one of many indicators that social structure and behavior is becoming ever more complex, with important implications for our understanding and planning of cities and regions.

This conference is mainly about the uses of computers in understanding and planning cities, and until quite recently, this was the predominant role for computers in planning. But it is increasingly clear that computers are now changing the very systems that we are seeking to understand using the same computers, and this in itself is generating important consequences for how we use computers in planning, consequences which have barely been raised to date. We will explore this conundrum here, suggesting that it is important to examine the ways in which computers are changing the methods for understanding as well as changing the structure and dynamics of the city itself. This is the phenomenon that we will refer to as The Computable City. In the past, the ways we have used computers to understand and help in planning the future city have been quite different from our concern for how computers and information technologies change the city, but with the recent proliferation of local and global information networks, these two domains have begun to collide and coincide. This is best seen in the fact that computer hardware and networks as well as software used traditionally to understand the city by professional researchers and planners are now being used by a variety of interests and agents whose concern is somewhat different, involving the very actions and behavior that we are usually see as composing the fabric and structures to be understood and planned for. Of course, this coincidence of those planning and those planned for is hardly new. Social scientists have, for long, pondered the intersection of interests between planners and the planned, and the degree to which an appropriate understanding of urban behavior can be seen in terms of one domain rather than the other. But in the context of computation and particularly the use of networks and computers, this difference and coincidence takes on a new significance which we will begin, albeit in a rather preliminary and somewhat cavalier way, to discuss here.

In the rest of this paper, we will first outline a simple framework for reconciling our knowledge about how we use computers to begin to plan for cities that are composed of these very same machines. The network paradigm that is emerging is helpful to us here and we will illustrate several examples of the ways in which computation is changing our perceptions of the city using the latest Internet fashion of the moment, the World Wide Web (WWW), to present our thoughts. We explore how we can use computers to watch real cities, how we can simulate abstractions of cities, how we can use computers to learn about cities, and how we can live and work within virtual cities. Our examples simply give us a glimpse of the kind of computable city which is emerging, one which is changing daily and which will have changed substantially by the time this paper is formally presented.

COMPUTERS FOR PLANNING : PLANNING FOR COMPUTERS

Most of our applications of computers for understanding and planning cities have been for purposes of analysis, modeling and design, for storing data, and perhaps more recently for communicating data and ideas. Only very recently has the notion that computers might be more than simply a means for a better understanding and that computation might be more than simply scientific analysis become significant. Over 30 years ago, Meier (1962) speculated upon the notion of the city as a formal mechanism for communi- cations and there has been a stream of work on the new geography of high technology production and services (Castells, 1989) but in a sense, this has remained very separate from the use of computers within the planning process. Different ideologies and interests pertain to these two domains and so far, their study has had little in common.

It is the synthesis of computers and telecommunications which threatens to change all this. The emergence of computer networks which are able to monitor systems in real time while simultaneously allowing analysis of those same systems is something which is very new, at least on a wide- spread scale. The increasing convergence of the non-routine and the routine in the same set of technologies is making possible the study and analysis of systems in a way that was not possible hitherto. Furthermore, the fact that so much routine behavior which has an impact on the city is now being ÔinfluencedÕ by electronic networks is changing the very phenomena which planners and spatial analysts have traditionally studied. It is a cliche to say that computers and telecommunications are annihilating distance as much they are changing time but there are substantial changes underway which will probably change the very notion of the city which we have grown up with. Moreover, what is going to be very clear in the future is that to study cities in any manner, it will be necessary to use diverse methods of computation which will vary from the straightforward browsing of digital data to much more sophisticated methods of simulating futures. To plan those same cities, we will have to complement our set of planning tools with those that involve the design, operation, and selection of different types of networks and information.

This is a complex prospect and we need some way of making sense of it all. First let us make a distinction between the material and nonmaterial worlds which in terms of cities we may think of as infrastructure and the way populations behave within that infrastructure respectively. The study and planning of cities has always approached this distinction with respect to how human behavior influences physical form and space although these limits have been the subject of intense debate in the last 30 years. Part of human behavior is our quest to understand and plan the city although many have argued that this activity must be treated no differently from any other type of behavior which affects the city. Nevertheless, a distinction must be made, otherwise there would be no distinct professional activity or concern. When computers were first invented 50 years ago, they gradually en- croached upon these professional concerns and by the 1960s had become significant enough to constitute a separate field of inquiry, notwithstanding the controversies such approaches implied. But only recently have computers through networks begun to dramatically affect both the infrastructure of the city as well as those other forms of behavior which planners and urban analysts view as determining spatial and social structures.

