A networked virtual environment, a type of virtual reality most commonly known as a "MUD" or a "MOO," was used at an internationally-attended scientific conference held on the Internet. Interviews with conference attendees indicated enthusiasm for the effectiveness of the medium, and revealed how the virtual environment can be modified to improve conferencing efficacy. Such alterations included novel input-output control management; automation of conference registration; control of anonymous or guest connections; simplification of conference center topography; an improved methodology for recording discussion sessions; use of moderated rooms to compensate for network lag; and providing buffers to update late arriving participants. It is concluded that networked virtual environments provide an inexpensive means for effective international conferencing on the Internet.
John F. Towell, Ph.D., Operations Management and Information Systems Department, College of Business, Northern Illinois University, DeKalb, Illinois 60115, Phone: (815) 753-6377, Fax: (815) 753-7460, Email: towell@mcs.anl.gov
John F. Towell (towell@mcs.anl.gov) completed his doctoral work in computational biochemistry at Colorado State University and is currently involved in postdoctoral research in Informatics at Northern Illinois University. His research interests include the use of networked virtual environments as teaching and conferencing tools. He is also an administrator at BioMOO, a virtual meeting place for biologists, which is located at the Weizmann Institute of Science in Rehovot Israel (bioinfo.weizmann.ac.il 8888 and http://bioinfo.weizmann.ac.il:8888).
Elizabeth R. Towell, Ph.D., Operations Management and Information Systems Department, College of Business, Northern Illinois University, DeKalb, Illinois 60115, Phone: (815) 753-6377, Fax: (815) 753-7460, Email: etowell@niu.edu
Elizabeth R. Towell (etowell@niu.edu) obtained her Ph.D. in Management Information Systems at the University of Wisconsin-Milwaukee. She is an Assistant Professor in the Operations Management and Information Systems Department of the College of Business, Northern Illinois University, DeKalb, IL, 60115. She teaches courses in Operating Systems and Database Management Systems. Her research interests are primarily in database interoperability.
Traveling to conferences and meetings is expensive and time-consuming. Recent advances in Internet technology have included the development of multi-media virtual realities which provide real-time communication in a programmable object-oriented environment. These virtual realities, referred to here as networked virtual environments (NVE's), are also known as MUD's (Multi-User Dimension/Dungeon). Although primarily used for games, the potential of NVE's as a serious communication medium is beginning to be recognized. An NVE was used at an internationally-attended scientific conference held on the Internet. Interviews with the users helped identify common difficulties encountered while using the NVE as a conferencing tool. This manuscript describes the NVE, the interview results, how the NVE was improved in response to these results, and further studies planned to improve the NVE as an effective Internet conferencing medium.
Networked Virtual Environments
A Networked Virtual Environment (NVE), is a server that provides information to clients about a database that it also maintains. The database contains information about a collection of objects that can be characterized as rooms, things and other connected persons. The information in the NVE database is organized in such a manner that the person using the client is presented with a virtual representation of a real room in which there could be other people with whom one may "talk." Talking in this context means using the keyboard to communicate with the other person(s) who share the same room with you; likewise, "hearing" means to see the NVE output. Typically NVE's are compartmentalized into rooms joined by entrances and exits. Hence, there is a topography which can be navigated with simple commands such as: "go south." When people move from one room to another they lose the communication they had with the person(s) in the previous room and can hear and talk only with those that are in the new room. Hence, the NVE is a type of virtual reality which is aptly suited for use as a conferencing tool in any internet, such as a university or corporation, or the Internet.
A recent survey of business schools indicated less than 5% considered "networked virtual environments (MUD's)" as very important topics (E. R. Towell, in press). Indeed, NVE's have been criticized because of their prominent use by college students for recreational activities (Evard, 1994). However, this criticism shows blindness toward the medium's capabilities (Thornley, 1994) and the potential of NVE's as a serious communication medium is gaining recognition. The National Institutes of Standards and Technology have expressed their interest in NVE's as a communication tool in the manufacturing industry .3 A systems administration group in the College of Computer Science at Northeastern University found that using an NVE for multi- user real-time interactions provided, in many instances, much better communication than electronic mail (Evard, 1994). An NVE was accepted by the scientific review committee at the first Electronic Computational Chemistry Conference as a serious medium for conferencing (J. F. Towell, in press).
