The Maturation of Computer-Mediated Communication by Gregory B. Newby Graduate School of Library and Information Science University of Illinois at Urbana-Champaign gbnewby@uiuc.edu ABSTRACT The communication norms of the major forms of computer- mediated communication are analyzed. These include electronic mail, mailing lists, Usenet newsgroups, virtual environments, interactive messaging, and information services. Then, the issues of network access are introduced, based on current trends in service provision by commercial enterprises. Forecasts for the future of information services and human communication by computer network are made. INTRODUCTION Computer-mediated communication (CMC) involves human interaction through the media of computers and computer networks. Today, at least 10 million persons worldwide use the Internet and associated networks to exchange messages and access information resources (1), and millions more employ home, office, or library-based computers not directly connected to the Internet. Because CMC is in many ways unlike traditional forms of human communication, such as face-to-face or telephone contact, norms for conduct on computer networks have been emerging and solidifying as user groups, services, and communication forums stabilize. This paper takes a broad look at the norms and uses which have emerged, and are still emerging, for various forms of CMC. It also examines some of the social and institutional forces which are shaping the way the global computer network is evolving. One goal of this paper is to take a step back from the Internet and its associated resources, and consider the extent to which the communication which is happening matches what we would like it to be. Another goal is to provide a prognosis for the future of the Internet as it becomes mainstreamed with other data distribution channels and institutions. The paper will also provide an overview of CMC for the new user, and a summary of service and resource types available now and in the near future. The future of the Internet is not certain, but many of the trends are toward a future that is far from ideal. As discussed below, the open forums for all people to communicate are not nearly as open as they could be; the tools for CMC available to the average user have not become much more sophisticated at all in the last 20 years; and the institutions and individuals that would like to get connected to the Internet find a chaotic, expensive, and ill-defined path to gaining access to the resources they hear about. In spite of the negative aspects of the Internet in its increasingly mature form, there is cause for optimism: people are able to cross geographical, social, and temporal barriers to communicate with others with similar interests, and scientists and researchers, at least, have better ability to communicate with each other. Unfortunately, the Internet in its mature form conforms much more closely to the restrictions and norms of the dominant non-CMC society than it steps outside of those pre-existing boundaries. CMC INTERACTION This section examines the major forms of CMC. There are thousands of discrete opportunities for people to engage each other, but they may be accurately grouped into a fairly small number of categories. Each type of CMC will be examined in turn. An earlier work (Rosenbaum & Newby, 1990) examined a similar list of forums. In that work, the qualities of each form of communication were listed. In this section, the focus will be on the norms and standards for communication which have emerged, in addition to the general nature and capabilities of each form. Email Electronic mail, or e-mail, or email, is a prototypical form of computer-mediated communication. Individuals can communicate with individuals or groups. Messages may be stored, forwarded, printed, or incorporated into other messages. Recent additions to email capabilities enable some users to send and receive multimedia mail, or to transfer data in non-text formats as part of an email message. The basic form of email is still text, though, and many email links and email software packages still reliably handle only text messages. A following section describes mailing lists, which is the mass-media form of email. Person-to-person email, and person- to-persons, is employed among people who know each other, or who know of each other, or when one person knows of another. Unlike office telephone or postal mail systems, email generally is routed directly to the mailbox of the person to who it is addressed (although a secretary or other person could screen the messages, just as for telephone or postal mail). This means that email may offer a more direct means of communicating with hard-to-reach persons than other media, and may also enable more communication across social strata. In fact, there is a good capability to identify and contact people famous in a particular area, people who you would have difficulty reaching on the telephone. Unfortunately, they may not return the contact -- making the "open channel" meaningless, since no messages are returned. Here, the social standing of the sender, if unknown to the recipient, may decide whether a message is returned: professors may not respond to undergraduate students, company executives may not respond to outsiders, and so forth. Because sending a message to an "open" forum such as a mailing list or newsgroup (below) constitutes some degree of openness to contact, and willingness to respond to messages of at least the topic of the message sent, people who post regularly to mailing lists and newsgroups are more likely to respond to a message "out of the blue" and indeed have some degree of social responsibility to do so. Although no research addresses this, it is likely that any personal email message will be responded to, at least briefly, than ignored. This is directly in contrast to sending out a broadcast request to a mailing list or newsgroup, which may go unanswered (especially if the question is a common or easy one!). Thousands of people who engage in public communication forums will be happy to engage in private one-on-one conversations on the side -- indeed, this is one of the qualities shared by mailing lists and newsgroups, that topics of great debate but peripheral interest may be relegated to personal email. Email is usually transferred rapidly from one person to another, although gatewayed mail may experience delays (especially to organizations connected by UUNet or store-and- forward networks such as Bitnet). In the past few years, mailer routing tables and gateways have gotten smarter, and the MX record format enables mail which looks like it is going to an Internet address to go to an address not