In short, computers which were once thought of as solely being instruments for a better understanding, for science, are rapidly becoming part of the infrastructure itself, controlling new infrastructure through their software, influencing the use of that infrastructure, and thus affecting space and location. In one view, the line between computers being used to aid our understanding of cities and their being used to operate and control cities has not only become blurred but has virtually dissolved. In another sense, computers are becoming increasingly important everywhere and the asymmetry posed by their exclusive use for analysis and design in the past and their all pervasive influence in the city is now disappearing. In both cases, the implication is that computers will have to be used to understand cities which are built of computers.There will be no other way.

It is easy to make too much of these events. For some, cities will never be perceived as being formed from computers and networks, and of course, ours is only a partial view. For a long time to come, the line between computers in cities and everything else will be a clear one and it will be necessary to extend our understanding across this line. In fact, this means that our understanding of cities and their planning is more complex than ever it was a generation of more ago, for now we need to consider new forms of network as well as the traditional ones based on material physical infrastructures. Besides adding to complexity through increased opportuni- ties for interaction and new forms of organization, cities are becoming increasingly invisible to study and analysis in traditional ways as more and more electronic media appear. As we will argue later, it is very likely that the degree of complexity will soon overwhelm traditional approaches to social control and this is already seen in the massive decentralization which computers are hastening (Kelly, 1994; Resnick, 1994).

In this paper, the lens through which we are viewing the city is the computer and we will direct all subsequent discussion this way. We will, in fact, make a distinction between real cities as viewed using computers and abstract cities as simulated on computers. By real cities, we mean those elements of cities which we might study using computers which detect and control elements such as transportation networks, utilities, remote sensors, and other data. In contrast, the study of abstract cities involves the use of computers for analysis and modeling, removed somewhat from the routine and nonroutine functioning of cities in which control is the modus operandi. Between these two poles lies the use of computers for interacting within cities, for buying and selling, for leisure, for work and for a host of activities that were previously carried out without computers. This is the virtual city, it is cyberspace, the world of email, the net, and such like. Connected to this, is the world of structured learning which in terms of our understanding of cities involves the real, the abstract and the virtual in diverse combinations.

Our perspective is hardly well-formed, nor is it comprehensive but our purpose here is to raise questions and illustrate examples of what is happening. To this end, we will discuss these different views of the computable city in terms of how we might use computers to watch real cities using online data, to simulate abstractions of cities perhaps using the same online data, to learn about cities using both these media, and lastly to explore the new world of the virtual city in which traditional forms of interaction are being both substituted for and complemented by the new media.

WATCHING REAL CITIES

From the beginning in the 1950s, computers have been used to store and process data of use to city planners. Before the dawn of the digital computer, as far back as Herman Hollerith, computational devices were used for such data. Indeed, the original momentum for widespread digital computation came from the development of information systems for transactions processing. Although most digital data for cities until quite recently has been available for nonroutine functions such as that collected by the population censuses, routine data is now becoming available. Real time data from industrial concerns has been available for a while as has financial data but now data pertaining to functions relevant to physical and spatial problems is online. Satellite data is routinely available at the global level over the Internet as illustrated by the numerous online weather reports but the most significant types of data pertain to traffic and movement. Real time traffic volumes and speeds, levels of congestion and accident hot spots are available at an increasing number of sites if one has the technology to access this, thus making possible analysis which is close to real time. Three examples which have appeared on the WWW within the last 6 months make the point.

Our first is available from the California Department of Transportation - CALTRANS - and associated sites involved in monitoring traffic in the State. Real time reporting of traffic speed and volumes in tabular or map form are available for a cluster of cities. We illustrate the case of San Diego in Figure 1 (http://www.scubed.com:8001/caltrans/sd/big_map.shtml)where we show traffic volumes on the main highways which change every five minutes as new data is fed across the net to the Web. The data is probably still only of novelty value and it is unlikely that much real time analysis can yet be done with this. But the fact that it is possible at all and that it is in the public domain is likely to change routine decision- making concerning trips while enabling longer term nonroutine analysis of congestion and capacity. Of course, with the development of navigational instrumentation in automobiles, this kind of data might be immediately useful and in time it is this which will be routinely available for all major highways and for all vehicles which can make use of it. Business and government, of course, have long used such data but this does begin to reveal that the problem will be how to use the data. The problem is no longer whether the data can be collected or exists but how to act on it. As with most remotely sensed data, the volumes are so great that most of it is lost and there is an urgent task to develop analytical software which can use it to best purpose.