Although many types of NVE's are available, this manuscript reports observations made using a MOO, an acronym for Multi-user dimension Object Oriented. In addition to the communication and navigation capabilities discussed previously, a MOO is extensible. Specific commands enable the user to create their own rooms and objects in such a manner that the environment becomes a medium of self-expression. Thus, the server supports multiple connections from Internet clients, provides database information, and modifies the database in response to user commands (Figure 1).
For additional background information about MOO's, the reader is referred to the MOO Programming manual, Pavel Curtis's exposition (1992) and other MOO- related documents, such as a mailing list which focuses on maintaining MOO's and the MOO language. The MOO server and standard database, which continues to be developed and supported by Pavel Curtis, and the information mentioned above are all available at the Xerox PARC Internet site.1 Also, an extensive source of MOO-related documents and writings can be found on the World Wide Web.2
The observations presented here were made at the first Electronic Computational Chemistry Conference organized by Steven Bachrach and colleagues at Northern Illinois University, November 7-18, 1994 (Krieger and Illman, 1994; Bachrach, in press; Hardy, in press). This conference was conducted on the World Wide Web with discussion of presented papers occurring via an electronic mailing list. To explore the use of NVE's in scientific conferencing, an NVE was created as the prominent feature of a paper submitted to the conference (J. F. Towell, in press). This NVE was prepared by the administrators of BioMOO, an NVE for professionals in the biological sciences, which has been recognized for its internationally-attended scientific seminars (Anderson, 1994). The NVE submitted to the conference was named the Virtual Conference Center (VCC). Our paper, entitled "Networked Virtual Environments and Electronic Conferencing," consisted of the following information which was accessible via the World Wide Web. The first part contained instructions for connecting to the VCC; the second part briefly discussed computational aspects of a networked virtual environment; and the third part contained information about the authors. Of the 294 conference registrants, 99 (34%) eventually registered to gain access to the VCC. During the meetings, extensive conferencing took place within the VCC. Recordings of those discussions were submitted to the mailing list so that all conference attendees could read and comment on the content of the discussions. The following discussion is the result of the conference attendees' reactions obtained either by interview within the VCC or electronically mailed comments. This paper discusses the problems associated with the use of an NVE, such as the VCC, at an Internet conference.
In the following discussion, words in bold type indicate commands that would be entered from the user's keyboard. For example, if the user were to issue a command to say, "hi," the reader will observe the command in bold type like this: say hi. It is assumed the user would type the command and then strike the enter key of the keyboard.
Connecting to a Networked Virtual Environment
The most generally available method for connecting to an NVE is with the use of telnet. However, this type of connection is generally accompanied by a surprising and confusing event. While entering a command or statement, output from the NVE disrupts the same line the user was trying to type. Observing one's entry disappear is baffling which compounds the confusion which usually exists from initial exposure to a virtual environment. Experienced users employ clients programmed specifically for interacting with NVE's. These clients separate NVE output from the user's input. Clients are available on the Internet by ftp, but some users do not have the skills or the time to acquire and install these programs on their own systems. To solve this problem VCC registrants were given instructions for connecting to an Internet site that automatically provided a client for connection to the VCC. Providing this client proved to be an effective means for bringing inexperienced users into an NVE.
The client, tinyfugue, was preset to the /visual on mode which presents to the user a screen divided by a horizontal line about three fourths of the way down the page. Output from the NVE is displayed in the top part of the screen while input from the user is displayed in the lower portion. Some participants complained of screen disruptions which were corrected when they entered: /visual off. Turning to /visual off mode no longer divides the screen into output/input panels but does preserve the user's entries when output from the NVE is displayed. Normally users type: @quit to break their connection to an NVE and then: /quit to discontinue the client. The client was programmed to quit following the @quit command which alleviated the problem of people being able to disconnect from the VCC but not knowing how to quit from the client.
We have programmed the VCC to provide telnet-connected users with a method that allows them to enter their statements without interruptions from the room. This novel method is called the Output Control Manager.