accessible via IP. This is good for end users, as the necessity for explicit routing through particular gateways has been greatly decreased: email to almost any email address in the world will usually go through with (relatively) few difficulties in addressing. Although the transmission of an email message is typically rapid, email conversations are asynchronous, similar to conversations using the postal mail system. The communicating parties do not need to respond immediately to a message -- the message will wait to be read and responded to. The absence of visual cues limited capability for textual emphasis (underlining, boldface, etc. are not generally available) mean that messages may be misinterpreted if they are not clear. With time, people who communicate by email will gain efficiency in communication, so that it is easy to exchange terse messages during ongoing conversation and be understood. At the outset though, especially with email between strangers, there may be some thrashing about and misinterpretation before messages become clear. As for other forms of communication, the specific rules or norms for communication which a pair or group of communicators follows may gradually diverge from the general societal standards over time. This means that work groups, couples, or others engaging in substantial communication over time may have their own standards for communication -- e.g., messages may be shorter or longer, or may be only of a serious nature or only of a frivolous nature, and so forth. Although there may be flexibility and evolution in norms later on, these generalities apply to new conversations or among people who are not well-known to each other. They are adapted from Rosenbaum & Newby (1990), Shapiro & Anderson (1985) and various network sources, with components added from the common usage of email (as opposed to the ideal). - messages should have a single topic, or label multiple topic clearly - A salutation and brief introduction is desirable when contacting someone you do not know. - Messages should be kept fairly short (a few dozen lines) unless it is known that the receiver has a reason to read more. - Sarcasm and humor may be misinterpreted and should be labelled or otherwise used with caution. - If you are asking someone you do not know for something, state clearly what it is that you want to know (or that you just wish to start a dialog). It is hard for people who do not know each other to extract from a few words the intent behind them. - If possible, reference explicitly or incorporate small portions of the previous discussion to which you are responding (not all email software facilitates this). - Do not make assumptions about the capabilities or knowledge of the person you are addressing. E.g., do not send an encoded and compressed sound file to someone unless you know (a) they know how to unencode and uncompress the file; (b) they have the capabilities to play the sound file; and (c) they have the necessary disk space and network bandwidth so that the file does not overload their local resources. Some other rules apply to almost all email: - Include a return email address in the text of the message. Message headers to not always include a legible return address, and you can not rely on someone else's mailer to generate a correct return address for you. - Include a descriptive subject line. All email software supports this. Change the subject line if the subject has changed substantially from the original. - Forward mail judiciously: - Do you have the consent of the sender to forward the message? - Are the recipients interested, or may they have already received the message? - If you are uncertain whether a forward is in order, or the message is long, instead ask whether anyone would like you to send them a personal copy of the message. - Email almost always goes through (or you will get a message stating that it went through), but if you do not get a response within a few days, send a followup message. Unlike postal mail, there are few laws which apply to the privacy of email. However, many institutions have statements concerning email privacy. Ideally, these statements will indicate that email is private and will not be examined without just cause and due process. Unfortunately, many organizations (especially commercial organizations) state the opposite: that email is owned by them and may be examined at any time. The guidelines set down by Shapiro & Anderson (1985) still stand: avoid putting anything in an email message that you would not like to see on the front page of a newspaper the next day. Mailing Lists Mailing lists enable groups of people to communicate on topics of common interest. Messages are sent to a central email address and distributed to every subscriber's electronic mailbox. Since email is currently the lowest-common- denominator for global network traffic, mailing lists are the form of networked CMC which anyone who has access to some sort of network connection may access. Email can be transported through most gateways between networks undamaged, and explicit routing can help messages to find even the most remote network (Frey & Adams, 1991). There is a tremendous variety in the ease of use and capabilities of programs for reading and sending email. While some people have no difficulty processing dozens of messages in a few minutes, many others have difficulty with a few dozen in a week. This lack of sophistication in email reading capabilities leads to strong social forces for most mailing lists which: - frown upon very long email messages; - do not favor cross-posting of messages to multiple mailing lists; - insist that messages be kept on the topic; - prevent personal discussions, or discussions aimed at a small subset of subscribers; and - look for significant contribution from each message, e.g., messages of the form "yes, I agree," are not allowed. There is some variation in the protocols for each list, but not as much as for other forms of CMC. Furthermore, different lists may have very different standards for the level of traffic -- both the length and the number of messages sent. Of the thousands of publicly accessible mailing lists, a sizable minority are moderated. Modearation enables a single individual or small group to insure the appropriateness of each message sent to the group. Ten to 20 new mailing lists are created every week (2), but many of these quickly fade in popularity so that only occasional messages are seen. Of the active mailing lists, 10-30 messages per week is the norm, but with many lists generating that many every day. Mailing lists do not usually generate as strong a sense of community as is found in other forms of CMC. Mailing list subscribers tend