Pedagogy vs. Competition in Higher Education Distance Learning
The Management School, The University of Edinburgh
50 George Square, Edinburgh EH8 9JY United Kingdom
Tel: +44 131 650 3817
Fax: +44 131 668 3053
* Manuscript received September 22, 1999; revised December 21, 1999
The Higher Education market in Europe might have been described as ‘mature’ during the 19th century, characterised by a relatively slowly changing number of suppliers and a relatively static number of students. However, as Europe has moved from agriculturally based economies in the early 1800s, through an industrial revolution and now increasingly towards a knowledge-based economy (Beniger, 1986), the role of Higher Education has become increasingly important for an increasing proportion of the community. The changing nature of work also means that for many of these ‘students’ full-time attendance at an institution of higher education is not an economic option and hence more flexible modes of delivery are required.
In the U.K. this increased demand for higher education under the constraint of limited budgets has created pressure for reform in the delivery of educational services which successive working party and government reports have noted (MacFarlane, 1992; Dearing, 1997; and Garrick, 1997), should take advantage of the ‘convergence’ in computing and communications technologies.
Investment in these technologies can improve the efficiency with which existing educational services are provided and with careful thought, the effectiveness with which education as a whole is delivered. In the short term, the impact on the institution will be registered on the cost side of the balance sheet, however in the longer term they can be viewed as providing direct access to new revenue streams within a widening educational market. The net effect on educational institutions could be significant, as although it has long been possible to attract students from a global market, these new technologies have for the first time raised the possibility of delivering a flexible educational service to that global market ‘at a distance’.
As educational establishments address these new markets with new products, some substitution will occur with existing products. Although the market is expanding as a whole, traditional market segments will shrink and new entrants with a different product portfolio and cost structure could replace some traditional providers.
This potential future has similar features to the current changes in market share experienced in the Financial Services industry between providers with a legacy investment in the ‘High Street’ and those who have moved to ‘Direct-Line’ operations. However, unlike the Financial Services industry, education providers target their customers on social as well as economic criteria. Changes in the competitive environment thus have policy concerns not only for educational institutions, but also for government, funding councils, and providers of educational products who ultimately balance economies of scale against increased substitutability of educational products.
This paper explores the potential impact of these new technologies on competition within the Higher Education market. It first considers technology’s role within the learning environment and explores how changes in technology and infrastructure have changed that environment locally and promoted the growth of a global education market. It then considers how university business systems might respond to this change and argues that technology which is used to improve the effectiveness of the educational service could also be used to provide a barrier against loss of market share. This would have the additional benefits of improving the transfer of existing educational skills into a networked environment, and of protecting income streams that support activities which address the Higher Education sector’s full range of social as well as economic objectives.
The Learning Environment Architecture
Higher Education institutions have evolved organisational structures and business processes in response to the perceived demands of the stakeholders within each institution and the markets they supply. Any changes in the technologies of production can have a much wider impact on, or be constrained by, these processes and structure and it is therefore helpful to review technology changes in the context of the ‘Learning Environment Architecture (LEA)’ (Ford et al., 1996). The LEA is shown schematically in figure 1, and is intended to represent the constituencies that are important in analysing the current state and direction of developments in the learning environment.
The LEA has a number of constituents:
External Environment - this is split into socio-economic and technological, and relates to trends in the business of education, which create pressures and opportunities for change.
Business Systems - which relate to the change drivers generated within an organisation, from Vision to Policy, and the business processes and objects which represent what an organisation actually does, and the resources and competencies required at each stage of that process.
Social Systems - this focusses on how people are organised, how they operate, and the structures and services that support them.
Technical - this focusses on the key I.T. components supporting the learning environment.
Perspectives & Qualities - this recognises that there are many stakeholders in the educational process who collectively define a value system. These values can be expressed in terms of quality dimensions against which quantifiable goals can be set and measured.
As figure 1 implies, these areas are intricately related, and so this model should not be treated as a unified policy instrument, but as a lens for providing a structured perspective on technology change within Higher Education.
The Technological Environment
The computer software and hardware industry is characterised by increasing production volumes and increasing rates of obsolescence. The number of microprocessor chips produced in 1997 alone was estimated by Gordon Moore, co-founder of Intel, to be 1017 and predicted to double each year (Jones, 1999). This incredible output has been made possible by significant investments in production technology that require short product lifecycles if those investments are to provide economic returns. Within U.K. universities, this trend has been reflected in budgetary cycles for capital investments in high-end computers (formerly referred to as ‘main frames’) being collapsed from 10 years in the 1970s to continuous review in the 1990s.