Our second example is more obvious but equally innovative. The Lawrence Berkeley Lab at the University of California have the entire 1990 US Census of Population online. It is possible for any user in the world who has access to an Internet account (and the cost for connection can be as low as $3 per week for a commercial service !) to download any element from this data set to a remote machine at a variety of spatial scales from the block group up to the entire USA. Reportedly, the LBL will do the same for the 1980, 1970 and 1960 Censuses in the near future, thus providing a remarkable resource which throws into confusion much of the debate about data costing and value added. However, the problem is not simply one of data volumes but of interpretation. Software is required and this means some form of GIS or nonspatial information system which can enable the data to be viewed either visually or numerically. Figure 2 (http://cedr.lbl.gov/cdrom/lookup/)shows the typical layout for this home page which outlines the service for 1990 data. Downloading can be slow especially at peak times on the information highway. The US Bureau of the Census and the US Geological Service (USGS) also have services which enable spatial data such as TIGER files, land use cover and digital terrain models to be downloaded for many areas within the US.

Our last example pertaining to data involves a network of pointers called GeoWeb which acts as a user-friendly guide to geographical data which is online. Devised by Brandon Plewe (1995) at SUNY-Buffalo as a pilot model for building spatial data infrastructure, it enables one to browse through a veritable treasure trove of spatial data available over the Internet. It is in fact more a resource to learn about what data is available and what software can be used in its access and analysis rather than a means for downloading the raw material. The home page is shown in Figure 3 (http://wings.buffalo.edu/geoweb/). As such, GeoWeb is an excellent illustration of the way the World Wide Web (which is a subset of resources available over the Internet) can be accessed and searched. The screens shown here are from the browser called Mosaic which enables different documents to be accessed using hypertext - hotlinks which point to other documents which summarize resources at different locations on the Web. By clicking on these hotlinks, the searcher moves across the net to the physical location of the document and resource, thus building up a web of interactions. This, of course, is the way the net can be explored but it also the way new webs can built across the net. The pointers soon explode as Figure 4, which is a map of four layers within GeoWeb

implies. This is the kind of complexity that is characteristic of the information society and which makes traditional responses to its study quite inadequate.

SIMULATING ABSTRACTIONS OF CITIES

The traditional use of computers in planning has been for analysis and forecasting from a data or a modeling perspective. All we can do here is to point to the kind of abstractions that are being developed as computation creeps out into society at large. The broad domain of knowledge in this area is encompassed by many of the papers at this meeting but the emergence of networking and online data resources is changing the field dramatically. Evidence of this comes from the use of serious computer games such as SimCity. Reportedly, SimCity was the most popular-selling computer game in the UK last Christmas (Macmillan, 1995), with many more people being exposed to the game than there are professionals concerned with the study and planning of cities. This contrast is even starker in that straw poles reveal that a majority of professionals in this area have never even heard of the game. The implications for who understands cities best and who should be involved in their planning are dramatic.

On the net, some interesting modeling applications are emerging. There are countless GIS tutorials but descriptive simulations, some with a Òhands-onÓ feel, are becoming available. We will examine two. First at USGS at NASA Ames in Silicon Valley, a group led by Len Gaydos have put together a simulation of urban growth in the San Francisco Bay Area as part of a wider project dealing with the human consequences of land use change. Urban development in the Bay Area from 1820 has been morphed together from old maps and since 1972 from satellite imagery. Some stills from this development are shown in Figure 5 (http://128.102.124.15/usgs/HILTStart/) but the real excitement begins when you click on the pointer beneath these maps which starts the mpeg movie player generating an animation of urban growth in the Bay Area. From this document, you can extract the data for this simulation and also explore a simple model of this growth developed by Keith Clarke and Lee de Cola based on cellular automata (CA) (Kirtland et al., 1994). As yet, you cannot run this model online but there are several documents on the Web where users who log on from remote locations can run applications if the software is mounted on their own machines. You can access Shane Murnion's ARC/INFO tutorial(http://www.geog.buffalo.edu/) which tells you how to construct an application and echo the results on the Web if you have the software online. By the time I present this paper at the meeting, we hope to have a similar simulation of the Buffalo region up and running which we are constructing in Imagine and ARC/INFO as a prelude to the application of various fractal and CA models of urban growth (Batty and Longley, 1994; Batty and Xie, 1994).

