Networked Multi-user and Multimedia Environments for Learning and Collaboration Jonathan P. Bowen The University of Reading Michael K. Houghton Entranet November 1999 Address for correspondence: Jonathan Bowen The University of Reading Department of Computer Science Whiteknights, PO Box 225 READING Berks RG6 6AY UK Email: [email protected] URL: http://www.cs.reading.ac.uk/people/jpb/

Abstract The dramatic increase in the popularity of the Internet, largely brought about by the World Wide Web, has significantly increased the need for environments to support remote collaboration, learning, and research. Existing network technologies can be used to partly service the requirement for self-paced teaching, but a greater level of tool integration is needed to support directed learning and collaboration. This paper discusses work done to marry traditional text-based conferencing with the facilities available through a graphical World Wide Web browser. The uses and needs of reconfigurable multimedia-based ‘virtual’ environments are explored, suggested from experience in producing a prototype development.

Introduction The World Wide Web (Berners-Lee et al., 1994) has brought an element of interactivity to an essentially static worldwide distributed document system. By providing for hyperlinking (Bieber & Vitali, 1997), these systems allowed the user to conceptually arrange information to fit their personal needs. This ‘hyperarchical’ system, proposed by Bush (1945) and Nelson (1990), allows the underlying data structure of the archives to be hidden under a collection of semantic organizations. The inclusion of the Common Gateway Interface (CGI) adds server interactivity to WWW documents by allowing programs to generate customized Web pages depending on selections by the client. This provides an element of dynamism to documents, as well as making scarce or remote services available through a unified interface. These developments make the Web ideal, and intuitive, for self-paced research and learning. However, the act of learning is not a solitary one, and these mechanisms are far from ideal for providing a basis for collaboration; the term ‘Web community’, often heard to describe the set of WWW publishers and users, is something of a misnomer. An alternative exists in text-based conferencing, but the lack of graphical ability has until now limited the use of these systems to disciplines which survive in a graphically-starved environment. In particular, they have not been significantly used by science, until Web servers became viable sources of support materials. There is a strong need to combine the collaboration inherent in systems such as these, with the user interface presented by the Web, in way that allows diverse use of shared documents and facilities. Such a fusion would provide a useful setting for meaningful interaction with intelligent agents (Kosoresow & Kaiser, 1998), programmed demonstrations, teaching software and experimental models, as well as with co-workers and co-learners.

A Virtual Laboratory The ‘Virtual Laboratory’ is one of the most important concepts in Computer-Aided Learning (CAL). It provides an experimental environment for demonstrating concepts not readily explored in physical ways. The demonstration can be done through sound, animated graphics and text, and might be shared in some way by several learners. The concept of the virtual laboratory draws heavily on the psychological theory of learning called constructionism (Harel & Papert, 1991), which has been built upon the work of Piaget. The basic hypothesis of Constructionism is that a person (Piaget studied child learning) understands a subject more rapidly and more fully when that subject can be easily assimilated into his or her view of the world (or explained in terms of familiar or relevant objects). Thus, a child familiar with gears could be helped to use this familiarity to aid in the understanding of many concepts in mathematics. Broadly speaking, one learns by doing, and the virtual laboratory allows one to ‘do’ theoretical, formal, or impractical things. The virtual laboratory facilitates open learning, distance learning, and could be of great assistance in classroom teaching, where a practice developed early in child education is independent study. It could display more substance than current ‘instructionware’, but must make use of multiple media to gain that substance. In addition, it should stress multiperson operation and collaboration, giving courseware and learners a place to happen together. A multi-user learning environment could more fully exploit the benefits of learning in groups. Multimedia displays have been demonstrated to 1