The relatively tight coupling or interdependencies between hardware, operating system and application (figure 2) means that changes in any one can be encouraged by changes in any other. For example, the MMX enhancement of the Intel Pentium processor was presented as reflecting a trend towards multimedia intensive applications, whilst new Windows operating systems have reflected the ability of processor chips to handle increasing numbers of ‘bits’ with each clock cycle. These lead to backward incompatibilities and a spiralling cost of provision in terms of new hardware, new software, and technical support.
The advent of the World Wide Web in 1990 (CERN, 1999) has been instrumental in providing an environment that is much more favourable for educational software developers. The World Wide Web is effectively an ‘open’ applications programming interface (API) that is portable across operating systems. The simultaneous availability of a platform-independent programming language, Java, which is associated with a public (and thus potentially open) specification, has created an environment where hardware and operating system obsolescence do not immediately make software obsolete too.
The fact that these applications can be run on browser software that is free for all users has contributed to the exponential growth in the user base. This combination of factors has made it possible to develop applications locally for one type of system and deliver that to a global user market which is already large, and continues to grow in both size and value.
Though the Internet represents an ideal delivery mechanism for software written for the World-Wide Web API, the exponential growth in number of users (IDC, 1999) and in speed of computing hardware has not been matched by growth in available network capacity (Clark, 1999). This means that the latency, or delays, associated with actually running applications across the Web can be orders of magnitude greater than those experienced within the local area network, with a consequently negative impact on their effectiveness and hence on their wider adoption. This issue has been addressed in a recent advance in the infrastructure available to Higher Education in Scotland, which for the first time can guarantee that network performance over a wide geographical area is equal to that experienced when running the application within the institution’s local area networks. This development, and its potential impact on the provision of educational services, will be explored in the following sections.
Infrastructure: The Scottish ATM MANs
Since 1996 there have been four Metropolitan Area Networks (MANs) operating in Scotland which have been developed expressly for the academic community (EastMAN, 1999). These MANs (figure 2) are additional to the high-speed U.K. backbone, SuperJANET. The MANs and their Wide Area Network (WAN) interconnects are based on 155 Mbps and 622 Mbps Asynchronous Transfer Mode (ATM) protocols.
ATM deployment in Scotland generally preceded the Fast and Gigabit Ethernet deployment within the Higher Education sector and was an important development in the network infrastructure for two main reasons:
These two features strongly interrelate. The availability of higher bandwidth makes it possible to run bandwidth-intensive applications, such as video-conferencing, but it is the guaranteed quality of service that will make it a success. This feature has enabled the creation of a very successful high quality ATM videoconferencing network between dedicated sites in Scotland (Martin, 1999). More importantly however, the focus on the IP protocol and investment in a common ATM infrastructure enables a much wider variety of computer systems to interact with each other, and for applications running across LANs to be delivered seamlessly across MANs. These applications will have varying requirements for bandwidth and quality of service. For example, email traditionally has a low requirement for bandwidth, and it is rare for a user to identify a qualitative difference between a message that is delivered in 1 second versus one that is delivered in 1 minute - something that is certainly not the case for video-conferencing. However, as email attachments grow in size the requirement for bandwidth also grows, and a service that takes over 10 minutes to deliver a message will be seen by many as qualitatively inferior! Hence bandwidth and QoS, whilst separable at a network level, are harder to differentiate from a user perspective and, for the purposes of this paper, where ‘bandwidth’ is discussed it should be taken to include both the raw bandwidth required and also the quality of service elements appropriate to support the application.
High bandwidth, a managed quality of service, and a common infrastructure means that the MANs can provide a high degree of interoperability between remote systems, a network optimised for the most demanding applications, and a suitability for ‘mission-critical’ applications. These have also been important factors in enabling connection of the MANs together. This effectively creates a nation-wide MAN that connects all of the 21 Universities and Higher Education Colleges in Scotland and, through its SuperJANET links, provides enhanced access to the rest of the information infrastructure of Europe and beyond.
Plans for the MANs of Scotland include integration with Colleges, Schools, Hospitals and the wider community. Such an infrastructure could have a significant impact on many organisational processes, from delivering local expertise to remote institutions, to sharing the distributed computing power of universities to create virtual supercomputers. High capacity MANs can deliver the required infrastructure to support these applications in the future, and are a valid base on which to project future scenarios.