Our second example is also taken from urban morphology although the analysis contained therein is rather different. Bill Hillier and his group at University College London have been applying ideas of space syntax to urban form for a decade or more, as initially illustrated in Hillier and Hanson (1984). These are models of urban structure in which space is explored which exists between buildings rather than within them. In a sense, the analysis seeks to interpret how space is formed by examining the external rather than internal disposition of functions and the way the space is configured in terms of its use. They have developed many applications for many cities and some of these are illustrated in their WWW server whose home page is presented in Figure 6 (http://doric.bart.ucl.ac.uk/index.html). Perhaps the most useful aspect of their pages so far is the easy-to-use tutorial concerning space syntax which they are developing and which provides a clear analysis of the modeling techniques in use.

Space syntax of this kind requires very elaborate radial route networks of cities, usually coded according to criterion associated with the definition of exterior space. The nearest thing to a world wide resource containing this type of data is the line feature data called the TIGER files which enable the geometry of the entire US to be reconstructed at Census block level. If you go into GeoWeb and point your way to the Census TIGER server (http://www.census.gov/tiger/tiger.html/) you will find a useful summary of these resources with some examples of this data online. How long will it be before the kind of space syntax analysis developed by Hillier can be linked to data in the TIGER files ? And will this be possible across the Web ? Much will depend on how data and software is made available and to whom but the prospect of other users engaging in such analysis is almost upon us. Although I cannot point to sources, I am fairly sure that some simple GIS packages exist on the Web which can already be used to analyze and visualize the kind of Census data which exists at LBL.

LEARNING ABOUT CITIES

In one sense, learning cuts across all of the applications we have intro- duced, for the very activity of accessing computer networks and exploring resources is so new that it will represent a basic learning experience for while yet. If you log onto the Buffalo homepage, then you will find that besides providing a catalog to the facilities, faculty and staff at the Geography Department and the NCGIA, there are a various resources connected to instruction. We have already noted structured tutorials such as Bill HillierÕs space syntax and Shane MurnionÕs ARC/INFO document, but on the Buffalo homepage, three courses are being developed with their own homepages in which students and faculty are thinking aloud, online, so-to- speak. We will examine my own course which is entitled The Geography of the Information Society and explain what we are doing. Figure 7 (http://www.geog.buffalo.edu/Geo666/) shows the homepage. As the title suggests, the course provides a wide ranging introduction to exactly the material of this paper - The Computable City by any other name (and this self-referential style which dominates this paper is in fact quite characteristicof the complexity which we are all facing with respect to the rapid computerization of society). The course is structured into four parts: first a review of the emergence of postindustrial society, and the growth of information networks, computers, software, orgware and so on; second an outline of the geography of high tech manufacturing, services and globalization; third, the development of information infrastructure in cities largely involving telecommunication but also smart buildings and electronic highways; and finally, the emergence of cyberspace and virtual communi- ties.

If you click on our homepage, you will see a logo (Figure 7), an outline of the course, and a list of participants: myself and students whose names when clicked reveal their own homepages on which they can experiment with anything they like - biographical details, pointers elsewhere revealing their preferences, papers they have written and so on. Also presented is a course schedule which if clicked provides a set of hierarchical pointers to papers written by each student for particular sessions, as well as pointers to the rest of the world. For example, in the section on technopolis and science parks, pointers are given to places like Tsukuba science city, Blacksburg Electronic Village, Research Triangle Park, Computer Manufacturing in Austin TX, and so on. Finally there are pointers to what we call Neat Cyberspace, examples of which we will give in the next section. The whole idea of this exercise is to provide us with ways not only to present our work to others and get feedback - you can send us email about this from within the Web pages - but also pointers to other parts of the Web where others have similar concerns to ours. Of course, to pretend that this is all there is to the course is fanciful and dangerous and there are many topic areas even within computer-oriented planning that would not be appropriately studied through the Web. But at least, this gives an idea of what can be done even though it is all Òunder constructionÓ.