to receive more than one list, and are much more likely to be a recipient, rather than an active sender, of messages to the lists they read. The decreased sense of community on most mailing lists may be partially because messages that arrive in a mailbox are things that happens to you (once you have subscribed) as opposed to other forms of CMC which require you to do something every time you want to participate. Mailing lists are used internally in many organizations, also by closed groups. On these, people may know each other and exchange messages using other means than mailing lists. For those, the rules that dominate the interaction will necessarily be inherited from the group. In the case of most public lists, though, the group communication will tend to be more akin to a mass-broadcast mentality than a personal communication mentality: people will write as though their message were an "announcement," instead of a message to a well-known recipient group. There are many mailing lists which have more of a distinctive personality or cohesiveness, but these are the minority. Usenet (and other BBS) Usenet, or Netnews, is a worldwide distributed bulletin board system (BBS) consisting of over two thousand discussion groups, called "newsgroups." Thousands of other groups are used only for local or regional discussions. Unlike email, Usenet messages are not sent to every "subscriber" to a given newsgroup. Instead, the user employs a sophisticated news program such as "nn" or "rn" to access the store of Usenet articles and peruse the contents. These interfaces enable people to rapidly scan through hundreds of messages, only choosing to read a few of interest, with great ease and little overhead. Since only one copy of each message needs to be kept for each news server (typically, an institution will have one news server, where the Usenet data are stored), and the interface is designed specifically for scanning through many newsgroups looking for subject lines indicating an interesting article, there tends to be a larger number of followup messages with fewer social restrictions on the size and relative merit of each message than for mailing lists. Usenet communication is asynchronous. Messages are propagated from server to server in a way such that hosts which are further away (more network hops) from the originating host will take longer to get a message. The order in which messages are stored on a server may not be the order in which they were created. Usenet is transferred over a variety of channels, including the Internet and UUNet. Since some of these are not connected 24 hours, propagation delays result. Delays are on the order of a few hours to a day or two. Most Usenet newsgroups are very informal, but individual groups have their own sub-cultures which may place social restrictions on the interaction. There are a small number of moderated newsgroups on which interaction may be more formal, but the majority of newsgroups are not moderated, so that anyone who can read the group can also post to it. Many newsgroups have a very large number of both dedicated and occasional readers. This leaves them open to new users, who typically do not identify themselves as such when they post. New users who do proclaim their ignorance of the group content are much rarer than on mailing lists, since anyone who wants to post has very easy access to recent messages on the list (unlike mailing lists, where people may subscribe, wait a day without receiving messages, then send a "what's going on?" query to the list). Some Usenet newsgroups have well-defined sub-cultures who may be less open to outsiders. These include highly technical topic areas and groups with sensitive discussions (e.g., talk.rape). On most newsgroups there are a few highly vocal members who tend to keep discussion rolling, provide answers to questions, or antagonize. Many Usenet groups distribute over 1000 messages per week and some reach over 10,000 readers. The Usenet newsgroup news.lists includes many statistics for Usenet traffic. The anonymous FTP site pit-manager.mit.edu keeps archives of these and other lists (cf. pit- manager.mit.edu:pub/usenet/news.lists). Usenet is not accessible by everyone who has access to the Internet. It is generally found only on UNIX machines, although packages for the Macintosh and VMS exist. Many of the topics of discussion found in mailing lists are duplicated on Usenet, sometimes with little overlap in subscribership. Another group of mailing lists is gatewayed to Usenet, and a few newsgroups are also distributed by the moderators as email. Discussions on the same topic will tend towards many more messages on Usenet than on mailing lists, but the messages will be in a conversational mode -- with one message adding slightly to a prior message. With mailing lists, what is added to a prior message tends to be more lengthy, with an ability to stand on its own. Interactive Messaging There are several different methods for communicating in real time with other computer network users. These include Internet Relay Chat (IRC), Relay, Bitnet line messaging, and the "talk" command. They all have in common that what one user types will be displayed with very little delay to the recipient(s). However, the uses for these methods differ. IRC and Relay are the most similar of the group. They both involve line-messaging, where a user types a line of text and sends it to a central site. Then, the message is distributed to all the users who are listening to that "channel" at that site. The discussion is similar to a citizen's band radio channel: users may find different discussions on different channels. Some channels have standing discussions, and both IRC and Relay offer the capability to form a private channel. The primary difference is that IRC operates over the Internet, and Relay operates over Bitnet. Both have multiple "servers" on which standing discussions may be found (dozens of public servers on IRC, about 1/2 dozen on Relay). IRC and Relay are most frequented by undergraduate students. They identify others with similar interests and engage in conversations with them. Different channels may reveal discussions on topics as diverse as politics, calculus, or current movies, with a definite tendency towards less academic topics. Users assume an "alias," similar to a citizen's band radio "handle," by which their messages are identified. The handle may promote a certain image or invite a particular response, e.g., "Gypsy Queen" or "Lonely Heart." Significant relationships have been created and supported by both IRC and Relay (Reid, 1991). Definite communities and sub-cultures exist in the IRC and Relay worlds. The participants are often very frequent users, perhaps