The Socio-Economic Environment
The U.K. Higher Education market is under increasing socio-economic pressures due to increasing student numbers against a backdrop of falling revenues. A sequence of working party reports (Macfarlane, Dearing and Garrick) have all reinforced the imperative for increased use of technology in education, and as the U.K. government’s response to the Garrick report notes, this will continue to place technological capabilities as an important element of competition within the local educational market (SHEFC, 1998).
One response to a reducing budget is to improve efficiency by sharing centralised facilities. With traditional libraries this raises problems of access, however on-line resources can still be effectively shared if network access can be provided and it is of a sufficiently high quality. This trend has already been established with the creation of JANET (Joint Academic NETwork) and SuperJANET (currently an ATM based backbone core), but their bandwidth was not equally accessible to all institutions. With the advent of the academic MANs the access to high quality external networks increases, and so does the opportunity for becoming consumers of and suppliers to a ‘distributed national collection’. This trend is supported by the increasing modularisation of degree structures that makes it easier to include materials in local courses that have been developed within other institutions.
In order to encourage the Higher Education sector in the U.K. to make use of this infrastructure a number of initiatives have taken place, the most recent being the ‘Use of MANs Initiative’ (Ritchie, 1999), which provided funds for development of educational facilities and applications for use within and between academic institutions. This funding was used by the Business and Management departments of the three Edinburgh Universities, The University of Edinburgh, Heriot-Watt University and Napier University to develop a multimedia library of applications which addressed generic aspects of the Business and Management curriculum. The installation of a server on the Edinburgh and Stirling MAN (EaStMAN) provided a facility for these applications to be developed and shared without duplicating the hardware resources required, again echoing the library theme. This was called the Omni.bus project, and provided an opportunity for exploring educational processes in a networked environment.
The Technical System
Projects like Omni.bus are best categorised under the LEA model as making a contribution to the technical system of an educational institution. If educational applications are going to replicate local services ‘at a distance’ then it is critical that the user interface is capable of supporting a full range of teaching and learning modes. Laurillard (1993) defines these as:
Discursive - supports discussion between students, or between student and teacher
Adaptive - a capability for logging information about a student’s performance to determine subsequent teaching activities
Interactive - an ability of the system to change its behaviour according to student input
Reflective - allows student to reflect on what they know, or on what they have experienced
Resource-based - self-directed learning using structured resources
Whilst it is clear that any or all of this functionality can be supported by the Web API, applications which support the resource-based, reflective and interactive modes may only require the application to be run on a local computer. The delivery mechanism can thus be either the Internet, or a physical medium such as a CD-ROM.
It is the discursive and adaptive modes that can only be delivered across a network. They require interactions to be mediated by communication between the Web server and the local computer(s), and this introduces a dependence on the speed (bandwidth) of the network. The discursive mode is particularly dependent on bandwidth, for example if full video-conferencing is required between a number of students.
In the Omni.bus project a number of applications were implemented that reflected this range of modes. These ranged from a structured collection of Web resources matched to the curriculum of undergraduate and post-graduate courses, known as the Business Information Resource, to configuring the server to act as a router for the MBONE (MICE, 1999) world-wide video-conferencing network.
Placing these applications next to the list of teaching and learning modes and delivery mechanisms in figure 4 illustrates that a relationship can be established between bandwidth and pedagogy. This relationship is by no means invariant. It will change as compression techniques improve, and processing performed locally is used to convey an impression of interactivity. This was demonstrated by Sir John Daniel, Vice Chancellor of the Open University in his transatlantic address of the International Distance Learning Conference in 1997. His presentation started with a live video link and was then switched to an Internet link where an impression of ‘presence’ was retained by substituting his live image for an avatar (Daniel, 1998).
Despite this compelling demonstration, it remains true that the minimum requirement of any conferencing interaction is that it is possible to hear each participant. Performing this over the Internet, or even locally using the internet (IP) protocol alone, becomes difficult when operating under conditions of high network loading when there is no preferential treatment of application data, and is not currently equivalent in performance to a telephone call. Many companies are consequently investing in ATM-based networks over which IP traffic that is sensitive to delays (such as audio) can be carried at a defined quality of service in order to achieve satisfactory performance (Mann, 1999). If telephone-quality video-conferencing is still viewed as an ambitious achievement over the Internet, then it is still reasonable to assert that pedagogy has a strong dependence on bandwidth, and hence on delivery mechanism.