Our second example is perhaps already more elaborate. Bill Mitchell and Mitch Kapor taught a course entitled Digital Communities at MIT last fall. Their homepage is illustrated in Figure 8 (http://alberti.mit.edu/arch/4.207/homepage.html). The design is similar to that which we have adopted although references to literature are more extensive and the document is more complete in that the course has already been taught and all the student papers prepared. What is important about these attempts is that they build on one another. Our page points to theirs, but not theirs to ours as yet, but nevertheless, the way the web is woven is an all important part of the learning experience. In a way, this is the network paradigm in its purest and most archaic form. The kind of complexity that is writ- large in these kinds of perspectives on the Internet give some sense of how overwhelming is the study of the information society. How to map and chart, indeed even explore this kind of complexity is something that we have not yet faced but of course, this is the very essence of the computable city. Moreover, how all this interfaces with more traditional infrastructures and behaviors as well as more traditional approaches adds other layers to this complexity.

LIVING IN VIRTUAL CITIES

We have already glimpsed the idea of the virtual city through examples such as GeoWeb and through our pointers to the information society. There is much written about this kind of cyberspace and how the net is being populated by nerds and hackers whose very existence involves living online. But the most exciting prospect involves the emergence of services which have traditionally been associated with non-electronic access and in this penultimate section, we will provide a couple of examples. But first some words about the Internet for it is important to see this in the broader context of the network society. The Internet has become the de facto symbol of the network society in that its growth, once begun, has spun out of control. It has become the skeleton for global computer-communications interaction. Although elaborate networks proliferate in the private and public sectors, most of then are gatewayed to the Internet where the range of services is enormous and increasing at rates often in excess of 10 percent each month. It is here that cyberspace is most fully developed, with an increasing range of routine and not-so-routine functions which enable users to shop, study, work, and play in a virtual world. In fact, the idea of virtual reality is perhaps a little extreme a way of characterizing cyberspace and the net for in essence, its functions simply replace traditional modes of communication and interaction with electronic.

Two examples simply touch the edge of this virtual world. In Singapore, at the National Computer Board, they have constructed an interface to the net which they call Cyberville. This is simply a way of exploring cyberspace but it is modeled on the analogy of a small town, a so-called electronic village containing all the kinds of physical land use functions or public/pri- vate buildings which one might associate with a well-balanced community. This image is in essence the homepage of Cyberville, shown in Figure 9 accessible at http://king.ncb.gov.sg/cyber/cville.html which has become the kind of icon used in many places for accessing and organizing entry to the Web. Click on any of the functions - stock exchange, science centre, university, town hall, movie theatre and so on and you move across the net to pages that display and point you to Web sites which contain those functions. In short, Cyberville is an intelligently organized entry point to the Web for users who wish to focus on the sorts of services one can access electronically. For example, if you click on the stock market, you can eventually find yourself at MITÕs AI Lab where they have an experimental stock market analyzer, indeed if you click on any of these items. you can begin to travel the Web moving from site to site according to the structure embedded into each homepage, but directed initially by Cyberville. This is a clever way of organizing entry to such electronic spaces and someday, all of our entry points will be so configured, at least for the end user, if not for most others.

One last example illustrates the point again. We will not show the homepage but if you click on this transports you to Cyberia which is an electronic shopping mall in Houston (no comment !). I have searched in vain in this and other such malls for places to actually enter credit card information and to make orders but invariably all the links are ÒdownÓ as users and owners begin to figure out not only the economics of the Web but more importantly its security. Once there however, on the Web, the variety of experiences can be overwhelming and the search can be endless. How many sites and documents there are on the Web is unknown. Plewe (1995) estimates that six months ago, over one million documents existed but this may be an underestimate for there are now 30 million or more users of the Internet. Once television becomes interactive and once video and other media can be shipped over the net on demand (for this implies that all media will be digital, hence each copy will be almost zero cost), the kind of complexity that we see here will be extensive. It is then that virtual cities will intermingle with real cities, when the abstract and the real will merge into one another, and it is then that our traditional concepts of understanding, let alone planning the city will simply disappear. This is a challenge that we must begin to address if we are to simply cope with these changes.

THE FUTURE

This meeting will be largely concerned with the use of new methods of analysis, modeling and design which are informed by computers rather than the broader kinds of activities which we have alluded to here. But there is an urgent need to generalize our debate in the light of the way cities themselves are changing, the way activities are decentralizing in both time and space and the way new network forms are dispersing and concentrating spatial activities in very different ways from the past. Those who work with computation are well placed to make important contributions to this broader debate for the use of new forms of computer and network, new digital data sources, and new software across the net in itself represents the way the city is changing. New insights will only come if new forms of computation are developed, requiring all who study the city to become immersed in its intrinsic computability.