Internet as a Professional Conferencing Arena Roger D. Smith Mystech Associates, Inc. Manassas, Virginia 22110 smithr@mystech.com 1.0 Introduction The simulation profession is spread all over the world and supports hundreds of models used to train and evaluate military operations and personnel. These are developed by the individual services, national laboratories, federal research and development centers, and commercial contractors. In this environment it is easy for development teams to become isolated and uninformed about what is happening in other areas. This results in duplicate work and duplicate mistakes which degrade the quality of training that can be provided and increase the costs born by the government, a.k.a. the tax-payers. In recent years several simulations have emerged as the primary training tools. These are organized under a program known as FAMSIM (Family of Simulations) at U.S. Army STRICOM. These have been or are being integrated to produce more realistic training environments. The ground (CBS), air (AWSIM), logistics (CSSTSS), electronic warfare (JECEWSI), intelligence (TACSIM), navy (ENWGS), and marine (MTWS) simulations will be networked to provide a single environment in which the strengths of each are joined together to produce an electronic battlefield which can not be replicated in any other way (figure 1). As the idea of interoperability between simulations grew the need for sharing knowledge, experience, and information increased. Unfortunately the funds and time to exchange this type of information continues to shrink. Schedules are so tight that it is often impossible for professional simulation builders to take the time to enhance their technical skills by exposure to conferences, courses, and meetings. Some method has to be found to keep these people current in their skills with limited budgets. One method that is used is sending a single representative to conferences and meetings. This person is then expected to bring the new ideas back to the project and brief others. In practice this often degenerates to discussions with close associates over lunch or "around the water cooler". The time to make formal presentations and the fact that many attendees do not have the speaking ability to convey the information conspire to diffuse good intentions. The Electronic Conference on Constructive Training Simulation (Elecsim) was conceived as an alternative to physical attendance at a conference. Given the global dispersion of simulation practitioners the conference was designed to bring quality presenters together with the dispersed audience at a virtual conference more easily and more often that could be done physically. We also hoped that the unique format would attract people who usually shun conferences as either uninteresting or inefficient. Elecsim was hosted on a single internet node and all information exchange was handled via E-Mail. Papers, graphics, and software were submitted by the authors and stored in a library. This library was made accessible to anyone with E-Mail access to internet, including dial-up users, CompuServe, America OnLine, Prodigy, and GEnie. All attendees then requested and received documents from this library. Finally, conversations between the authors and audience and among members of the audience was facilitated or mediated by the list server software. These tools allowed us to reconstruct the primary aspects of a professional conference within the internet domain. 2.0 Organization The original idea for an internet based E-Mail conference was born from the need of members of our simulation project to be knowledgeable about events transpiring in other parts of the simulation world. Having participated in several mailing lists and requested documents from list process libraries, it was a natural step for us to combine the recognized need and the network method to form a new application for list servers. Initial research into the idea led us to Dr. Tom O'Haver at the University of Maryland who had conducted such a conference for chemistry educators. Dr. O'Haver was able to provide background information and the locations of publicly available software to aid in the venture. 2.1 Selecting and Installing the Software The three most common packages for managing a mailing list and library are Majordomo, Listproc, and FTPmail. Studying all three of these we determined that Majordomo best fit our needs. Though Listproc seemed to provide more power and flexibility, the Majordomo commands were more intuitive. Since the simulation community is dominated by scientists, mathematicians, and military officers we wanted to provide as short a learning curve as possible in introducing them to internet conferencing. Majordomo was acquired via anonymous ftp and installed on a local area network of Sun Sparc workstations running SunOS 4.1.3 and sendmail. We then spent two to three weeks becoming familiar with the software and establishing mailing lists similar to those we intended to use on the internet node. Experimentation in this mode allowed us to make mistakes without impacting users who expect to have access to their internet accounts at all times. We created a new account and group to own and operate Majordomo. This isolated the mailing system from normal users who did not want to risk sharing their work with the outside world. Working on our own machine we also had unlimited access to root privileges to create users and groups, change aliases, and rebuild the Network Information Service (NIS). In reading the Majordomo instructions and examples all steps in setting up the system are not immediately clear. A reference language has developed among Majordomo users which is not common to those uninitiated to list servers and sendmail. As a result the writer sometimes slips into this language and becomes incoherent to new users. Several trials and errors were required before the meaning of some of the instructions were clear. As an example, one operating constraint of Majordomo that was omitted from the instructions was that all mailing lists must have lower case names. We started out using "Elecsim" and received several failure messages before being