Footnotes: Medium as Process


Bailey (1978), for instance, offers concise overviews of the advantages and disadvantages of a number of research methodologies, including surveys, interviews, experiments, field studies, indirect observation, ethnomethodology, content analysis, and other methods.

Computer Conferencing and Communication Theory

The term "Global Village", frequently quoted but rarely cited, is a compact restatement of the central thesis of McLuhan's (1964) introduction to Understanding Media: The Extensions of Man: "After three thousand years of explosion ... the Western world is imploding. ... after more than a century of electronic technology, we have extended our central nervous system itself in a global embrace, abolishing both space and time as far as our planet is concerned. ... as electronically contracted, the global is no more than a village." (McLuhan, 1964, p. 19-20).

"Not simultaneous; not concurrent in time" according to Webster's New International Dictionary, Second Edition.

These features sometimes include notification of participants when new participants actively join a conference, the ability to find out who is active in the conference, and the ability to engage in a one to one dialogue with other participants. This kind of feature generally depends, however, on a user interface that is tightly integrated with the computer conferencing software. The tight binding most often restricts participation to the users of a single centralized computer system. None of these features are available on the computer conferencing facility that is the subject of this study. These and similar features are built into Turoff's EIES computer conferencing software as documented in Hiltz and Turoff (1978).

Other computer-mediated communications vehicles can and do require simultaneous participation. An overview of computer media, some of which require synchronized participation, can be found the appendix. The requirement of synchronized participation is essentially antithetical to computer conferencing, however, as it can only reduce breadth of participation, geographical diversity, and other essential characteristics of the medium.

Reardon and Rogers do not specifically mention computer conferencing, but cite similar media including teleconferencing networks, electronic messaging systems in organizations, and computer bulletin boards. One notes, moreover, that Rogers (1988) frequently uses the terms "electronic bulletin board" and "computer conference" interchangeably.

In current implementations, in which the only practical means of input remains a keyboard and the only practical delivery vehicle remains a text display. Future versions of computer conferencing can be expected to broaden the range of input and delivery vehicles to include voice, graphics and video. More will be said about these possibilities in the conclusion of this study.

Hiltz and Turoff (1978, p. 123) observe that a "critical mass" of at least 8-12 active participants from at least three geographical areas is required to assure the success of a computer conference. "Below this critical mass, there are not likely to be enough new messages or conference comments entered. Above the minimum size and dispersion, enough activity and controversy is generated to motivate members to sign on frequently and to participate actively in the exchanges." There is no formal confirmation of this critical mass hypothesis, and it isn't clear that geographical distribution of participants is a necessary feature of critical mass, but the need for a number of committed and vocal participants is generally consistent with the author's experience.

A "userid" is the "name" a user is known by on their local computer system. It is generally two to eight characters and may be the users real name, a personally selected self description, or other alphanumeric code. A "node" is the name of a computer. The user id and node, taken together, provide a complete electronic address to which mail can be sent with reasonable assurance that it will reach a particular person.


An individual who starts and/or subsequently takes on responsibility for a forum's care and maintenance is known, on IBMPC, as the forum's "owner".

A forum on IBMPC is an individual conference, one of many distinct conferences maintained on the conferencing facility. Forums generally have a specific purpose -- a focus on a specific program or other general topic.

PCTOOLS is a sibling facility of IBMPC which distributes employee authored IBM Internal Use Only software within IBM.

Individuals who receive contributions to various forums on IBMPC via electronic mail.

Individuals who request changes to the various forums on IBMPC via electronic mail.

Individuals who read forums on IBMPC by directly editing copies of the forums maintained on their local computer systems.

Electronic mail that notifies participants of changes to individual forums on IBMPC.

Medium as Process: An Overview

This point concerning the equivalence of paper and audio tape may be regarded as controversial. The reader is therefore asked to suspend judgement until the typology chapter, where the matter will be discussed in more detail.

Also frequently referred to as an actor, transactor, speaker, presenter, or communicator.

Also frequently referred to as an audience, transactor, or communicator.

Sometimes in conjunction with such concepts as encoder, decoder, and message.