spending entire evenings and late nights online several nights per week. Various shorthand methods for signaling affect are used, and users are adept at following multiple chains of discussion at once, reading others' responses while typing their own. People may sometimes insult each other or gang up, but most groups are very receptive and helpful to newcomers (Reid, 1991). IRC software exists for UNIX, VMS, and some other computer types, and requires Internet connectivity. Bitnet line messaging is the transport mechanism for Relay. Users form a single-line message and send it to a target username on a Bitnet node. The Relay program uses Bitnet line messaging. IBM's mainframes running CMS and DEC computers running VMS which are connected to Bitnet have line messaging capability. People who know each other, and who are both logged on at the same time, can exchange line messages. Compared to Relay, where plenty of people can keep the conversation going, line messaging with a single person can be tedious. Delay is introduced as each sender types in a message, and then there is a delay (usually a few seconds) as the message is transmitted to the receiving Bitnet site. This form of communication is used best for short messages to a known person. It may also be used to interact with LISTSERV and associated mail and file server programs (3). Bitnet line messaging may be used to send a message to people who are not known to the sender, in the hopes of starting a conversation. Although some may welcome this, most users do not -- such usage is frowned upon socially, as well as forbidden by CREN bylaws (4). The last form of interactive messaging to be mentioned here is the "talk" command. This works in the UNIX environment and some installations of VMS, and requires an Internet connection. "talk" sets up an interactive "conversation" with a remote user where everything that each user types appears on the other's screen as it is typed (mistakes and all). This is not line messaging, as with Bitnet, but character-at-a-time full-screen messaging. Most "talk" programs allow more than two people to communicate, but beyond 4 users the screen becomes cluttered. Like Bitnet line messaging, "talk" is most used for short conversations with someone you know. Longer or more detailed messages would probably be sent by email. However, the "talk" command, because of the immediate nature of the display, is not so limited as line messaging for having a real conversation. Friends who are separated in distance might find enjoyment in using "talk," in that it seems more like a telephone conversation than email. All forms of CMC interactive messaging are best suited for interactive discussions on non-complicated topics. What is typed is especially open to misinterpretation, and it may take seconds or minutes before participants come to agreement on what is being said. For more detailed interaction, another forum is indicated, such as the telephone or email. MUDs & other Virtual Environments Several score multi-user interactive environments are accessible for remote login over the Internet. Most of these are games in the tradition of role-playing and adventure, but several are more professional or academic. MUD stands for "multi-user domain" or "multi-user dungeon." Most are based on plain text interfaces, but one or two currently offer graphics. In a MUD, the user assumes a character and engages other characters in whatever interaction is appropriate for that forum. Each MUD is based on a motif, typically a swords- and-sorcery scenario, but classrooms and spaceships are also found. MUDs are based on physical environments -- textual descriptions of roads, paths, buildings, other characters (including some which are played by the computer) and so forth are used as a basis for the user's navigation through the environment. Users login to a MUD and can interact with other players by (essentially) sending line messages, but augmented by a variety of simulated affect and body movements: Estragon bows. Estragon says, "I am pleased to meet you, and would be happy to discuss our common interests in this quest." Estragon looks intently at Vladimira. MUDs have some well-established rules of conduct, and offer an unusually wide range for self-expression in a CMC environment. Frequent players may have a few MUDs that they favor and strive to advance in, but also visit other MUDs. Of the 1/2 dozen or so major variants on MUD software, all enable users to create their own portions of the game (e.g., a magic castle). Some MUDs offer some creation capability to new users, others require ascension to a high level of experience first. Unlike most network-based activities for group participation, especially those mentioned in the following section, MUDs are typically run without the participation of a system administrator (but some sufferance by the sysadmin is required). MUD programmers can download the basic MUD software and game configurations by anonymous FTP, then install the code and create their own game area. MUD software is self-contained, including the maintenance of usernames and passwords. This enables the MUD administrator to leave the MUD software running and listening at a specific UNIX port (5) for incoming traffic. By choosing a port open for general use by all system users, no administrative intervention is required to maintain the MUD usernames (these are independent of the general system usernames). Like Relay and IRC, most MUD users are undergraduate students. Also high-school students and some graduates. MUDs are a large part of the social lives of many players, who may develop strong relationships with other players. For some players, there might be a fuzziness between the role-playing character and real life. This is typical of non-computerized role-playing games as well (Fine, 1983). Mass-Broadcast Media New network users are often exposed to Gopher, and told about anonymous FTP, Archie, WWW, and WAIS. They may also hear about electronic journals and file servers. This collection of resources, and a few others, constitute the information provision services of the network. Some require Internet access (WWW, anonymous FTP, Gopher), but others enable email access as well as interactively through the Internet (Archie, WAIS, mail and file servers. Anonymous FTP mail servers are also available). These forms are necessarily asynchronous, and usually involve people communicating with an unknown audience across time and space. As discussed below, there are only limited capabilities for individuals without system administration connections to create mass-broadcast information sources using these tools. There are technical standards that enable the tools to exist, but