corrected by members of the Majordomo users group and changing it to "elecsim". The basic architecture begins with a directory identified as the home of the mailing lists. Under this, individual lists and document repositories are built. The Majordomo configuration file identifies these directories as the place to look for requested documents and command scripts which must be run. The mailing list itself is an alias. All incoming mail addressed to "listserv" is redirected to the command scripts which transmit documents from the library in response to the senders requests. All mail to a name such as "elecsim" is redirected to the file which contains a list of all E-Mail addresses subscribed to the conference. This is a mail reflector, sending copies of incoming messages to everyone on the list. Once the software was operating on the local machines we downloaded it to the internet host and installed it there. The system administrator supplied the root password and endured a few inconveniences as we created new users and aliases on his machine. Majordomo was completely installed and configured via a dial-up connection to the host, not once did our fingers or eyes touch this machine. This demonstrates the lack of geographic limitations even in creating and operating a list server of this type. Unfortunately, differences between the internet machine and our local machines resulted in additional problems. The most annoying was the incorrect format for the sendmail aliases for mailing lists. The Majordomo distribution comes with two different sets of examples for the format of these aliases. The one we chose initially and tested on the local machines worked only for local network mail. Again, recourse to the Majordomo users group exposed and solved this problem. Once running, the system manager changed the root password leaving us with our own user accounts, including the new majordomo account which owned the mailing lists and software. In this configuration we discovered a characteristic of SunOS 4.1.3 which we had missed when operating as root. Mail sent by local users of the internet machine to the majordomo account change the owner of the mailing list to "daemon.daemon" with "-rw-rw-r--" privilege. Once this happens neither the majordomo account nor our own accounts can edit the list and make manual changes, though sendmail is still free to work with the list. 2.2 Conference Standards Surmounting the software and network hurdles we were free to move on to issues concerning the conference itself. The first step was to establish standards for the information exchange that would allow all E-Mail users access to the document library. Though many academic works are distributed on the internet in PostScript format, this limits the readers to those with access to PostScript printers or document display translators and our targeted audience was much more varied than this. To serve them, we chose to store text and graphics in separate files with different formats. ASCII was the natural choice for text since it can be displayed on any screen and loaded into all word processing packages. We chose UUencoded GIF files for graphics. GIF viewers for almost any computer abound on the internet, CompuServe, AOL, and Prodigy at a very reasonable price (often free). A list of anonymous ftp locations for these software packages was taken from Dr. O'Haver's chemistry conference and provided to all participants. Authors also needed screen capture or format conversion packages to get graphics from DrawPerfect, PowerPoint, MacDraw, etc. into GIF format. Again these are widely available for all common hardware platforms. During the preparation phase none of the authors complained of the inability to produce graphics in this format. Had any software executables been submitted these too would have been UUencoded for delivery via E-Mail. UUencoding reads in any file and produces an ASCII output file which UUdecode can then translate back into the original file. This works for text, graphics, binaries, etc. Once encoded into ASCII the file can be delivered via E-Mail without being corrupted by the delivery mechanism. All papers and graphics were submitted via E-Mail to verify their transportability before placing them in the Elecsim library. 2.3 Finding Authors Since internet conferencing is still an experimental venture we opted to make attendance free to all simulation professionals and students. We further extended invitations to administrators and internet watchers who had other interests in the conferencing process. In order to make the idea a reality we contacted the major players in the field; described the idea; and invited them to join by presenting papers. The response was almost completely favorable with all but one project agreeing to submit a paper. The need to build an event of this type without a precedent practically dictated that the papers be accumulated by invitation rather than an open call to the community. And, since two days were being allocated for discussions of each paper, it was not possible to accept the dozens of papers that one would find at an event like the Winter Simulation Conference (WSC). Therefore, we were boxed in by the threat of paper flood on one hand and a drought on the other. The papers presented came from academia, industry, and government. It was very encouraging that academics who have a vested interest in publishing their works in "credible" journals and conference proceedings were eager to be a part of this conference. These cited the educational promise of the idea and the possibility of future recognition. As word of the conference spread we had requests from several others to be included. These we accepted as alternates should any of the invited papers cancel, which a few did. By the time the schedule was finalized we had papers from the following organizations: University of Virginia, University of Central Florida, U.S. Army STRICOM, Naval Research Laboratory, NASA Jet Propulsion Laboratory, Oak Ridge National Laboratory, Los Alamos National Laboratory, MITRE, Mystech Associates, SAIC, Loral, MRJ, Rolands & Associates, and VisiCom Labs. 2.4 Enlisting Sponsors In approaching