A "proto-medium" is a collection of mediators which, if used together, could be used to enable communication. There are an infinity of such combinations, only a few of which ever attract widespread support. The range of such possibilities is left to the imagination of the reader.

Prior Studies of Computer Conferencing

Johansen, Vallee, and Spengler note that this result runs contrary to other evidence, and may be open to question.

Mediators and Media

Coincidence is used here not as an indication of accident but of intersection. Working from an evolutionary perspective, this coincidence is a matter of natural selection. The optimization of message capture interfaces to message production facilities provides an inherent natural advantage that ultimately selects matching interfaces for continuation. One wonders, of course, about intersections that have not been made. The brain puts out low level energy patterns, but the means to interpret these patterns does not appear to be an element of human interface (or is it; although hardly demonstrated, reports of extra sensory perception remain persistent.

One wonders, moreover, of the impact the lack of intersection of interfaces may have on alien encounters if and when they should happen. Well matched interfaces between humans and aliens would be a most surprising outcome. One notes, relative to this, that poorly matched interfaces may be the biggest obstacle to meaningful communication between humans and dolphins.

Characteristics of Media

Potential Recipients are operationalized as follows:

  • low -- potential recipients measured in ones to tens.
  • medium -- potential recipients measured in hundreds to tens of thousands.
  • high -- potential recipients measured in units greater or equal to hundreds of thousands.

    Potential Feedback is operationalized as follows:

  • low -- feedback unlikely for any given recipient
  • medium -- feedback slow and/or indirect
  • high -- feedback is rapid and direct

    Effects of Media

    This is obvious a very crude statement of bandwidth. The written (paper and pencil) channel is clearly different than a spoken (voice and audio tape) channel. Still, the relative bandwidth of these channels is hard to gauge. The written channel is limited to words without inflection, but can be supplemented with pictures, underlining, pen pressure, punctuation and other analogic devices. The spoken channel can transmit inflection, including emotion and emphasis, but is limited to just words and inflection. Punctuation must be imagined from inflection. Pictures aren't possible. The value of such a measure, however crude, is in its ready observability. The number of channels inherent to a medium is about as observable as the number of feet on an animal. A similar measure of bandwidth will be used in constructing this studies typology of media.

    A Typology of Media

    This stretched definition of publishing is backed by some legal standing, as books, movies, paintings, and even computer software fall under the purview of the same copyright laws.

    My thanks to Dave Chess for this insight.

    Completing the Circle: Media Practices

    Practice is only one of several words that provide a reasonable description of this common ground. They might, insofar as they exist as a means of exerting "control" over a medium, be referred to as "controls". They might, insofar as they are mutually agreed on by participants in a medium, be referred to as "agreements", "understandings", "concurrences", "consensuses", "accords", "treaties", "protocols", or "contracts". They might, insofar as they constrain behavior, be referred to as "accommodations", "adaptations", "stipulations", or "constraints". Hence the term "practice" is not selected because it describes either rule or generic process perfectly. It simply describes the common heritage of these media-defining social innovations at least as well as any other.

    Innovation, as used here, describes not only the moment of invention, but the fact of invention. Media practices do not simply exist, even when they have been conventional for hundreds or thousands of years. They exist as human innovations and survive because those innovations remain functional for the people who use them. Hence while one might, with almost equal correctness, define media practices as "social conventions" directed at the content of media, we regard "social innovation" as a better description, especially in a study where we will directly observe the evolution of some such innovations for computer conferencing.

    Trivial is not used to indicate that these measures are particularly easy to compute. Frequency of word usage, for instance, entails a highly repetitive collating task that is probably only practical when accomplished via computer program. We use trivial only to indicate that the measure is in no way complex, a fact that is readily verified by the ease with which a computer can be trained to compute the measure.

    The author has discovered, since first writing this paragraph, that the news summary was not discontinued, but moved to page two of section A. Failure to discover this relocated feature in reading hundreds of New York Times issues over the last several years is probably no great complement to the authors observational skills, but it does provide an interesting commentary on the power of media practice, even in the process of viewing a mass medium. The author never thought to look for the feature elsewhere in the paper, but simply assumed that, no longer located where it belonged, the news summary had been discontinued. Some of this was, perhaps, a function of time. The author generally reads the news paper quickly, skipping over large amounts of the paper rather than browsing everything. It is suspected, however, that many people read the newspaper, and approach other media, in much the same way, developing media practices that allow them to minimize the time and other investments they make in media.