It is clear therefore, that bandwidth remains the major limiting factor in the variety of learning environments that can be supported at a distance. It also remains generally true, that as the physical distance over which communication takes place increases, the bandwidth available to each user decreases (figure 5). This is a natural consequence of the Internet’s resilient method of routing traffic through a series of interconnected networks, with the maximum speed of the interaction being limited by the slowest or most heavily loaded intermediary. In general this means that the most bandwidth hungry applications are run over internal or private networks rather than across public networks. Paradoxically however, with the advent of the Scottish MANs and guaranteed performance, it became possible to run faster applications across the metropolitan area than across the ‘legacy’ 10 Mbps networks that currently pervade most HE institutions.
This was certainly the case in the Management School at Edinburgh, and in recognition of the potential limitation this would place on discursive and adaptive learning environments, steps were taken to replace the LAN.
This resulted in the installation of the ATM sub-network shown in figure 6. This facility gives a number of operational benefits by eliminating any local bottlenecks between application server and application user, and provides enough bandwidth to contemplate network intensive multi-user activity such as video-conferencing sessions centrally managed by the server. The link with EaStMAN and the Scottish Cross Connect allows full advantage to be taken of applications running on the Omni.bus server by other Scottish Universities. The link to the U.K.’s academic backbone, SuperJANET, allows the scope of potential interactions to include the whole of the U.K. without passing through any network running at less than 155 Mbps.
Business Systems - Distributed vs. Integrated Models
Higher Education institutions that are now starting to view their markets as global in both recruitment of students and delivery of educational services have to decide how to respond to and develop these markets. Though new technology has reduced the difficulties of servicing a geographically dispersed student body, the business models adopted often fail to take advantage of any new opportunities offered by these technologies. Instead they rely principally on models established for ‘correspondence’ courses, in which self-contained materials are delivered to the student, on-line support is asynchronous (with letters being directly replaced by e.mail), and direct contact intermittent.
The stand-alone nature of the resulting product along with the current move within the U.K. towards nationally accredited curricula (QAA, 1999) and increasingly standardised and modularised courses, will ease substitution between internally-produced modules and those produced by other organisations. These organisations may be other educational institutions collaborating in the provision of a shared collection, or perhaps educational publishers, who can offer economies of scale. This should not only reduce the cost of the materials themselves, but the relatively low level and asynchronous nature of contact between staff and students means that students can be supervised across time zones without any increase in staff complement, giving global reach at a low marginal cost.
This ‘distributed’ model therefore has institutional appeal both as a ‘cost-efficient’ rationalisation of internal processes, focussing staff effort where they add most ‘value’, and as a revenue generating re-packaging of existing materials as ‘distance learning’ products. However, as the preceding section has argued, this model reduces the number of learning modes that can be supported, and hence has a direct impact on pedagogy. This not only limits the potential effectiveness of the learning materials, it also reduces the number of qualities on which a student decides to purchase one institution’s offerings versus another. In doing so it establishes a market which is particularly hard to protect as educational products become harder to differentiate between on the basis of either their content or method of delivery. Ultimately the buying criteria could reduce to a trade-off between the reputation of the institution and the cost of the course. Under such circumstances, it not hard to predict that if a distance-learning MBA from Harvard Business School was priced at the same level as a distance-learning MBA from a nearby college, then only one institution would be running a profitable MBA programme! In a truly global market, such a ‘fully distributed’ approach to the market would leave room for few suppliers.
Though the fully distributed ‘distance learning’ model is popular it is not the only business vision which can be supported by information technology. The Omni.bus project reflects a strategy of using information technology to bring the Teacher - Student and Student - Student interactions ‘nearer’ to each other, by increasing the flow of information between the two and increasing the opportunities for using that information to structure and support a wider range of learning environments. This reduces the ‘pedagogical gap’ between local and remote students, allowing a wider range of learning styles to be accommodated and the learning environment to be adapted through on-line synchronous contact with the teacher.
The wider range of interactions supported under this ‘integrated’ model not only improves the educational experience from the student’s perspective, it also allows the experience of existing staff to be more readily applied to a networked environment. For example, a lecturer who is experienced in coaching students with special needs on a one-to-one basis is more likely to be able to apply that experience in a private video-conference, than to be able to program it into an application delivered on a CD-ROM. Even with large groups, a lecture theatre that is also a video-conferencing suite, enables a wider student body to participate across the network, with less need to modify existing lecture material, and lecturer expertise, for that market.