To understand the following discussion it is necessary to introduce the
commands for making statements within the VCC (and most NVE's). Assume the
user's name is Beth. To make declarative statements, Beth would type: say
The output control manager is activated when users enter the say (") or emote (:) command without any text following. They are then shielded from VCC output until their statement has been entered at which time they are given a playback of any statements that had occurred while they were in the input mode. This feature has been received with great enthusiasm by many NVE managers who have adopted the code.4
There was some frustration with the VCC registration procedure because of several complicating factors associated with creating and debugging the VCC. Although these problems would not have occurred if the bugs had been discovered earlier, it nevertheless prompted us to consider better ways to conduct registration.
There are two standard registration procedures. In the first, requests are sent by NVE-connected users which contains their choice of a user identification name (userID) and their Internet address. The server checks the userID for database redundancy and returns to the user the approved userID along with a password via electronic mail. In the second, an administrator of the NVE performs the process manually. The former is undesirable because the user often selects a userID that provides insufficient user identification. Likewise, the latter requires excessive time. Having a suitable identification is important because recordings were made of the discussions. When users talk in an NVE, their name precedes their statement. For example, if Pavel_Curtis says, "Hi, I'm new here," everyone in the same room plus readers of the recording transcript will see: Pavel_Curtis says, "Hi, I'm new here." Hence it is important that an informative userID is assigned to the person. If one plans to use an NVE as part of an electronic conference, we suggest pre-registering the participants and sending them an e-mail message containing their userIDs and passwords for the VCC. Pre-registration was not done in the conference reported here because the VCC was not part of the conference organizational structure. Automatically issuing userID's is advantageous because the userID can be set to the conference registrant's real name and thus provide adequate identification of the participant within the VCC. Also, with regard to correct identification in the VCC, the @rename command should be deactivated to prevent a particular user from assuming another identity.
The VCC database requires a server that can utilize external files. Normally information stored in the standard MOO database is kept in memory but the VCC keeps much of it's information on disk to be loaded when needed. This external file utilization requires the use of the File Utilities Package (FUP) developed at the Weizmann Institute of Science by Gustavo Glusman and Jaime Prilusky.5 During registration for an Internet-based conference, registrant information can be written to an external file which in turn can be read by the VCC and used to create a character with an appropriate userID.
Generally most NVE's permit a type of connection which does not demand a userID or password. Such anonymous connections are called guest connections and they may be problematic. The VCC was configured to permit guest connections for several days preceding the commencement of the meetings. This provided an opportunity for new users to come into the environment and explore it anonymously. However, once the meetings commenced, guest connections were shunted into a room which instructed them on registration procedures and barred them from the VCC discussion rooms.
It has been suggested that guests be given more opportunity to explore the VCC anonymously after the meetings have started. The VCC has been altered by extending the guest room area to include rooms for running tutorials on using the VCC. Another suggestion was to allow guests to hear the discussion taking place in various conference rooms without participating in them. Guests could be provided with a radio-like device with channels for selecting among the various discussion sessions currently underway. By experiencing the interactive discussions, guests might eventually want to come into the VCC as participants. Allowing guests to take tutorials and hear conference discussions will be studied at the upcoming conference discussed in the discussion section of this paper.
Virtual Conference Center Topography
The topography of the VCC was intended to be as simple and straightforward as possible. The VCC can be configured so that it brings a newly connected user into any room. The VCC starting point was a central room called the PLAZA (Figure 2). The PLAZA was used for posting scheduling information and, as in real conferences, contained bulletin boards which could be used for a variety of messages such as employment opportunities and personal messages. From the PLAZA, the user needed to issue only one command to enter any of ten rooms, eight of which were named according to their direction from the PLAZA (the COFFEE SHOP was down from the PLAZA and the UPPER ROOM was up from the PLAZA). For example, to attend a meeting in the NORTH ROOM, users entered: north or n following their connection and arrival in the PLAZA. Each of the conference rooms, except for the PLAZA, had only one entrance and one exit. Thus, to go to the PLAZA from any of the conference rooms, one merely had to type: out or exit.