enhance the quality of on-line information, and improve a learner’s ability to retain and understand that information. An excellent simple example of a virtual laboratory is Logo, the mathematics and programming package for children by Seymour Papert (1980). Using Logo, powerful concepts in Computer Science, such as recursion and procedures, can be explained to young children in a way which hides their formal complexity and emphasizes their use. However, although Logo employs graphics to explain programming concepts, in its basic form it could hardly be described as multimedia. Nor is it multi-user, although several children would usually share the keyboard in an attempt to program their latest picture of a house or a flower. A virtual laboratory might be divided into areas, or ‘rooms’, based on material that needed to be illustrated or discussed. For example, a Computer Architecture laboratory might be divided so that each room concentrates on parts of the architecture of a computer at different levels of ‘magnification’. One room might concentrate on the operation of the Central Processing Unit (CPU), where learners could manipulate the registers in the processor. Leaving the room, the users might enter a room where the connections between processor, bus, and memory, are related to them, and the CPU is viewed as a whole, the contents of the CPU room no longer visible. This same approach could be adopted to allow users to browse through any complex computer system, such as a machine, an instruction pipeline, a compiler, or simply an algorithm. Such a virtual laboratory architecture need not be limited to teaching systems. The following are other possible uses of the system:

 A collaborative document system. A shared environment in which documents can be produced, then distributed via the Web will make a good environment for collaborative document authoring. Access to databases through such an environment would be an advantage, allowing technical documents to directly reference, or even react to, the data around which they are written. In such a system, a technical report need not be a static document.  An advanced conferencing system. Integrating voice and video streams into a multi-user system could provide an advanced conferencing system where effective shared tools can be more tightly integrated into the conferencing environment. Such an environment could have shared Internet search tools such as browsers and search engines (Masinter & Ostrom, 1993), and even become a strong environment for the collaboration between users and software intelligent agents (Kosoresow & Kaiser, 1998).  An advanced desktop interface. A virtual environment of this form could be usefully embedded into operating systems, where objects in the virtual environment may appear as representations of devices or applications in a distributed computing environment. The collaboration facilities could provide a framework for enhanced communication between user and machine, perhaps introducing simple natural language communication of the form already seen in text-based multi-user environments.

2

The World Wide Web and teaching The WWW offers a strong mechanism for distributing many kinds of information, and arranging this information in many ways, via the process of hyperlinking. Little special knowledge is needed to navigate the Web; its mechanism is hidden from the user, who only sees the documents and their logical links. The simplicity and intuitiveness of arranging topics in this hyperarchical way makes the WWW an excellent medium for the dissemination of course notes. Material other than straight text can also be delivered in this way. The WWW has become a strongly graphical medium, and in its development from a text-based to multimedia document system, it has set many standards and adopted many other, to encourage wide access to on-line material: in short, the Hypertext Markup Language (HTML) has become ubiquitous. Material not handled by the browser is catered for with additional applications on the client machine. HTML is the ‘code’ of the pages of the Web, and as such is as deep a level of technology as most Web authors need to understand. More recently though, an additional document language, called the Virtual Reality Modeling Language (VRML), has arrived on the Web. This provides a standard way of describing three-dimensional objects such as graphs, car engine components, or rooms, and has been designed with hyperlinking between scenes in mind, and the ability to anchor other documents, such as Web pages, to objects in the scene. For the dissemination of more dynamic document forms, however, understanding of HTTP (the Hypertext Transfer Protocol) is needed. HTTP forms the infrastructure of the Web: HTML documents refer to other documents via hyperlinks, and HTTP is used to handle the collection of these documents from remote servers. HTTP is a simple transmission mechanism, which in its early stages was essentially arbitrating one way traffic (the bulk of traffic is incoming from the client perspective, and outgoing from the server perspective). A document is requested from a server with a ‘GET ’ request, and returned to the client, which formats and displays on-screen according to the type of the document and any markup it contains.