relatively few standards for communication indigenous to CMC environments. Instead, the resources tend to be modelled after existing forms: technical documents, paper journals, databases, etc. New network users may have difficulty in using these resources because the access mechanisms seem very different from familiar methods. In fact, though, the real problem is that the tools do not provide the same sort of information about what each, say, menu item contains as people expect. Without a good model of network contents drawn from experience, users do not know when to expect a technical document, another menu, a one-line citation, or some source code. Or to be logged in to another resource. Or to suddenly leave "gopherspace" and be put into "ftp space." If the interface and underlying mechanisms are different, but the contents are generally familiar, it should be incumbent on the information provider to give cues which help users to employ their pre-existing skills for non-networked resources in CMC data access situations. This section has not included the various forms of database access which exist. Although a case may be made that they are asynchronous forms of broadcast communication, a discussion of databases, WAIS servers, and so forth would probably not be illuminating. There are also a number of hybrid forms of communication which included some aspects of CMC, including fax and fax-to-email communication, voice mail, teleconferencing, computer-aided instruction, and groupware. The current discussion has limited itself to what might be called "pure" CMC activities, but recognizes the Internet- based radio, video, and multimedia email (Borenstein, 1992) are removing some of these distinctions. THE TOOLS FOR CMC AND THE WIDENING GAP BETWEEN THE HAVEs AND THE HAVE NOTs Most of the Internet gurus and standard makers, including many of the computing support staff at Universities, businesses, and government institutions, have desktop UNIX, X- Windows, MS-Windows, or Macintosh workstations, often connected directly to the Internet via high-speed local area network (LAN, e.g., Ethernet). The creators of the tools of tomorrow are high-end users. They have high-speed network connections to their desktop, plenty of disk space, and the most sophisticated client programs available. Given the history of computing and telecommunication, it is perhaps justified that work progresses on gigabit-per- second computer networks in the U.S. while most countries in the world do not have reliable electricity or telephone systems. However, there are still tens of thousands of institutions in the U.S. for which Internet connectivity is as yet out of the question, either because the expense of the connection to the network is too great or because the necessary internal infrastructure -- LANs and computers -- does not exist. The low-end Internet user of today has comparable equipment to a high-end user of 15 or 20 years ago: a vt100 compatible terminal or terminal-emulating microcomputer (text only) with a 1200- or 2400-baud modem connection to a remote host. This is adequate for text editing, command line interaction with remote resources, and access to many of the popular services such as Gopher and WAIS. What the low-end user can not access is growing. This includes more sophisticated interfaces to WAIS, Gopher, anonymous FTP, and WWW, as well as weather maps, simulations, and page display applications. Both a graphical display capability and a local capacity for IP networking is required for many applications currently available or under development. This reality creates some difficult situations for Internet trainers or those seeking to connect institutions to the Internet. One talks of point-and-click interfaces to WAIS, and for FTP, but these do not work in vt100 environments. A T1 link to the Internet may get overloaded (along with the local disk drive) if many users are employing remote resources. The inability of most computer users to download and compile Internet-accessible resources means that they will hear about things they do not have an immediate capacity to use. As discussed in the next section, network users are handicapped by a lack of system administration access and software installation skills to use some of the resources available or under development. Disk space is only one of the final frontiers which network users encounter between themselves and the applications they hear about, frontiers which connectivity alone is insufficient to cross. Who Can Create and Provide Information Resources? Ultimately, any individual can create any sort of information resource -- software, a listing of services, a paper, or whatever -- and make it available to other Internet users. Tools such as Gopher (discussed below) make resources available globally. In fact, though, the most useful distribution channels for information resources require the cooperation of a system administrator to create and maintain. These include dedicated usernames for remote login, anonymous FTP sites, Gopher servers, and WAIS servers. A "regular" network user can not decide they want to make something accessible via anonymous FTP, for example. Instead, they need to identify someone with control over an FTP site who is willing to store their file. Sometimes this needs to be solicited, other times someone may volunteer (as when a resource is posted to a mailing list or newsgroup). Although the creation of a new FTP directory and storing a user's file there is nearly trivial, an FTP site with several thousand files can be no small task to administer. This typically leads to an environment where frequent updates of materials is not desirable, and culling of older materials seldom occurs. Individuals are empowered for information access on the Internet, but are not nearly so empowered for information provision. Depending on the friendliness and accessibility of the local computer administrators and the degree to which an individual's interest area seems appropriate to the administrators, it may not be easy at all to distribute resources via the Internet. Mailing lists and Usenet group archives are still only semi-accessible: archives, when available, are not always easily searchable and almost never topically indexed, yet these are really the only places where everyone can put information (other than their own username's directory, of course). If you ever wondered why there seems to be so much information created by and for the more computer literate user population, the first obvious conclusion is that those are the users of the Internet. This is of course no longer the case: Internet users come from all backgrounds, as demonstrated by the diversity of mailing list and Usenet newsgroup