The VCC contained recorders in each of the rooms, except the PLAZA, and recorded discussions were kept and archived. The recorders were stopped and reloaded daily. Recordings from the previous day were then edited and sent to the conference mailing list for all to read. Some conference participants had difficulty reading the transcripts from recorded discussions. One common problem was discomfort with personal comments intertwined with serious discussion.
The interactive aspects of NVE's provide a surprisingly natural social environment, and humans, being playful creatures, often behave as such. This playfulness was generally reflected in the recordings of discussion sessions even though the topic of discussion was serious and of a highly technical nature. Part of the playfulness is derived from the discovery of the new and stimulating NVE environment. Even those who were present at the discussion and thought the personal comments were not distracting at the time felt that reading the transcript afterwards was difficult. The interlaced personal comments seemed mere distractions later on. Hence, this phenomenon seems appropriately described as, "You had to be there to appreciate it."
How can we capture only the serious part of the discussion for the transcripts? This problem may be solved in several ways. First, the VCC has commands that permit interpersonal communication which is not captured by the recorders. These methods can be used effectively for side comments and are used extensively by the more experienced NVE user. As discussed previously, the say command may be issued with a double quotation mark ("). We have modified this by designating two double quotation marks ("") and emotes (:) when preceding text to be recognized by the recorder as an off-the-record statement. The recorder responds accordingly by placing a pound sign (#) at the beginning of the statement. Hence, when a transcript is generated from the recording, off-the-record statements can be excluded from the transcript. Thus, the smiles, gestures, jokes and other personal comments are kept as part of the moment but not formal history. A description in each of the conference rooms informs the users that statements that are not intended for the record should be preceded with two double quotation marks and that communications made via the emote command will also be excluded from the record. Other changes have also been made to filter messages generated by the server from the transcript of the discussions. For example, if someone in the room disconnects, an announcement is made to everyone in the room, including the recorder. The VCC has been modified to place the pound sign in front of any general announcements made by the server during discussions. The results of the modifications to the recording and transcript-generating procedures will also be tested at the upcoming Electronic Glycosiences Conference discussed later in this paper.
Internet Lag and Moderated Rooms
A second common problem with reading transcripts involved statements that seemed out of place or were not in the flow of the conversation. The appearance of disjointed comments can be due to many factors. Depending on the time of day, Internet lag can physically prevent a person's statements from appearing on the screen in the temporal order of a natural conversation. Internet lag is defined here as simply a delay of signals sent between the client and server due to communication delays on the Internet. Delays can also be due to the server, but that was not the cause of the lagging reported here. Hence, when such delayed signals are received, they are not synchronized with the rest of the discussion and appear disjointed. During the conference, the European participants talked of considerable transatlantic lag that worsened towards their afternoon. What can be done to accommodate those who are having difficulty with Internet lag?
Experienced NVE users typically announce when they are experiencing lag so others in the conversation will understand and compensate by waiting for the lagger's responses. Inexperienced users will probably have difficulty adopting this compensatory mechanism, and the VCC itself can be used to help. During seminars at BioMOO and discussion sessions at the VCC, we have experimented with the use of a moderated room.6 This room permits communication only after the user has entered: request. Following the request command, users are placed in a queue and must wait for their turn to talk. Current speakers have the floor for their discussion until they are finished at which time they enter the command: yield. Then the next person in the queue is allowed to talk. The number of people who are permitted to talk is adjustable. A moderator can monitor the queue to see who is waiting to talk and can encourage users to yield or, if necessary, force them to yield. With the use of the moderated room, the discussion was focused, and those with lag problems, once they gained the floor, could make their comments in the natural flow of the conversation. More experimentation is needed with this technique, but there were no complaints from VCC users about adjusting to the request and yield commands in the moderated room.