There is still a role for third parties, such as publishers, under a distributed model to use economies of scale to provide learning materials of high quality at a cost that no institution can match. However the definition of quality needs to reflect the level of support for the above range of interactions and adaptations. Educational ‘content’ products that support different delivery methods, are ‘visible’ to both the student and the lecturer across the network and can be adapted for different learning environments by that lecturer. Such products will have a market almost regardless of the chosen strategy of the educational institution.
Business Systems - Pedagogy, Communities and Overheads
The ‘integrated’ model is undoubtedly superior to the ‘fully distributed’ model in terms of improving the learning experience for any remote student. It also helps establish a wider range of qualities on which to sell an educational product and hence makes a revenue stream from this activity easier to protect. However the economics associated with providing a ‘fully distributed’ product, an ‘integrated’ product, or no distance learning product at all, have to be viewed in terms of the needs of the existing communities represented in the Higher Education sector.
The approach taken to creating a teaching and learning environment has a clear dependency on the subject being taught, the characteristics of the student community being addressed, and the requirements of individual students. Differences between subjects are recognised at an aggregate level by the funding councils who differentiate between the cost of provision for ‘Arts’ versus ‘Science’ and by institutions themselves through differential allocation by subject area of nearly every factor of production. Differences in community include full vs. part-time students and undergraduate vs. post-graduate vs. adult returner. The requirements of the individual student will range from sensitivity to a particular learning style, to the obvious constraints imposed by physical impairment.
Institutions have traditionally reacted to these differences by selecting from a range of didactic approaches, however it is clear that the ability for computer-mediated education to substitute for these traditional approaches is not uniform across the whole didactic range, nor is it by discipline, by community, or by student. In these circumstances, an institution might be encouraged to choose products for translation into computer-mediated formats on the basis of their ease of translation, to keep the capital investment required down, and commercial value of the resulting qualification, to maximise the resultant revenue.
These criteria would lead to the selection of existing courses that are taught in large classes, to students who are prepared to pay above average fees. Though many courses might be described as such, it is clear that they will all make a significant contribution to the overheads of the institution, and hence represent an important income stream to be protected. The act of translation is not only a capital intensive investment in new factors of production, it produces a substitute product for the existing course that could reduce its enrolment, and hence its net contribution to overheads. In a commercial environment, this would lead a company to segment the market and potentially decide to move its total operations from the traditional factors of production into the new factors of production, at which point it can offset the loss of one market against a saving in its cost structure. When a low cost, fully-distributed product is viewed against a traditional product with its high staff costs, an analogy might be drawn with Financial institutions that are moving their investments from a legacy ‘High-Street’ presence to a ‘Direct-Line’ presence, with a resulting saving in overheads. However, unlike the Financial Services industry, the loss of educational customers within one institution cannot be directly traded against a reduction in overhead, because many of the staff and facilities are still required to address the subjects that cannot be tackled ‘on-line’ and to service the communities that are tackled not for economic reasons, but for social ones.
Hence an institution investing in ‘on-line’ education will find its fixed costs unavoidably larger in the short to medium term. This makes the business model adopted for on-line education doubly important, as distributed and integrated approaches will both add to fixed costs, but only one is likely to lead to the sustainable competitive advantage required to service those costs.
Discussion and Conclusions
The Garrick report on the future of Higher Education notes that the use made of computing and communications technology will be an important factor in competition between local educational providers (Garrick, 1997; SHEFC, 1998). A general convergence in these technologies and the specific advent of the World-Wide Web and Java has made this increasingly true about competition within a global educational market.
Though there are relatively few institutions in the U.K. that deliver educational products to a world-wide student base, all U.K. higher education establishments should view the globally available products as potential substitutes for their own. These may erode their market share in activities that make a critical contribution to the fixed cost structure of their institutions. If institutions should choose to respond to this threat, or perceived opportunity, with distance-learning products that follow a ‘fully distributed’ approach, they will reduce their ability to differentiate their own products from those of the competition, and promote a competitive environment that will leave room for few global suppliers.