Dynamic documents and other applications While the HTTP protocol as it stands provides a suitable mechanism for the distribution and linking of large databases of static documents, it is not suited to the delivery of documents based on user input. Dynamic documents are covered by an extension to HTTP called the Common Gateway Interface (CGI). This extends the protocol with two methods: PUT (the reverse of GET), and POST, which is the most used mechanism, designed for submitting the contents of a form to a program on the server. CGI also extends the GET method by allowing the submission of simple parameters for searchable indices and the like. When a user fills in a form on Web a page, and submits the filled in data to the server, the client sends a POST request to the server process on the remote machine, which encodes the contents of the form. The program is invoked on the remote machine, the input sent to it, and the resulting output is returned as a Web page. Resulting data may be a text file, a picture, or a fully graphical Web page. Such a system is more than adequate for simple shared access to databases: database documents can be retrieved with GET and modifications sent to the server with POST. From the teaching perspective, simple queries are ideal for on-line dictionaries and other searchable indices, and fill-in 3

forms can be put to good use for quizzes and tests, especially multiple-choice questions. In addition, work could be submitted by students via the PUT request. Other, more advanced applications using more of the multimedia facilities of HTML and CGI have included a multimedia atlas, and shared interfaces to real remote devices such as telescopes and robots, which respond to user interaction and return a document including images, which describes the current state of the device. Dynamically generated VRML will soon find its way into systems such as these. Thus the Web, through HTML, HTTP and CGI, has produced the first easily used standard interface to remote applications on the Internet. Managers of shared resources need not now provide their own interface software for applications whose interfaces can be readily realised in a Web browser window.

Failings of the Web CGI is, unfortunately, quite limiting. More advanced applications, such as two-way or multiway communication, or real-time system operation, are extremely difficult to implement and manage. The principle reasons for these difficulties are as follows:

 There is a lack of ‘state’ in the browsing supported by HTTP. A request for a document carries no state about documents requested beforehand; while the client knows which links have been followed before, the server has no means to find this out without a significant amount of added coding. This can be done for small amounts of state, but larger amounts become more difficult.  There are no mechanisms for synchronous, incremental document updating. Once a resource has been requested and delivered, there is no way a server can send additional information to be added to the document, nor can it easily instruct the client to perform synchronous activity as such.  HTTP, in its current form manages only a single request per connection. A document which includes thirty icons from one site will open and close thirty-one connections, which introduces a heavy resource overhead to the server, and makes predictions of the length of time to deliver a composite document very difficult. Measures are being taken to allow sockets to remain alive until all components are delivered, and to allow simultaneous streaming of several requests.  Two-way streams are not supported – an HTTP connection manages a connection in a halfduplex fashion – a request goes to the server, followed by a document sent from the server. Most collaborative applications require a two-way stream, with no enforced order of transaction.

Summary In summary, it can be said of the Web protocol and language suite that it exceeds the expectations for which it was originally designed, and is a powerful and system-independent document distribution system. HTML provides a very suitable, simple markup for structured documents, and HTTP 4

delivers these documents reasonably effectively. CGI is an effective gateway mechanism for simple databases and elementary shared services. However, for many applications, in particular the applications which form the focus of this paper, virtual laboratories, HTTP is a long way short of being a satisfactory protocol.