topics. A remaining impediment to a ratio of non-computer-related information which better matches the user population is the barrier to information provision using existing tools by those not well-connected with their local computer administrators. The solution to this inequality is not immediately forthcoming. It requires more sophisticated information technology on the desktops of people who do not do number- crunching, and departmental computing resources in departments currently at the bottom of the list for network connectivity in academic, business, and government institutions. In the meantime, there are some large- and small-scale information resources appropriate to workers in the humanities, social sciences, literature, and so forth, and a number of friendly system administrators to continue these resources. The number of FTP or Gopher sites providing literature or literary materials, for example, is still not nearly in proper proportion to the number of sites related to computer software -- yet compare the number of English to Computer Science majors at any university. Economic Realities: The Commercialization of the Network Internet historians can trace the path of the Internet, as we know it today, from early days under the Defense Advanced Projects Research Agency (DARPA) to the modern days of NSFNet, with funding primarily from the National Science Foundation. During this twenty-year history (Marine et al., 1991), the Internet as such became a collection of networks, all capable of using the Internet Protocol (IP) and associated transport protocols (TCP, SMTP, etc.) to communicate with other Internet hosts. During the 1980s, a large number of regional and national networks emerged around the world, each with its own administrative structure. Many of these used IP and maintained gateways to the rest of the world, and were therefore logically part of the Internet. Other types of networks, including Bitnet (which, like the Internet, has links to other networks with similar store-and-forward methods for data transport in other countries), Fidonet, and UUNet flourished and, at least to some extent, developed the capability to exchange data with IP-based and other types of networks (Quarterman, 1990). Computer networks, within and between institutions, have become mainstreamed technology. Shrinkwrapped LAN solutions enable even the smallest institution to support internal computer networks, and a growing number of Internet service providers (in the U.S.) make getting connected to the Internet only a matter of willingness to pay an appropriate monthly fee. Individuals can get some level of Internet connectivity as well through a variety of commercial providers, many of which provide value-added services such as email and online database access. The success of IP networking is the success of the NSFNet experiment. Currently, the NSF is decreasing centralized funding for the U.S. data superhighway, and instead seeking cost recovery and second-party providers for additional Internet connections and expansion (notably Merit, Inc. and Advanced Networking Services or ANS). NSFNet funding seems to be headed for high-end goals, such as gigabit per second transport speeds and a national metacomputer (Smarr & Catlett, 1992). There has also been a quiet transition in most areas of the U.S. away from the mid-level service providers toward commercial providers as the source for Internet access for new Internet connections to both commercial and academic institutions. What does all this mean? It means that the days of the Internet as a research-oriented network where commercial gain is not permitted are rapidly coming to an end. It means that the distinction between commercial and non-commercial networks and traffic has faded. Indirect evidence to this effect includes the number of subscription-based services that are being distributed over the Internet (including NSFNet and the mid-level networks proper -- there is very little attempt to route commercial traffic via commercial networks). These include newsletters and Usenet hierarchies. More direct evidence may be found in the recent assignment of several businesses including American Telephone and Telegraph (AT&T) as Internet directory providers. If one logs into the new service provided by AT&T (telnet to ds.internic.net, follow login instructions), he or she will see that AT&T will provide a free yellow-pages entry for any business up to 16 lines long. For a fee of $300US / month, the business may purchase several pages for their entry. Compare this NSF supported use of the Internet to the NSFNet Acceptable Use Policy (Appendix A), especially paragraph 7. This example is one of the more insidious due to the direct involvement of the primary funder of Internet activities in moving the Internet toward commercial purposes. The future of "free" services on the Internet appears to be gloomy. In the next section, services and resources for Internet information will be considered as they related to these changes. Free and Fee Services: Is There Room for Both? The 1990s to date have seen the rapid growth of more than just the number of Internet users and connected computers. It also has seen the proliferation of several tools for using the network more efficiently. The most notable of these are Archie, WAIS, WWW, Hytelnet, and Gopher, with their associated enhancements (e.g., Veronica for Gopher, and Mosaic for WWW). Hundreds of library card catalogs (OPACs), dozens of campus- wide information systems (CWISs) and numerous databases of various types have also been made publicly available during this short time period (6). Something fascinating has occurred during this time -- the big picture is most readily obtained through several examples. First, consider Archie. Archie is a tool for finding items available via anonymous FTP. It access the directory listings of more than 600 public FTP sites around the world (many more exist, but Archie does not access all of them). Since Archie has become available, the redundancy of files available via FTP has decreased markedly. It used to be that any of several dozen "good" FTP sites could be accessed and most of the well-known FTP resources would be available. A recent Archie search for the most recent version of Emacs, for example, revealed only three sites in the U.S. which possessed it -- and typically zero or one site in other countries with good Internet connectivity. This decrease is certainly due to several factors, including the increase in the number of items available worldwide via FTP. A reason which seems intuitive, though, is that the various sites do not have reason to be altruistic about network service provision, and instead are happy to