This paper has considered an alternative, ‘integrated’, approach in which a greater dependency on network bandwidth in the delivery of educational products allows a much wider range of learning modes to be supported. The availability of higher bandwidth for the students also enables them to interact more directly with teachers and other students, and hence allows an easier translation of existing teaching expertise and expectations into the networked educational environment. These factors should provide a much better educational experience for the student, and if the pedagogical benefits can be expressed as values that can be communicated to students, then their intrinsic dependence on network bandwidth creates a competitive barrier that cannot easily be overcome by distance learning products from other institutions. This is illustrated in figure 7, where it is clear that developments such as the MANs could play a significant role, as they provide a potential mechanism for high-speed access to individual homes through third parties, such as the cable television providers, or through new ‘copper-based’ technologies such as xDSL, which offers speeds up to 52Mbps across ATM network architectures down traditional residential telephone wires (ADSL, 1999).
For institutions about to enter this market, a critical concern will be the sustainability of this barrier to competition. Current investments in global telecommunications infrastructure, such as the 1.9 Terabit per second Project Oxygen with landing points in every continent except Antarctica (Bogler, 1999), represent a significant increase in the physical bandwidth available to Internet transactions. This might be expected to bring up the average bandwidth available to users on the Internet, however recent reports from IDC have claimed an increasing divergence between the demand for bandwidth and its provision in wide area telecommunications (Clark, 1999). They observe that the bandwidth being generated and used on local area networks is following Moore’s Law (Jones, 1999) by doubling every 18 months, but demand for wide area telecommunications bandwidth is doubling every 6 months. This observation, if true, suggests that the impact of increasing numbers of users will outweigh increases in physical bandwidth, actually causing the average bandwidth available per user across the Internet to fall. Bandwidth therefore looks set to remain a scarce resource for the immediate future and this might explain why the world’s largest supplier of electricity and natural gas, Enron, recently revealed plans to establish a market in which bandwidth will be traded as a commodity (De La Rosa, 1999). A barrier to competition established on the basis of imbalances in the provision of wide area versus metropolitan area bandwidth may in future become practically easier to overcome through this developing market, but it will then move from being a physical barrier to an economic one.
Though it may be observed that central funding constraints and quality assurance initiatives from government have already contributed to reduced differentiation of educational products by promoting increased uniformity and substitutability of educational modules in order to share development costs between institutions. Centrally funded facilities such as the U.K. MANs do provide a method for achieving these objectives without promoting the creation of educational products that can be just as easily administered by remote global competitors. It is still possible to achieve economies of scale in the creation of educational products, however, rather than converting a text-book into a stand-alone interactive application delivered on a CD-ROM, educational suppliers could provide products that allow institutions to customise the learning environment. For example, if publishers provide application environments that directly support different learning styles, and allow all on-line teaching, learning, assessment and further resource support to be provided by local institutions, then product differentiation will exist on a local scale as well as a global one.
The above approach might be criticised for helping to preserve the status quo, however it does so in a way that improves the educational experience for remote students, develops flexibility in the provision of education as a whole, and helps individual institutions to balance economic objectives with social ones. This will also help to preserve a physical infrastructure that provides a much wider set of benefits for the U.K. as a whole. Taking higher education alone, it may be observed that even the operation of the Open University, one of the heaviest users of bandwidth across all public media, would suffer if access to the facilities and staff of institutions with traditional overhead structures was reduced.
Whether institutions with a global reputation for excellence in traditionally delivered courses will choose to enter the global market with a ‘fully distributed’ distance learning product remains to be seen. However, the recent U.S.-U.K. tie-up between Stanford, the University of Chicago, Columbia and Carnegie Mellon with the London School of Economics to "distribute management education over the web" (Bradshaw, 1999) makes this scenario worthy of careful evaluation by all suppliers to this market.
The Omni.bus project referred to in this paper involved a large number of people at The University of Edinburgh. The development team included Stephen Kitt, Ian McGillivray, Eric Watkins, Nikos Massios, Mark Mackenzie and Ian Graham. Support is also gratefully acknowledged from Edinburgh’s Unix support team (Graeme Wood and Paul Haldane) and the Network Services Team, Scott Currie, Bill Byers and especially George Howat who commented extensively on this paper.
The project would not have been possible without funding provided by the Scottish Higher Education Funding Council (SHEFC) through the Use of the Metropolitan Area Networks Initiative (UMI), The University of Edinburgh Management School and the Faculty of Law & Social Sciences. The author is also grateful to Graham Lovell of Sun Microsystems for arranging support under their Academic Equipment Grants Programme.