Internet collaborative systems The main alternative to the Web for collaborative environments lies in text-based multi-user systems. Unix has for a long time provided the ‘talk ’ system to allow two, and sometimes more, people to type into a window that each user can see. A more advanced system, called Internet Relay Chat (IRC), was modeled on Citizens Band radio, and provides a conferencing system organized by topic, where a ‘channel’ is reserved for a given discussion. Similar facilities such as EarthWeb Chat are now available directly as interactive Web pages. One of the most popular text-based multi-user systems is MOO (Curtis, 1996) (MUD Object Oriented), an object-oriented successor to the Multi User Dungeon (MUD) games of the late 1970s and early 1980s, developed by Pavel Curtis at Xerox Palo Alto Research Center (PARC). It is a reconfigurable system capable of supporting many other applications other than games, but is frequently used as a sort of virtual reality where the reality is based squarely in the imaginations of the users. MOO is popular because of the notion of a persistent shared space that is presented to its users (Bruckman & Resnick, 1993). This space consists of a number of ‘rooms’, with exits leading from room to room. The first and most popular of the MOO systems is LambdaMOO, hosted at Xerox PARC. The LambdaMOO environment is rich, intuitive, and entertaining, and many have sought to emulate its atmosphere when developing their virtual classrooms and even whole on-line universities. A MOO server is accessed by its users with the telnet program, or a piece of specially-designed client software. The system itself is divided into two parts: the server, and the database. In the database are representations of objects, which have methods (called verbs) which can be invoked by the users and by other objects. The MOO server performs simple time-slicing which allows the processing of jobs from many users simultaneously. MOO objects are written in MOOcode, an object-oriented language with list-processing, with syntax similar to C or Java code (Flanagan, 1997). Verbs are invoked by the users through a simple English-like parser, which allows for commands such as: put tin in box take tin from box press button on box point browser at www.cs.reading.ac.uk say hello This simple interface makes many tasks very easy to perform, and it is flexible enough to allow diverse uses. Interactivity is its strong point, and many MOO administrators (known often as Wizards or Janitors) implement sophisticated verb structures for communicating with people beyond simple ‘say’ commands. It is also a very easy environment to develop and modify: the inheritance system 5

makes changes to all objects descended from a given object possible by modifying only that object. Thus, all rooms in a system could have a new feature added to them by adding it to $room, the generic room object. However, the MOO system has many failings, many of which are only apparent in the light of the extraordinary success of the Web. The most significant are as follows:

 The most significant failure is the limited and complicated support for binary data. This makes the delivery of many media difficult, particularly image data. As such, MOO is a firmly textbased system, and although there are many implementations of MOO Web servers, these usually rely on an external Web server for delivery of images.  The server is a large process, with its own memory and task management. This resists portability to many systems. The entire MOO database is held in memory (real or virtual), and process sizes of over 10 Mbytes are not uncommon in many MOO systems.  All editing of MOO objects must be done within the MOO itself – the database is not stored in a human-readable format. This can make porting of objects from one MOO environment to another very difficult.  Most MOO systems are plain text based, since there is no standard client, nor is there a standard markup protocol. This means that hyperlinks (or gateways) from one MOO to another are not really possible. Several parties are, with the arrival of WOO (Webbed MOO) servers, developing clients to handle markup added into the text stream (Newberg, Rouse III & Kruper, 1995), but the clients are not widely used, or available on many systems. An experimental CoSMOO system has been developed to combine Java client capabilities, the WWW, and other Internet technologies, to create a multimedia teaching and collaboration environment, which can be associated with traditional document systems and databases to provide shared access to information. The client supports multiple fonts and font sizes through an HTML-style markup language (see Figure 1). Browser control is implemented allowing a CoSMOO object to instruct the client to display Web pages. This could, for example, be used by a teacher interacting with potential geographically distributed students. This idea has been further developed as part of an MSc thesis (Cox, 1998). Other MOO systems have been aimed at virtual worlds built by students and instructors. For example, MiamiMOO is an interdisciplinary project that links a text-based virtual reality, a MOO, with the World Wide Web from Miami University, Oxford, Ohio, USA. Objects within the MOO are viewable through the Web and these may link to multimedia graphics, sound, and video. A traditional text-only interface or a multimedia Web browser such as Netscape may be used to navigate around the MOO.

Other technologies The Internet is developing at a staggering pace, so there are many solutions to the problems of networked conferencing. Recently, development of the bandwidth capacity of Internet backbones

6

has allowed use of real-time audio communication and the experimental use of video conferencing (Boyer, 1996). Several packages have arrived for voice conferencing, and Cornell University developers have produced a popular video conferencing package called CU-SeeMe. These technologies demonstrate that many of the problems of real time conferencing can be solved, but they only provide voice or video transfer. These technologies must be merged with shared file editing and virtual environments before the full potential for voice and video conferencing as a collaborative medium can be realised.