    Medium as Process: An Overview

    This and other key dates in the evolution of the IBMPC computer conferencing facility, many of which will be mentioned in this chapter, are taken from the IBMPC LANDMARKS file on the IBMPC Computer Conferencing Facility.

    This and other statistics are calculated from archival records of IBMPC activity.

    Much of this dissertation is written on a PC that was delivered in this first week.

    The author was among the organizers of this group, and served as its first president. Dates are taken from meeting notes and The CPC Newsletter (the group's newsletter).

    As written by Mike Cowlishaw, who is also responsible for the REXX language and the REQUEST EXEC.

    From a practical perspective, the user ID must be owned, managed, and administered by somebody, but the operation of TOOLS makes it accessible for use by many people.

    From the machines perspective, this user ID is simply the assigned name of a virtual machine whose access to processor, disk and other machine resources allows the application to run. Perhaps the most important function of the user ID is in providing a unique destination for any electronic mail that is sent to it.

    A step interaction variable constructed by multiplying a dummy variable corresponding with the weeks in the period against an incrementing count of the weeks in that period.

    This step interaction variable better represents the acceleration in appends when it is transformed to an exponential. The current model accounts for 92.48% of the variance (R-squared) in appends to IBMPC (F=1450, df=3/354, p<.00001). Transformation of this variable to an exponential relationship allows the model to account for as much as 92.60% of variance (F=1475, df=3/354, p<.00001).

    The nature the step interaction variables in this analysis is such that the parameter estimate corresponds to a rate of acceleration in appends to IBMPC during the period.

    A dummy variable corresponding with the weeks in the period.

    A step interaction variable constructed by multiplying a dummy variable corresponding with the weeks in the period against a incrementing count of the weeks in that period.

    This lack of significance for the intercept is expected, as the absolute baseline for future growth on IBMPC is zero appends per week before December, 1981.

    Growth in the use of IBMPC during this period can be represented adequately with a single linear component. It is represented somewhat better, however, at least for growth in appends and files on IBMPC, by a mildly exponential component. The superiority of this exponential component in representing the growth of IBMPC is an indication that the acceleration in the use of IBMPC is itself accelerating during this period. Once that is understood, however, the linear component is easier to understand (as a real world acceleration).

    "Per week per week" is a correct statement of a rate of acceleration. If, for instance, IBMPC attracted 300 appends in one week and 307 appends two weeks later, the append rate or velocity for the first week would be 300 appends per week and the acceleration over the two week period would be 3.5 appends per week per week.

    The table on append growth includes footnotes that apply to this model as well.

    The nature the step interaction variables in this analysis is such that the parameter estimate corresponds to a rate of acceleration in contributors to IBMPC during the period.

    The second week of 1985 is fairly representative of the period. It is selected in preference to the first week of 1985 or the last two weeks of 1984 because these three weeks experience persistent declines in the number of appends. This decline is presumably related to the effect of holidays and vacations on employee use of IBMPC.

    The table on append growth includes footnotes that apply to this model as well.

    The nature the step interaction variables in this analysis is such that the parameter estimate corresponds to a rate of acceleration in changed files on IBMPC during the period.

    Originally named VMPC, but only briefly (a matter of days).

    A Simple Query

    An assessment of costs will be made at the end of the chapter.

    What Kind of Event was this Simple Query

    A reference to R.D. Laings "Knots", a book of poems about the intricasies of human relationships whose style this offering imitates.

    CompuServe is a publicly available consumer timesharing network that is unrelated to IBM.

    A case might also be made for cave graffitti and other early media.

    Genres of Computer Conferencing

    Belongs to the first cluster of means, which includes six value questions with means ranging from 1.68 to 1.86. None of the questions in this first cluster of significantly different than any other according to t-test comparisons. All are significantly different than the other seven means in the analysis.

    Belong to both the first and second clusters of means. None of the four questions that load in both clusters are significantly different than the eight top means in the analysis, ranging from 1.68 to 1.96.