professional associations with the idea of spon- soring the conference we found ourselves alternately attractive and repulsive to them. To some this idea captured their attention and motivated them to participate in the first of what could be many successful conferences in the future. To others the conference as a flash in the pan spawned by the current excitement about the internet. Typically, large, established societies had developed methods for planning and sponsoring conferences which were not conducive to the speed and organization possible in the electronic environment. Championing Elecsim to these was almost impossible, but hopefully the success of several such conferences will cause them to reevaluate their position. To our surprise, the opposition was not to the professional credibility of the conference itself but to the electronic nature of it. To quote the Vice-President of one large computer organization, "If the participants do not come together physically it is not a bonafide conference and we are not interested in being a part of it". In the end our sponsors were : The Society for Computer Simulation, The Military Applications Section, Operations Research Society of America, and The International Journal in Computer Simulation. Each of these printed an announcement, free of charge, in one of their publications. Separately the Military Operations Research Society agreed to publish the announcement even though they do not sponsor conferences other than those which they organize themselves. The primary contribution of society sponsorship was credibility and free announcement space. The electronic format negated the need for assistance with funding and the organization of hotel reservations, meeting rooms, airline fares, mailing announcements, and the publication of proceedings. 2.5 Promotion The primary form of conference promotion is usually the publication of announcements in journals and newsletters. But, for an electronic conference we found that other, less traditional, announcements drew more registrants. Electronic mail is an excellent medium for announcing electronic conferences. We began by compiling a list of all e-mail addresses from the business cards of those with whom we had professional contact. Then we broadcast the announcement to all accounts on the Defense Modeling and Simulation Information System (MSIS) and the Army Software Reuse internet nodes. We also posted the announcement on the MSIS Gopher which lists upcoming simulation events. These gave us a distribution of over 1900 people who were specifically or generally related to the conference topic. The next stop was the USENET news group "comp.simulation" which is distributed all over the world. Announcements here are also picked up and added to the monthly Electronic Simulation Digest. The announcement was then sent to the "Calendar of Events" section of IEEE Computer, IEEE Scanner, Communications of the ACM, Distributed Interactive Simulation - Joint Newsletter, Issues in Air Force Simulation and Analysis, Washington Technology, and Internet World. Following the conference, papers and presentations were prepared describing the event and its benefits. Once a participant is registered we continued to promote the conference to him/her. Big numbers on the attendance board meant nothing if the people registered did not participate and come to understand and appreciate the usefulness of the format. To remain in their minds we published an "Elecsim News Update" every two weeks which provided helpful hints on using the list server, introductions to the authors and topics, and descriptions on new documents and graphics added to the library. A living schedule was published along with the announcement and was continuously updated as authors changed the name of their papers or alternate papers were added. The library also included documents containing author biographies, paper abstracts, and attractive military graphics images. All of these were living documents right up to the start of the conference. Changes were driven by the dynamic nature of an experimental conference and by our desire to demonstrate the flexibility of the format. 3.0 Execution In creating a mailing list and the associated library, Majordomo allows several different types of access. The primary difference among them is the level at which membership and information is controlled. The loosest of these is the "auto" list where every subscription request is accepted automatically. The most strict is the "private" list where only members of the list may access information for it, and all membership requests must be authorized by the list owner. Between these extremes lie the "open" and the "closed" lists. Elecsim was handled as an "auto" list in which any user could perform any function. No abuses of this power were detected, but "open" lists will probably be used in the future. By the time the conference began we had over 400 participants, 15 papers, and 2 background essays. Upon registration every user received an information package which provided more details about the event. It described the steps for proceeding in conference participation and the internet locations for graphic viewers and uuencoders. As we mentioned, Majordomo was selected for its simplicity and we believe this approach paid off in users' ability to instantly begin accessing the conference. Few participants requested further help in operating the list server, and only a hand-full unsubscribed before the conference began. Following the conference, participants were asked to fill out an evaluation form. These have not been analyzed yet but the return rate was very high. The system was setup such that the list owner received error messages regarding mail for that list. These included failures to subscribe or the transmission of documents to non-existent e-mail addresses. On the other hand problems with the list server itself were sent to the owner of the majorodmo software. This created a clean break in dividing responsibilities and helped in locating errors during the set-up phase. As registration and the conference unfolded it became very obvious that a large number of the