Internet programming languages The most recent development in applications for the Internet has been the so-called ‘Internet programming language’. Essentially a reworking of the p-code standard object code concept of Pascal, these languages are advanced high-level languages which target an architecture neutral environment by producing machine code for a virtual machine. The virtual machine is implemented on several platforms, and once this is done, all the platforms can run all the applications compiled for the virtual machine. The most popular of the Internet programming languages is Java (Flanagan, 1997). from Sun Microsystems. This language has been designed over a long period to address many of the criticisms of traditional languages, such as:

 Security. In a networked environment, users must be able to trust that their applications will be unable to carry out hostile actions, such as opening the machine to virus attack, or transmitting sensitive data to unknown remote sites. Java implements security managers to limit what actions an application can carry out to those deemed to be secure.  Robustness. Applications should not crash. Java addresses traditional program failures by implementing a robust memory management system.  Uniformity of interface. Java implements a GUI (Graphical User Interface) system called the Abstract Window Toolkit (AWT), which consists of interface components implemented in terms of standard components available on all target systems. This helps to ensure similarity of interface on every platform. The Java system addresses most of the issues which arise when code is downloaded unsolicited from the Internet. Having achieved this, it is essentially safe to embed the Java virtual machine into Web browsers, which is exactly what major browser developers are doing. This promises a future where customized interfaces for Internet applications are possible without specialized client software. Clearly such technology plays an important part in the virtual laboratory concept. Java is an immature technology, but its development is closely tied to the Internet collaboration concept, and could advance these systems significantly.

7

Fusing the technologies A generic virtual laboratory system would clearly be a fusion of many of the concepts described above. This section is dedicated to describing some of the most important features of such an environment, the work that has already gone into achieving these features, and possible future work in this area. It is of paramount importance that the virtual laboratory architecture incorporates a full interface to the Web, through HTML, HTTP and CGI, and possibly also VRML. Objects should be able to present themselves in a variety of document forms, whether static or dynamic: the systems must be a location on the Web, which can be accessed by following a hyperlink to a standard Universal Resource Locator (URL) from a traditional document. It is essential that objects in the virtual environment can as far is as possible be addressed by HTTP: the objects must be browsable in the normal way, but also should be able to accept input via CGI form methods, etc. This one way Web integration may suffice for the interface of the environment, but to fully enable the environment to react to the Web, there should be incoming as well as outgoing HTTP activity. Objects should be able to retrieve documents from the Web, to configure themselves, process data, access search engines, or simply retrieve a picture of the user currently operating it. We could envisage a virtual laboratory complex, where each laboratory is handled by a different server and covers a different topic, with hyperlinks or gateways between them, to represent semantic links between subject areas. Existing virtual environment systems have tended to use simple plain text interfaces, with few facilities for emphasising or highlighting text. They also rely totally on the natural language style interface, with perhaps a simple menu selection system. A modern Web-integrated environment must present the fully interactive parts of the system to a similar standard to the static Web documents it can display. This means that the interactive streams should employ HTML-style formatting markup, and graphical user interface components, in combination with the natural language interface. In order to embed the virtual environment into existing networked systems, it should be able to communicate with databases, applications and devices through the most reasonable means (or through the only means supported by those applications). This means that the virtual laboratory should augment its traditional telnet stream interface with facilities to communicate via remote procedure calls (RPC), object request brokers (such as CORBA or Sun IDL), and even traditional Unix protocols such as FTP (File Transfer Protocol) and NNTP (for news transfer). The objects in the environment must have full connectivity with the surrounding network in order to fulfill their roles in the integrated system. As the multi-user environment becomes more involved (more media streams, more synchronization activity), it is clear that the server environment will become larger. Distributing the processor activity of the environment will soon become essential. A remote controller for the Netscape browser has been implemented using the reference material provided by Netscape Communications. This remote control system is limited to Unix (although equivalents exist on most platforms). This controller implements the most important remote control features (most importantly to instruct a browser to retrieve a page). Remote control facilities have been implemented as an extension to Tcl, the Tool Command Language available for Unix systems. The extension allowed for extensive control of a single browser. Simple interactive client software was written to link text-based server applications with WWW information, so that the WWW information could dynamically update as progress was made in the 8