Scheduling International Conferences
One major problem with holding internationally-attended conferences is deciding when to have them. The time-zone distribution of the presenters at the conference (Figure 3) indicates that a meeting scheduled at 8:00 Pacific Standard Time (PST), which is Greenwich Mean Time (GMT) minus 8 hours, would occur between 8:00 am and 5:00 pm for most of the people who were presenters at the meetings. However, Internet lag was most prevalent for the Europeans at 8:00 pm PST which was also a time outside of normal working hours for those located in Australia, Hawaii, and Hong Kong. One group of conference attendees decided to resolve these problems by scheduling two conference times, the first at 08:00 GMT, the second at 23:00 GMT. The transcript from the morning session was sent to the mailing list so those attending the later conference could discuss the topics from the earlier session. Having two conference sessions worked very well and should be considered as a useful alternative to scheduling only one session when most attendees are in their normal workday. It also revealed the importance of the interplay between the VCC real-time conferencing and the communication via the mailing list. There is no need to have one without the other, each has distinct advantages.
When a person is connected to the VCC, the object that represents that person in the virtual sense is referred to as a player-object. There is a hierarchical classification of player-objects that progressively allow greater control of the NVE environment. A "player" can only connect to the NVE, navigate it, talk with other users and manipulate the objects there. A "builder" has all the capabilities of the "player" and the additional ability to create new objects such as rooms and things that can be placed in the room. The VCC registrants were automatically given "builder" status and many of the scientists, with the use of the extensive help system, quickly learned how to use the create commands. Soon the VCC was littered with lamps, sofas, cappuchino machines, liquor bottles, etc. While the ability to make objects is undoubtedly a fascinating property of the VCC for many people, the result was the creation of distractions for others who desired only to discuss science and not pass around wine bottles. Thus, while it is difficult to sanction restriction of personal freedom in the VCC, it is perhaps best if conference attendees are given "player" status to minimize potential problems with differing personal opinions of a proper VCC environment.
Stage talk is the term used to direct a comment or question to a particular person in the same room. For example, if Franz were in the same room with John, and John typed: to Franz What do you mean?, all present in the room would see on their screens: John[to Franz]: What do you mean? Stage talk is a useful tool when involved in multi-person discussions. However, the VCC did not explain how to use this feature in its help messages nor were instructions provided in the preliminary instructions sent to VCC registrants when they first registered. The conferees were quick to learn it when they were with someone else who was using it but we suggest including this feature and appropriate instructions in any NVE to be used for conferencing. The VCC now provides use of the command to which permits a combined ability for stage talk while protected by the output control manager discussed earlier.
As can be expected at any conference, people arriving late would often ask to be updated about the current topic of discussion. Usually one or more of the conferees would attempt to explain the current discussion to the late-comer which was disruptive, wasteful and inefficient. Hence, the VCC rooms have been modified to contain buffers that can update a tardy attendee with a replay of recent conversation in the room. The buffer was set to replay the last 50 lines of the conversation. Each room description, which is shown to users when they enter, contains instructions for using the replay buffer. However, this information should also be sent to users upon registration. Further work is planned to determine how many lines of recent conversation are necessary to update a late-arriving participant.
This discussion has centered on the experience gained during the presentation of the Virtual Conference Center(VCC) at the first Electronic Computational Chemistry Conference in 1994. No formal scheduling was attempted by the VCC presenters. All VCC discussion sessions were either spontaneous or organized by conference registrants. Undoubtedly, greater participation would have occurred if the VCC had been part of the conference organization and prior planning and scheduling could have been accomplished. The first Electronic Computational Chemistry Conference was an experiment in the use of the World Wide Web as a conferencing tool. The conference organizers had decided on the use of electronic mail as the means to discuss the papers. We think each medium has distinct advantages and the use of one does not preclude the use of the other.7 With the use of recorders in the VCC, transcripts were made which were sent to the conference mailing list. Hence, all conference registrants shared the discussions of those that occurred in real-time in the VCC. The first Electronic Glycoscience Conference to be held from September 18 to October 13, 1995 at Oxford University will include, as part of the conference, the VCC. The integration of the VCC into this conference will provide an excellent opportunity to test those VCC changes that have been discussed here, and identify new aspects of the NVE that can improve Internet conferencing efficacy.