    Belongs to the second cluster of means, which includes six value questions with means ranging from 1.80 to 1.96.

    Belong to both the second and third clusters of means.

    Belongs to the third cluster of means, which includes three value questions with means ranging from 1.96 to 2.16.

    Belongs to the fourth cluster of means, which includes two value questions with means ranging from 2.52 to 2.61.

    Belong to both the fourth and fifth clusters of means.

    Belongs to the fifth cluster of means, which includes two value questions with means ranging from 2.61 to 2.75.

    Formal and Informal Rules

    In describing IBM, Harvard Business School (1988) writes that "IBM prided itself on its corporate citizenship and its strong cultural norms. Permeating the culture were the three "basic beliefs":

    1. Respect for the individual -- caring about the dignity and rights of each person in the organization;
    2. Customer Service -- giving the best service of any company in the world; and
    3. Excellence -- believing that all jobs and projects should be performed in a superior way.

    A series of rules changes had been made to IBMPC MEMO between 1981 and 1984. This revision represented a major rewrite, however, and resulted in a distinct rules document, IBMPC RULES.

    None of which have been much of a problem on IBMPC anyway. This reflects, to a very large extent, the culture of IBM, in which such behavior has always been regarded as unacceptable. On those rare occasions when the IBMPC administrators have encountered such language, social pressure on the appender from other IBMPC participants usually results in the modification of the append before the IBMPC reviewer even sees it. Consistent enforcement of the expectation of expression with business courtesy and good taste has, however, undoubtedly reduced the use of unacceptable language on the facility.

    One notes that the IBM experience in this area is not unique. One can point to any number of academic computer conferencing facilities (COMSERVE is certainly an example) where inappropriate language is very much the exception. Neither, however, is it general. One can also point to any number of non-IBM computer conferencing facilities in which rude language is the norm.

    The Administration of IBMPC

    Chess' interest in the IBM PC also reflected his principle job responsibility as the PC consultant for the IBM Thomas J. Watson Research Center. Chess read IBMPC in part to increase his knowledge of PC's. Indeed, he was widely regarded as the most knowledgeable PC consultant in the company, and became, through IBMPC, a PC consultant to the entire company.

    One should not, in combining this estimate with the prior paragraphs estimate that the reviewer may read as little as ten percent of all appends, conclude that 1 append in 65 is a problem. The IBMPC reviewing software and informal reviewing process, which has already been described in this chapter, work to insure that the reviewer who only reads 10% of all appends sees those appends which are most likely to be problematic.

    This analysis of the complete contents of IBMPC on December 2, 1989 showed a total of 149,327 appends to IBMPC. This represents about nine working months of submissions at current submission rates. It is not a true nine months, however, as some forums are "pruned" (edited to remove older appends) more frequently than that. A complete review of modify actions could be performed by accessing the IBMPC archives, which contain all pruned appends.

    Of the appends analyzed, individual appenders had modified their own appends 5,691 times (changed about one in twenty-two appends). Forum owners modified another 458 appends. The IBMPC administrators modified or rejected only 232 appends. This rate, about 1 reviewer modified append in 650, is probably inflated somewhat by appends that have been modified at the request of an appender who either cannot, or cannot figure out how to, make a desired modification.

    These raw numbers clearly do not tell the whole story. They do not, in particular, count archived appends in the many forum fragments on the IBMPC archive. Reviewer actions may actually be overcounted, as there is no discrimination of reviewer owned forums (hence some reviewer actions may really be owner actions), and some reviewer actions are taken at the request of appenders.

    The statistics generally ratify a pre-survey estimate that reviewer actions occurred at a rate of no more than 1 in 500 appends. Owner actions were, as has been proposed by other IBMPC participants in response to the 1 in 500 estimate, somewhat more frequent, but still only affected 1 in 326 appends. Forum owners and conference reviewers, when counted together, changed 1 in 216 appends.

    Benefits of Computer Conferencing

    Paraphrased from Bezilla and Kleiner (1980).

    The paper referred to is Discovering the PC User Interface: Design lessons of FILEMAN, STP, and the Yorktown PC User's Workbench, a draft paper written in 1986 by one of us (Foulger).