participants were internet and e-mail novices. They made very basic mistakes, like not knowing their own e-mail address. Those who use the internet actively, taking advantage of tools like Gopher and Mosaic, often forget that they form an elite among all who have internet access. Some of these elite wanted, and almost demanded, access to Elecsim materials via anonymous ftp or Mosaic. They did not understand that the conference was geared to the lowest common denominator - electronic mail. This may seem like "internet low tech" but it was intentional in order to reach the greatest number of people. The experiment was not in internet technologies, but in the ability of internet to serve people in intellectual need. During the conference all communication was geared around a few e-mail addresses (figure 2). As detailed "listserv@mystech.com" served as the point of registration and document retrieval. The documents included information on conference functionality and the technical papers on constructive simulation. The address "elecsim@mystech.com" was the mail reflector which allowed any individual to address comments to everyone attending the conference. Traffic to this address was extensive and could at times bury participants in e-mail. "elecsim-owner@mystech.com" served as the help desk. This reached the conference organizers who could answer questions not covered in the online documents. It was also the point of contact for providing comments and possible improvements, which were received and appreciated. Finally each author could be reached at his/her personal e-mail address which was provided in each of the papers being presented. Participants were instructed to send questions on each paper to the author. The author would then digest all questions, provide answers, and post a single response to "elecsim@mystech.com". Directing questions to the authors eliminated a few hundred messages from everyone's mailbox. Finally, we should point out that the creation and execution of the conference was carried out without ever sitting down at the keyboard of the host machine in Falls Church, Virginia. Nor was the responsibility of organizing it allowed to interfere with our regular work and travel schedules. As a result all of the work described above was carried out from Manassas, Virginia; Kileen, Texas; Orlando, Florida; Frankfurt, Germany; and Seoul, Korea. 4.0 Mistakes In organizing the conference we made several mistakes which we will not repeat in the future. These are enumerated here so that others may avoid them as well. 1) Perl was not installed correctly on the host machine. It seems there had been little call for Perl until Majordomo came along. This resulted in Majordomo failing because certain mandatory files could not be found. These types of problems can be resolved readily by referring to the O'Reilly book on Perl. 2) Lower case list names. We learned through trial and error that all Majordomo lists must be lower case letters. 3) In the announcement we requested that people provide the e-mail address which they wished to subscribe. "subscribe elecsim " This proved to be a problem for some users who did not know their e-mail address. If the address is not provided in the subscription request Majorodomo will use the address that the message came from. This should have been the advertised version: "subscribe elecsim"; with exceptions being handled by the list owner upon request. Some people also used their real name according to the format of the Listproc software. 4) Used the address "listserv" for Majorodmo. This caused confusion with people who had used Listproc under the "listserv" name. These were the users who tended to provide their human names instead of their e-mail addresses. Other lists commonly use "majordom" or "majordomo". Something unique to Majordomo, but indicative of the functionality would be preferable, such as "library", "registration", or "infodesk". This may also reduce confusion with the name of the list (next item). 5) People confused the functionality of the "listserv" and "elecsim" addresses. The first handled registration and document delivery. The second was a conference call to all registrants. As soon as the existence of the "elecsim" address was announced people began to use it to request library documents. In the future this name may be withheld until the very beginning of the conference when it will be used. 6) The "wrapper" program may be owned by any account that is a trusted user to sendmail. The daemon account is standard for this or the account which owns the Majordomo software. We selected the daemon account but found later that when adding e-mail addresses to the mailing list the ownership of the file was given to daemon.daemon. The protections were "-rw-rw-r--" which prevented other users such as majordomo and the list owner from editing the list to correct mistakes or remove incorrect addresses. It would have been better for the majordomo account to own the "wrapper' program. 7) Journal Typesetting Errors. Journals editors tend to see character strings as dictionary words and computer terms may prompt them to approximate to the nearest English equivalent. Therefore, "listserv@mystech.com" became "listserve@mystech.com". Upon detection it was too late to change the copy that had been sent to the printer. Therefore, this new address was added to the list of aliases which invoke the Majordomo software. 