interactive session. The extensions used in this system are simple markup commands, which are obeyed as soon as the client (a modified telnet client) receives them. The client could be used on many kinds of multi-user system, but provided very limited interaction between browser and server. All interactive documents were introduced with the aid of a conventional Web server, and CGI scripting. The next stage in the development of the client technology was to reimplement this client in Java, so that it could be accessed through a netscape browser. Browser control is simple in Java, so attention was turned to experiments in text formatting markup. A simple HTML-like markup language has been implemented, taking into consideration the multiple streams of information that are sent to the client from other users as well as from the programs which constitute the environment. A MOO system has been modified to send this marked-up text to the telnet client. Experiments suggest that the inclusion of text markup significantly improves the text-only experience, although care must be taken to ensure that separate streams of text formatting do not clash to create erroneous on-screen formatting. Other developments in the experimental MOO system include a shared browser object that has formed the basis of incoming HTTP facilities to allow objects to configure themselves from files retrieved from remote Web servers.

Possible directions Future work in this area should involve a full analysis of the requirements of multi-user systems with regard to updating them with graphical user interfaces and multiple media and document formats. Components of this analysis are discussed below. The MOO system has several strengths and weaknesses as a server for more advanced multimedia environments. Full analysis may uncover a method for augmenting the server to handle multiple media types, or could result in a design for a new server.

 Proxies. For example, one plausible way of adding more access methods to MOO is via a proxy system. Additional servers for new access methods such as RPC could be placed around the MOO, and be connected to the MOO via a continuous connection. The proxy might then translate a remote procedure call to a MOO verb call and return the result to the caller. A proxy object within the MOO might arbitrate requests from several proxies, so the integration of new access methods will not require the reimplementation of the environment object database. However, it is not clear how robust this solution is, nor how effective it will be operating through MOO’s rather idiosyncratic time-sharing system.  Communication streams. Another area where significant analysis is required is the issue of the information stream to and from the client. The design of a markup system for text formatting (and cueing of browser activities and perhaps animation and sound) has been discussed earlier. What is unclear at this stage is what is the best way to manage several streams of input from the client. In the current MOO system, the user interacts with the MOO a line at a time, but there are situations resulting from the implementation of a graphical interface to the MOO that require the management of several simultaneous streams. For example, a large text document 9

has been composed in an edit window and is being sent to the MOO at the same time as interaction is required on a control panel for another object. Control actions should not be delayed, so some mechanism for interleaving the actions should be implemented. This can be done through the management of several connections, one for each significant form of message to and from the server, or through the use of connectionless communication along side one stream-oriented connection. The advantages and disadvantages of each of these are unclear at the moment, as is the effectiveness of implementing such a system over MOO.

 Graphical User Interfaces for shared objects. Further analysis is due in the design of GUI interfaces, and their control by the server. It seems likely that two kinds of interface, generic and specific, must be provided. The choice of which mechanism to use would depend largely on the application. A generic interface might respond to a script sent to it, much like an HTML form description. Generic interface components might include audio managers, animation objects, and button panels. Specific interfaces would be hard coded for specific purposes – an interface to a compiler, or a robot, could be implemented so that it need not be sent an interface description.

