Guidelines for Network Communication

Janusz Buda


The present article is the fourth in a series of studies of network communication, specifically the communication that takes place in open-forum discussion groups. Previous articles have focused on defining some of the characteristic features of this new medium of communication and identifying some of its limitations and constraints.

These limitations and constraints affect both the form and content of the messages carried by the medium, and also the form and content of the communication thus mediated.

The relationship between medium and messages, and medium and communication, is an exceedingly complex one. In the present study, an attempt is made to clarify these relationships by offering a number of guidelines for more effective network communication.

Principles of Computer Communication

The modern computer is an efficient device for transmitting, processing and storing information. It can handle words and images as easily as numbers, and it is rapidly replacing older information-dissemination technologies.

In the early days of computer development, information was carried to and from computers via discrete media such as punched cards, magnetic tapes and magnetic disks. Although magnetic (and, to an increasing extent, optical) disks are still in widespread use, most data now flows directly from computer to computer via electronic networks.

Direct Link

The simplest form of computer network is created when two or more computers are linked by cable.

A user sitting in front of one computer can compose a message on his or her screen, and then send it to one or more users on the same network. This kind of basic network can be found in schools, universities and offices throughout the world, and is a far more efficient (and more environmentally friendly) method of communication than writing or typing messages on paper and then carrying them from room to room, or from person to person.

This kind of computer network does, however, have some disadvantages, the most serious being the constraint on physical distance. Connecting computers located in the same room, or in the same building, is relatively easy and inexpensive. Laying cables to connect computers in different buildings, or in different parts of a city, is a different matter altogether.

Phone Link

This limitation of distance can be overcome by making use of existing world-wide telephone systems. In the same way that telephones are used to link people throughout the world, irrespective of distance, they can be used to link computers separated by thousands of miles. In addition to a computer and a telephone, all that is required is a small electronic device called a modem. This device is necessary because computers and telephones transmit data in different forms. Computers transmit information in digital form, as a stream of numbers; telephones transmit information in analog form, as a stream of modulated sound. A modem converts digital signals into sound, and then feeds this sound into the telephone. At the other end of the link, another modem picks up the sound and converts it into digital signals before passing it on to the receiving computer.

Although this sounds complicated, for the average user it is as transparent as sending a fax message from one machine to another.

This comparison with fax transmission does, however, underscore one of the significant limitations of long-distance computer communication. Whereas fax machines are usually left switched on twenty-four hours a day, ready to receive incoming messages, most personal computers are switched off when not in use, and will not respond to incoming calls.

This is not usually a problem when information is being transmitted between computers in the same time zone — most people can be expected to be working similar hours, and the likelihood of establishing a computer connection is very high. When the computers are in different time zones, on the other hand, the likelihood is that the recipient computer is switched off, and the user asleep.

It is for this reason that modern computer communication makes use of an alternative paradigm of information transmission, one that overcomes constraints of both time and distance.

Message Centre

Before examining this alternative paradigm, it may be useful to digress for a moment and consider the problem from another perspective.

Let us imagine that a large university of, say, ten thousand students needs to suspend lessons at short notice. What would be the best, the most efficient way to communicate this information to the students?

Several options come to mind. One would be use of the telephone system. The university office could call the students by phone and inform them of the suspension. This would be a very quick and direct method, but it would also be impractical. Even if we imagine every single member of the office dialing all day long, from nine to five, it would still be physically impossible to make ten thousand calls in the time available. Even if it were possible, there would be no guarantee that all the students would be at home when the calls were made.

Another option would be use of the mail system. The university office could print an announcement and mail it to the students at their home addresses. This would ensure that almost all the students would read the announcement sooner or later, but it could hardly be called efficient. It would take time and labour to have the announcement printed, and even more time and labour to address and fill the envelopes—not to mention the enormous postal charges involved.

Both of these options make use of a centrifugal model of information transmission. The impulse for information transferal comes from the centre, and the information is broadcast outwards, from the source to the recipient.

An alternative option is use of a centripetal model of information retrieval, in which the impulse for information transferal comes from the recipient.

Throughout history this model of information transmission has been found to be the most effective and secure method of overcoming limitations of time and distance.

In the form of the message centre, it has been used in times of war and disaster, and by populations on the move. In the form of the message drop, it has been used by spies and agents working under cover.