Networked virtual environments (NVE's) are a type of virtual reality that are aptly suited for real-time conferences and seminars on the Internet. We have studied the use of a particular type of NVE, called a MOO, by scientists attending an international conference held on the Internet. Real-time interviews and electronically mailed comments indicated that the NVE was enthusiastically received. Criticism from the scientists helped identify many aspects of the NVE that can be modified to improve conferencing efficacy. The following are areas of concern when using an NVE for an internationally-attended Internet conference: providing input-output control management; automation of conference registration; control of anonymous or guest connections; simplification of NVE topography; methodology for recording discussion sessions, use of moderated rooms to compensate for network lag; and providing buffers to update late participants. NVE's catalyze the excitement and synergism that results when people communicate in real-time and hence provide an inexpensive means for effective international conferencing on the Internet.
We thank the Administrators of BioMOO and Jaime Prilusky, Head of Bioinformatics at the Weizmann Institute of Science, for their cooperation during the preparation and presentation of the VCC. A special thanks to Paul Hansen and the Together Foundation who provided the site for the VCC, and to Lisa Towell for first introducing us to LambdaMOO.
1. Available FTP: Hostname: parcftp.xerox.com Path: /pub/MOO/. Also, Pavel Curtis was recently recognized as among the top 50 influential people to watch in cyberspace. Staff (1995, February 27). 50 for the future, Newsweek, pp. 42-45.
2. Available http: lucien.berkeley.edu Path: /moo.html.
3. This was indicated in announcement NIST 52SBNB4C8231.
4. The output control manager has also been referred to as SMOC (Spam Manager Output Control; spam referring to unwanted text appearing on the user's screen). Gustavo Glusman has recently extended the SMOC such that all commands may be issued from the input mode, not just statements. Code for SMOC is available upon request from John Towell at m50jft1@hayek.cob.niu.edu.
5. The File Utilities Package (FUP) contains a program that is added to MOO server programs at compilation time. The latest versions of the File Utilities Package and the MOO server combined are available FTP: Hostname: bioinfo.weizmann.ac.il Path: /pub/software/MOOsupport/FUP File: MOOFUP1.6.tar.Z.
6. The code for this room was obtained from MediaMOO where it was programmed by Dan Rose. Available TELNET: Hostname: mediamoo.media.mit.edu 8888.
7. Although it was not used as such at the conference, the VCC is also a World Wide Web server and hence can be used to render real-time conferencing as well as World Wide Web capabilities. For more information, http: Hostname: bioinfo.weizmann.ac.il:8888.
Anderson, Christopher (1994). Cyberspace offers chance to do 'virtually' real science. Science 264, 900-901.
Bachrach, Steven M. (in press). Electronic conferencing on the Internet: The first Electronic Computational Chemistry Conference. Journal of Chemical Information and Computer Science.
Curtis, Pavel (1992). Mudding: Social phenomena in text-based virtual realities. Proceedings of the 1992 Conference on Directions and Implications of Advanced Computing. Available FTP: Hostname: parcftp.xerox.com Path: /pub/MOO/papers/DIAC92.ps
Evard, Remy (1994). Collaborative networked communication: MUDS as systems tools. Proceedings of the Seventh Systems Administration Conference (LISA VII). Available http: Hostname: www.ccs.neu.edu Path: /home/remy/documents/cncmast.html
Hardy, Barry J. (in press). Tales from the launching of global electronic conferencing in computational chemistry. Chemical Design Automation News.
Krieger, James H., & Illman, Deborah L. (1994, December). Internet offers alternative ways for chemists to hold conferences. Chemical and Engineering News, pp. 29-40.
Thornley, Blair (1994, July). Travels on the net. Technology Review, pp. 20-31.
Towell, Elizabeth Rose, Yim, Anna, & Lam, Terry (in press). Utilization of Internet Services and the Teaching of Internet in Business School. The Journal of Computer Information Systems.
Towell, John Foster, Hansen, Paul, Mercer, Eric, Leach, Martin, Rubin, Irit, Prilusky, Jaime & Glusman, Gustavo (November, 1995). Networked virtual environments and electronic conferencing. In S.M. Bachrach, D.B. Boyd, S.K. Gray, W. Hase, and H.S. Rzepa (Ed.), Proceedings of the First Electronic Computational Chemistry Conference [CD-ROM]. ARInternet: Landower, MD.