 Distribution. In-depth study of the MOO system will hopefully uncover ways in which a distributed multiuser system can be created. The possibilities for this are that either load-balancing technique will prove popular (essentially dividing the environment up evenly between several processors or machines), or a role-based division may be more acceptable (dividing up the process according to task).  Implementation in Java. The analysis may lead to a better design of server, and one possible implementation strategy is to implement the server itself in Java. This would create a portable server, as well as portable clients, and may improve the coherence between client and server. A design in Java, with its multithreaded approach, may also be more readily distributed across processors. In addition, the prospect of objects that can readily be ported from one MOO environment to another, is very appealing.

10

Conclusion This report has presented the principles underlying multi-user and multimedia networked environments to aid collaborative learning. A rapid prototype system has been developed to explore the possibilities and a number of directions which it would be worth pursuing have been suggested. It is hoped these will help provide constructive ideas for future collaborative teaching systems. Development is ongoing and there is increasing interest in the educational use of MOOs (Haynes & Holmevik, 1998). Remote collaboration using the Internet is a continuing topic for research (Rhee, 1999) and has especially exciting possibilities in developing countries where access to education is difficult (Muller, 1999). Developments in video conferencing (Boyer & Lukacs, 1996), and computer-supported collaborative work (CSCW) (Takeda, Inaba & Suguhara, 1996), are opening up new teaching opportunities if used imaginatively. Improved multimedia technology could be extremely useful for practical distance teaching and learning (Woolf & Hall, 1995) when enough potential users have network access of sufficient bandwidth and reliability. The World Wide Web Consortium (W3C) is developing new Web standards such as XML (eXtended Markup Language), a generalization of HTML, and in particular SMIL, a new Synchronized Multimedia Integration Language aimed at the Web (W3C, 1998). Such developments are likely to further enhance the facilities available to users via the Web, as browser support and on-line material become available. The vast amount of information available on-line means that efficient and effective search techniques to search for relevant educational material will required. The Internet can be viewed as a single, vast and ever changing digital library of knowledge (Fox et al., 1995), but with no overall control or indexing, and of course with greatly varying quality and usefulness. Many organizations, such as museums (Bowen, 1997 & 1999), make material available that is suitable for teaching purposes if organized and presented in a suitable manner. There are already keyword search engines and hierarchical directories freely available that help in the search for such information. However the automated techniques of today are not entirely satisfactory and improvements will be required. Categorization and searching of Internet resources remains a subject of continuing research (Clever Project, 1999). Although difficult to estimate accurately, there are over 100 million people with access to the Internet and the number is approximately doubling each year. WWW usage initially doubled every three months, an increase of 1% per day. The nature of the WWW user population has been a subject of study (Pitkow & Kehoe, 1996); the average users still tend to be relatively well off young males, although this is likely to change as other sections of society gain easier access. Only when access to the Internet is as universal as the telephone is today will the true benefits of on-line learning and collaborative environments to society as a whole be fully assessable.

11

References Berners-Lee, T., Cailliau, R., Luotonen, A., Nielson, H. F. & Secret, A., (1994) The World-Wide Web. Communications of the ACM, 37, 8, 76–82. Bieber, M. & Vitali, F. (1997) Toward Support for Hypermedia on the World Wide Web. IEEE Computer, 30, 1, 62–70. Bowen, J. P. (1997) The World Wide Web and the Virtual Library museums pages. European Review: Interdisciplinary Journal of the Academia Europaea, 5, 1, 89–104. Bowen, J. P., guest editor (1999) Museums and the Internet (1). Museum International, 51, 4, 4–41. (See Only Connect!, pages 4–7.) Boyer, D. G. & Lukacs, M. E. (1996) The Personal Presence System: A wide-area network resource for the real-time composition of multipoint multimedia communications. ACM Multimedia Systems, 4, 3, 122–130. Bruckman, A. & Resnick, M. (1993) Virtual professional community, Results from the MediaMOO Project. In Proc. 3rd International Conference on Cyberspace, Austin, Texas, USA. Bush, V. (1945) As we may think. The Atlantic Monthly, 176, 101–108. Clever Project (1999) Hypersearching the Web. Scientific American, 280, 6, 44–52. Cox, R. (1998) Developing a multimedia MOO for application as a distributed collaborative education system. MSc thesis, Department of Computer Science, The University of Reading, UK. Curtis, P. (1996) LambdaMOO Programmer’s Manual, Version 1.8.0p5. Xerox Palo Alto Research Center (PARC), California, USA. Flanagan, D. (1997) Java in a Nutshell, 2nd edition. OReilly and Associates Inc. Fox, E. A., Akscyn, R. M., Furuta R. K. & Leggett, J. L. (1995) Digital libraries. Communications of the ACM, 38, 4, 23–28. Harel, I. & Papert, S., editors (1991) Constructionism. Ablex Publishing, Norwood, New Jersey. Haynes, C. & Holmevik, J. R., editors (1998) High Wired: On the Design, Use, and Theory of Educational Moos. University of Michigan Press. Kosoresow, A. P. & Kaiser, G. E. (1998) Using agents to enable collaborative work. IEEE Internet Computing, 2, 4, 85–87. 12