    Four Hours in Community

    In append 113 to SHUTTLE FORUM, made at 3:56 PM EST on January 31 we learn that the author, located is Austin, Texas, "was at home Tuesday and saw it live on CNN."

    Eyewitness reports are given in three appends, all made by IBM employees located in Florida on January 28. Append 8 (12:53 PM) reports watching "the shuttle explode from my office window" in Boca Raton. Append 30 (1:41 PM) reports seeing "the explosion via binoculars from my front door" near Tampa. The writer of append 78 (3:44 PM; Boca Raton) reports that "I watched the shuttle blow up with my eyes."

    In append 116, 2:26 AM on February 1, we learn that a Kingston, NY employee was "resting ... when my clock-radio came on somewhere in the background of my sleep. When it came on, there were a few words spoken and then I heard those horrible words: "...the shuttle disaster..." and I was ***INSTANTLY*** WIDE awake."

    A Cary, N.C. appender writes (append 114) that "I first heard of the disaster from a friend and co-worker who had just received a call from his sister with the news."

    In append 102, Cary, NC employee tells us that "a friend came into my office and asked if I'd heard that the shuttle blew up."

    A Kingston, N.Y. employee (append 26) reveals that "It was announced a few minutes ago here in Kingston." Append 43 reveals that "shortly before 11:00am Tucson time, the announcement about the shuttle came over the P.A. system."

    The average length of appends grows by about 10% over the first four hours of the forum. Appends grow by about 20% over the course of the last six days. Between the early and the late forums, however, the length of appends jumps by about 40%.

    Append 70 of SHUTTLE FORUM, posted at 3:25 PM on January 28th.

    Append 56 to SHUTTLE FORUM, posted at 2:50 PM on January 28th.

    Append 92 on January 30.

    Expressiveness in a Digital Medium

    697 lines (28%) are associated with appends that contain news. 478 lines (19%) are associated with appends that contain analysis.

    The term "grief work" can be traced to Lindemann (1944). Gorer (1977) describes Lindemann's conception of grief work as "emancipation from the bondage of the deceased, readjustment to the environment in which the deceased is missing, and the formation of new relationships.

    From Gordon's (1975) description of the shiva.

    Gorer reviews the work of Freud (1917), Klein (1940), Lindemann (1944), Eliot (1955), Engel (1961), Marris (1956), and Bowlby (1964a, 1964b).

    One wonders if this decline in the importance of the community isn't a better explanation for the decline in funeral traditions than the high World War II casualties that Gorer (1977) advances as a central cause of the "social denial and individual repudiation of mourning" in Britain.

    The New Electronic Community

    From Robert Frost's "Mending Wall".

    Medium as Process on IBM's IBMPC Computer Conferencing Facility

    This practice is expertly satirized in the play and movie "The Music Man" when a huckster sells a town on buying musical instruments and starting a band by invoking a vision of otherwise unoccupied children hanging out at the pool hall.

    Overview of Fifty-Two Media

    In a computer conference, an append is the equivalent of a message. When a message is sent to a computer conferencing facility, it is "appended" to the end of an existing conference. In some computer conferencing systems, like the TOOLS computer conferencing software on which IBM's IBMPC computer conferencing facility is based, the append is quite literal. A conference is a document, and messages are appended to the end of that document. In other systems, like the COM and PORTACOM computer conferencing systems, the append is less direct, with the message maintained as a separate file, and a pointer to that file attached to the conference file.

    From December, 1984 and March, 1985.

    IBM's IBMPC computer conferencing facility is arguably one of the largest computer conferencing facilities currently in use. At the end of March, 1985, it contained over 100 million characters (15 million words) of information, distributed between over 1000 different forums, or individual computer conferences. During a four month observation beginning in December, 1984, nearly 3000 people appended, on average, roughly 120,000 characters (17,000 words) a day. That's the equivalent of a 350 page novel about Personal Computing every week. At the end of 1985, that rate had increased at least three fold, to well over 1000 pages a week.

    The fastest growing forum on IBMPC, ISTHERE FORUM, was growing at about this rate at the end of March, 1985.