8) More Alternate Papers. Though the conference consisted of only 15 papers, most of those invited, we expected some of the authors to cancel their presentations because of other business commitments. To cover this we had two alternate papers in reserve. This number was too small for comfort. At one point four papers looked like they would not be written, though in the end we lost only two. For future conferences we plan to have a reserve to cover a 25% paper cancellation rate. 5.0 Benefits By far the greatest benefit of the conference is the ability to extend participation to people who are otherwise beyond the reach of traditional conferences. Attendees came from 16 countries: Australia, Austria, Belgium, Canada, Finland, France, Germany, Great Britain, Ireland, Portugal, Singapore, South Korea, Spain, Sweden, Turkey, and the United States (figure 3). Evaluations solicited after the conference indicated that a large number of the 400+ participants could not have attended the conference physically because of schedule conflicts and budgetary issues. The fact that these inaccessible people participated indicates that the potential to reduce duplicated simulation development efforts as a result of isolation does exist. Through exposure and the ease of electronic communication, relationships were developed that might not have been in a physical conference. These may lead to a greater degree of synergy and cooperation between the parties involved. For some, learning to use Internet for this conference will open this powerful form of communication for use in other areas. Though many had access to internet and e-mail they had not used it until the conference provided an event which motivated them. Since electronic conferences can be held on a much smaller budget and shorter time-lines, it is possible to convene them on specialized subjects on a regular basis. Physical conferences must serve a broad audience in order to attract sufficient attendance to cover the high costs involved. Electronic conferences have the potential to renew people's knowledge base more frequently since they are not tied to traditional budgetary constraints which dictate that individuals may attend only one or two conferences per year. An individual may participate in a new conference every week and it have little negative impact on their productivity. Finally, the electronic domain is a great leveling tool. Those who are normally quiet in a public setting are free to speak as forcefully as the boisterous, who are in turn muted to the volume of the rest of the audience. In an electronic conference no one is white, black, blind, deaf, or handicapped. All participate on even ground. 6.0 Future Plans With these conferences we hope to contribute to the growing electronic community. But, we want to facilitate the exchange of in-depth technical information in a more structured format than is accomplished via e-mail and newsgroups alone, which are much less formal and tend to serve more immediate needs than professional growth and development. A moderated conference could be created which runs continuously, much as educational television does. The moderator could organize submitted and accepted papers; publish the schedule; and host the server software. Such a project would require much more commitment than can be done via volunteers. An electronic conference may be scheduled in conjunction with a traditional conference in order to encourage more detailed discussions of the topic by any segment of the conference's physical participants. Such a supplement may also provide an alternative for distributing proceedings and collecting evaluations and suggestions. The traditional conference format has been shaped by experience and the limits of serving large audiences. Some of these limits can be overcome with electronic supplementals. Future electronic simulation conferences will attempt to use more of the internet tools available. The heart of the conference will remain in e-mail until more than that is available to simulation professionals around the world. But, e-mail can be supplemented by scheduled live talk using Internet Relay Chat (IRC). This portion of the conference would be limiting in that all participants must be able to access their computers during that time period all around the world. We would also like to create a paper that is an MPEG movie. This can be distributed via e-mail and viewed by the freeware MPEG viewers that are beginning to proliferate. Such a paper will probably be an exception in the 1995 conferences but may become more common later. The same may be attempted with audio files. The MBone could provide a combination of live talk and movies, allowing at least one-way transmission of papers being presented. This format is also very limiting in time and in hardware/software capabilities on the receiving end. We believe that a VR conference may be of interest, whether live or recorded. Something similar to Carl Loeffler's Networked VR Art Museum out of Carnegie Mellon University is possible. Though exciting technologically it is also limiting and its benefit over mail and some of the other methods needs to be demonstrated. Though we like to present grand plans for our own future, we suspect that reaching a wider audience with the current technology is more valuable that adding video, audio, and VR. We will strive to create and encourage the use of electronic conferences in more professions and on more specialized topics. Intending to reach people who are being missed by traditional conferences rather than trying to impress those few who are able to experiment on the leading edge. Bibliography Chapman, D. Brent, "Majordomo: How I Manage 17 Mailing Lists Without Answering "-request" Mail", Proceedings of LISA VI, Long Beach, NJ, 1992. Comer, Douglas E., Internetworking with TCP/IP, Volume I, II, & III, Prentice-Hall, Englewood Cliffs, NJ, 1991. Davenport, Thomas H., "Saving IT's Soul: Human-Centered Information Management", Harvard Business Review, Boston, MA, March-April, 1994. Krol, Ed, The Whole Internet User's Guide & Catalog, O'Reilly & Associates, Sebastopol, CA, 1992. O'Haver, Tom, Proceedings of ChemConf 1993, University of Maryland, College Park, MD, 1993. Rheingold, Howard, The Virtual Community: Homesteading on the Electronic Frontier, Addison-Wesley, Reading, MA, 1993. Wall, Larry and Schwartz, Randal L., Programming Perl, O'Reilly & Associates, Sebastopol, CA, 1991. Author Roger D. Smith is a Principal Simulation Engineer with Mystech Associates. He is responsible for developing simulations and tools to support the training missions of US and Allied forces. These have included air and ground combat models, intelligence collection and analysis algorithms, after action review systems, and simulation management tools. He has an M.S. in Statistics from Texas Tech University and a B.S. in Applied Mathematics from the University of Southern Colorado.