Masinter, L. & Ostrom, E. (1993) Collaborative information retrieval: Gopher from MOO. In Proc. INET’93, Annual Conference of the Internet Society. Mueller, M., guest editor (1999) Emerging Internet infrastructures worldwide. Communications of the ACM, 42, 6, 28–72. Nelson, T. (1990) On the Xanadu Project. BYTE Magazine, 15, 9, 298–299. Newberg, L. A., Rouse III, R. O. & Kruper, J. A. (1995) Integrating the World-Wide Web and multiuser domains to support advanced network-based learning environments. In Proc. Educational Multimedia and Hypermedia, Graz, Austria, 494–499. Papert, S. (1980) Mindstorms: Children, Computers, and Powerful Ideas. Basic Books, New York. Pitkow, J. E. & Kehoe, C. M. (1996) Emerging trends in the WWW user population. Communications of the ACM 39, 6, 106–107. Rhee, I., guest editor (1999) Collaboration – Internet-style. IEEE Internet Computing, 3, 2, 30–73. Takeda, K., Inaba, M. & Suguhara, K. (1996) User interface and agent prototyping for flexible working. IEEE Multimedia, 3, 2, 40–50. Woolf, B. P. & Hall, W. (1995) Multimedia pedagogues. IEEE Computer, 28, 5, 74–80. W3C (1998) Synchonized Multimedia Integration Language (SMIL) 1.0 Specification, Technical Report, World Wide Web Consortium, MIT, USA. URL: http://www.w3.org/TR/REC-smil/

13

Some Relevant On-line Resources The following resources may be of interest to readers of this report:  Reading University Multimedia Group http://www.fmse.cs.reading.ac.uk/mmg/  Virtual Worlds http://www.fmse.cs.reading.ac.uk/mmg/virtual worlds/  CoSMOO http://www.fmse.cs.reading.ac.uk/mmg/cosmoo/  MOO home page http://www.moo.mud.org/  MOO-Cows FAQ (Frequently Asked Questions) http://www.moo.mud.org/moo-faq/  Lingua MOO – An Academic Virtual Community http://lingua.utdallas.edu/  Yahoo MOO links http://search.yahoo.com/bin/search?p=MUD+MOO  Common Gateway Interface (CGI) http://hoohoo.ncsa.uiuc.edu/cgi/  Virtual Reality Modeling Language (VRML) Consortium http://www.vrml.org/  World Wide Web Consortium (W3C) http://www.w3.org/  Hypertext Transfer Protocol (HTTP) http://www.w3.org/Protocols/  Hypertext Transfer Protocol – Next Generation Overview http://www.w3.org/Protocols/HTTP-NG/  Synchonized Multimedia http://www.w3.org/AudioVideo/  Diversity University, Inc. http://www.du.org/  Java Home Page http://java.sun.com/  Entranet http://www.entranet.co.uk/

14

Figure 1: Screen-shot of CoSMOO

15