rely on others to store model collections. A second example is WAIS. WAIS is the brainchild of Brewster Kahle, Craig Stanfill, and others at Thinking Machines Corporation (TMC). In 1992, Brewster left TMC and started his own company, WAIS, Inc. The Coalition for Networked Information Discovery and Retrieval (CNIDR) became the home for the "free" version of WAIS, and WAIS, Inc. immediately announced more sophisticated capabilities for their commercial product than "freeWAIS" offered. A final example is Gopher. The University of Minnesota announced in early 1993 that Gopher software would be subject to a $5,000 - 10,000US licensing fee for commercial uses. Although businesses and other users could still employ Gopher free of charge for free information distribution, any for-fee service provision would require licensing. Archie, WAIS, and Gopher are three information services which have greatly eased information access on the Internet. Each in their own way, though, they provide example of the increasing distance between free and for-fee Internet resources. Let us now turn to the information itself. There are a large number of one-time listings, directories, resource guides, and so forth which are created by individuals or small groups. This is the essence of the Internet as we know it: individuals adding to the collection of available resources. There are two unfortunate facts of this type of service, though. The first is that these resources are often not updated (although many are, of course). This is only important because of the second: the resources tend to be recirculated and stored on Gopher servers, FTP sites, etc., and to not go away when their information is no longer current. Since we do not yet have good methods for knowing how old a file is or whether it was updated, unless the file has some lines describing an update history, non-authoritative resources may be perpetually redistributed. They may even prevent others from creating a more up-to-date resource. In the meantime, many of the best resource providers (in terms of completeness and maintenance) have published their works via non-electronic means. Some continue to be updated and made freely available in electronic form (e.g., Strangelove & Kovacs, 1993), others are no longer provided at all on the Internet (e.g., Kehoe, 1992). Another set of authors publish materials about the Internet that would make excellent Internet-based resources, but are not provided over the Internet. The role of the publishers in Internet-based information provision is not yet clear. While future sales of print materials may be inspired by free extracts -- or even the full text -- provided on the Internet, publishers fear that this may also reduce the number of customers willing to pay money for a paper version if the electronic version is comparable. The role of information creators, though, seems more clear: there is less inspiration to create and maintain high- quality information for free distribution on the Internet. The tools for navigating network information resources create a false feeling that 1) everything is out there, if you take the time to find it, and 2) the information available is up- to-date and authoritative. This section has looked at the growing differences between information services and resources on the Internet which are free or for-fee. The public library model of information access, where almost everything is free to the patrons, does not seem to be under development on the Internet. Instead, it is likely that the number of fee-based resources will continue to grow (with the backing of primary Internet policy shapers), and individuals will find themselves, at least in the short term, with less inspiration to create their own free resource. In the long run, there seems to be room for both free and for-fee resources. There are a number of publishers investigating methods for the distribution of free information via the Internet, and (despite the gloomy facts in this section) continued emergence of new free resources. In the short term, though, Internet users may find that the concept of "value added" for information resources is strongly associated with "for-fee." NREN NREN legislation as passed in 1991 (U.S. Congress) gives about one one-hundredth of the funding to education as to research and science efforts such as the Grand Challenges. "NREN" would more properly be written "NReN," given the emphasis in the near term on connecting schools, libraries, and the general public outside of high-tech corporations and academe. There is hope that networking will be made available to a much larger portion of the population -- those who are currently the information poor -- in the form of new legislation under consideration. Non-government projects, such as the Free-Nets, are already active in reaching out to these populations. There seems little doubt that some sort of computer network access will be ubiquitous in the U.S. home, office, and school by the end of the millennium, but there is cause for concern over who will provide that access and what form it will take. In the government sponsored race to provide high- technology infrastructure to the U.S. population, the current leaders are the telecommunications and cable industries. The nature of CMC may shift dramatically if commercial enterprises shape the services and communication methods of the networking future. The shift does not need to be a negative one, but it might result in a perpetuation of Bitnet, the Internet, and the associated networks and services as a service to the information and education elite -- and possibly relegate them, eventually, to footnotes in the history of telecommunication. The role of commercial enterprises in the near future for the provision of computer networking and information services is the wild card in more than the nature of CMC -- it has the capability to turn the good intentions for education and international competitiveness on their heads, creating a generation steeped in virtual reality MUDs and pay-per-minute information services that is even less likely to read a book than the Nintendo(tm) generation. CONCLUSION This paper has not focused on the information resources of the Internet, instead taking a human communication approach. Newcomers to the Internet are shown data resources through Gopher, databases, and so forth. However, if the analysis in this paper is correct, there will be a decreased emphasis on Internet-accessible resources which are both of a high quality and free. What this leaves the Internet with is its greatest resource: the people who use it. Interaction with others through computer mediated channels -- mailing lists, newsgroups, interactive media, and methods to come -- will remain the sure thing during the maturation period the global computer networks are in. These numerous communities will continue to produce various information resources of pertinence to them, even as commercial enterprises step in to provide monolithic services with appeal to a wide market. The box containing the new forms of human communication which have emerged with the growth of computer networks can not be closed, and it does not seem likely that the paths of change from a free non-commercial network will be diverged from. As the social forces associated with these two phenomena interact, current network users will be surprised at the decreased freedom for interaction and information access they have, but will also find a host of new acquaintances waiting to be made. * * * * * * * * * * * * * * * Notes: 1. The estimate of 10 million is gathered from network statistic sources such as the anonymous ftp site at ftp.merit.edu and the Usenet newsgroup news.lists. It is unclear whether these numbers reflect the number of people with network access, or just those who actually use the network. In mid-1993, the total number of hosts accepting IP packets is about 1.5 million. Many other hosts are accessible via email only, or through corporate or government firewalls. My estimate is that the total number of people with at least email access to Internet resources exceeds 20 million persons. 2. As announced to the Listserv-maintained mailing list new- list@vm1.nodak.edu. To subscribe, send a message of this form to listserv@vm1.nodak.edu: subscribe new-list Your Name. 3. A typical example is to subscribe to a mailing list, e.g. by typing (on an IBM computer running CMS): TELL LISTSERV@NDSUVM1 SUBSCRIBE NEW-LIST GREGORY B. NEWBY. 4. CREN is the governing body for Bitnet. Bitnet is the proper name of the U.S. portion of the world-wide network using the same transport mechanism (based on IBM's Remote Spooling and Console System or RSCS), although the term is often used to refer to the network as a whole. 5. "Ports" are used by Internet hosts to communicate with each other. Well-known ports are employed for such processes as email (port 25), telnet (port 23), and FTP (port 21). Under UNIX, ports under 1024 are only accessible by privileged programs, but ports over 1024 are available to any user program. MUDs use ports such as 9999, 4444, 6916, etc. 6. The Hytelnet software offers an excellent and up-to-date access point for many of these resources. Software available for many computer platforms via anonymous FTP: ftp.usask.edu:pub/hytelnet. References Borenstein, N. 1992. A User Agent Configuration for Multimedia Mail Format Information (RFC 1344). Available: nic.ddn.mil:rfc/rfc1344.ps or rfc1344.txt. Fine, Gary Alan. 1983. Shared Fantasy. Chicago, Illinois: The University of Chicago Press. Frey, Donnalyn; Adams, Rick. 1991. !%@:: A Directory of Electronic Mail Addressing and Networks. Sebastopol, California: O'Reilly & Associates. Kehoe, Brendon. 1992. Zen and the Art of the Internet. Englewood Cliffs, New Jersey: Prentice Hall. Strangelove, Michael; Kovacs, Diane. 1993. Directory of Electronic Journals, Newsletters and Academic Discussion Lists 3rd Ed. Washington, DC: Association of Research Libraries. Marine, April; Kirkpatric, Susan; Neou, Vivian; Ward, Carol. 1992. Internet: Getting Started. Menlo Park, California: SRI International. Newby, Gregory B. 1993. Directory of Directories on the Internet. Westport, Connecticut: Meckler. Quarterman, John. 1990. The Matrix. Bedford, Massachusetts: Digital Press and Prentice Hall. Reid, Elizabeth M. 1991. Electropolis: Communication and Community on Internet Relay Chat. (Thesis). Melbourne: University of Melbourne. Available: ftp.ee.mu.oz.au:pub/text/IRCThesis/electropolis.ps.Z or electropolis.txt.Z. Rosenbaum, Howard; Newby, Gregory B. 1990. An Emerging Form of Human Communication: Computer Networking. In: Proceedings of the American Society for Information Science Annual Meeting Volume 27. Medford, NJ: Learned Information. Shapiro, Norman Z.; Anderson, Robert H. 1985. Toward and Ethics and Etiquette for Electronic Mail. Santa Monica, California: The Rand Corporation. NSF Report R-3283-NSF/RC. Available: rand.org:pub/old/Reports/r-3283.ps or r-3283.lpr. Smarr, Larry and Catlett, Charles. 1992. Metacomputing. Communications of the ACM 35(6): 44-54. U.S. Congress. 1991. The High Performance Computing Act of 1991. Appendix A: NSFNet Acceptable Use Policy (Available: ftp nnsc.nsf.net:nsfnet/netuse.txt) Interim 3 July 1990 NSFNET Acceptable Use Policy The purpose of NSFNET is to support research and education in and among academic institutions in the U.S. by providing access to unique resources and the opportunity for collaborative work. This statement represents a guide to the acceptable use of the NSFNET backbone. It is only intended to address the issue of use of the backbone. It is expected that the various middle level networks will formulate their own use policies for traffic that will not traverse the backbone. (1) All use must be consistent with the purposes of NSFNET. (2) The intent of the use policy is to make clear certain cases which are consistent with the purposes of NSFNET, not to exhaustively enumerate all such possible uses. (3) The NSF NSFNET Project Office may at any time make determinations that particular uses are or are not consistent with the purposes of NSFNET. Such determinations will be reported to the NSFNET Policy Advisory Committee and to the user community. (4) If a use is consistent with the purposes of NSFNET, then activities in direct support of that use will be considered consistent with the purposes of NSFNET. For example, administrative communications for the support infrastructure needed for research and instruction are acceptable. (5) Use in support of research or instruction at not-for-profit institutions of research or instruction in the United States is acceptable. (6) Use for a project which is part of or supports a research or instruction activity for a not-for-profit institution of research or instruction in the United States is acceptable, even if any or all parties to the use are located or employed elsewhere. For example, communications directly between industrial affiliates engaged in support of a project for such an institution is acceptable. (7) Use for commercial activities by for-profit institutions is generally not acceptable unless it can be justified under (4) above. These should be reviewed on a case-by-case basis by the NSF Project Office. (8) Use for research or instruction at for-profit institutions may or may not be consistent with the purposes of NSFNET, and will be reviewed by the NSF Project Office on a case-by-case basis.