In the more salubrious environment of the university campus, the same model of information transmission is to be found in the university announcement board.

Students are told where the board is located and instructed to check it as often as possible. The university office has only to type out a single sheet of paper and post it on this notice board, where it will be read by students the next time they check the board.

Although both telephone and mail have distinct advantages over the announcement board, considerations of time and economy make it likely that the latter would be used in the hypothetical situation described above.

It is not surprising, therefore, that computer networks have adopted the same principle of the message centre to facilitate rapid communication between users separated by both time and distance.

In a university, the message centre takes the form of a physical board made of wood or plastic. In a computer network, a computer replaces the announcement board. Instead of paper messages pinned to a wooden board with drawing pins, electronic messages are stored on hard disks inside the host computer, and network users connect (or ‘log in’) to this computer to access that information. Because of the similarity between these two kinds of message centres, small- and medium-scale computer networks are often referred to as Bulletin Board Systems.

Single-node Networks

At one extreme, a small Bulletin Board System (BBS) can consist of little more than a personal computer located in someone’s home, accessed via a single telephone line and able to support only one call at a time. At the other extreme, a large BBS may make use of powerful workstation computers, and a complex electronic switching system can allow dozens of users to access it at the same time. Whatever the size, however, all such systems are basically autonomous, and all incoming and outgoing message traffic is processed at one site. Because they make use of only one site, or ‘node’, such systems can be referred to as single-node networks, a description that would also encompass many giant commercial information services run on massive mainframe computers and able to support simultaneous calls from thousands of users.

Multi-node Networks

In the same way that messages can be transmitted from bulletin board users to the host computer, they can also be transmitted from one host to another. One of the limitations of a single-node bulletin board or information service is the fact that it is a closed system, and its store of messages is available only to users of that system. A user of such a single-node bulletin board can send private or public messages to other users of the same system, but cannot send messages to the users of other systems. Linking single-node hosts computers and exchanging messages overcomes this limitation, and joins users of single-node bulletin boards into a far greater multi-node system.

If we return for a moment to the example of the university announcement board, we could imagine a large institution with several widely-flung campuses arranging for all messages posted on one board to be copied and posted on all the other boards on other campuses, thus ensuring that the messages were read by all the students of that university.

It might be useful to refer to this kind of arrangement as a single-level multi-node system, one in which all the nodes are connected directly to one another, and perform the dual functions of providing access to users and exchanging messages.

Such a simple multi-node system is, however, commonly found only in grassroots networks linking relatively small hobby boards. Such grassroots networks work efficiently only in small configurations: the more individual nodes join the network, the more time it takes to exchange (or ‘gate’) messages between them. In an extreme situation, a small BBS belonging to such a network would be spending more time gating the messages of other boards than processing the messages of its own users. It is for this reason that large multi-node networks make use of a complex hierarchy of nodes linked by high-capacity, high-speed lines to exchange messages. Such a hierarchy can be compared to the hierarchy found in commercial production and distribution systems.

In practice, users do not access the upper-echelon nodes of such multi-level, multi-node networks in the same way that consumers do not, as a rule, go all the way to the factory to purchase a can of tinned fruit, or all the way to a farm to purchase a dozen eggs.

Characteristics of Computer-mediated Communication

Forms of Computer Communication

Computer communication takes many forms. The expression ‘computer communication’ is itself misleading, for although computers can and do communicate with each other, when we speak of computer communication we are usually referring to communication between the users of those computers. For this reason, some writers prefer the expression ‘computer-mediated communication’.

Computer-mediated communication includes the exchange of private correspondence (E-mail), the exchange of files and other digital data, and the exchange of public messages. The present study focuses on the latter, and subsequent discussion will concentrate upon the exchange of public messages that takes place on the open forums of the world-wide Usenet network.

Before doing so, however, a word needs to be said about terminology. Many network-related terms are loosely defined and often used indiscriminately. The open-forum newsgroups known as Usenet are often viewed as part of the Internet, when in fact the Internet system of networks is only one (albeit the largest) of the many networks, or network systems. participating in Usenet. Likewise, ‘the Net’ can sometimes be used to refer to Usenet, and sometimes to the Internet. It can also be used to refer to an even larger system encompassing all interconnected nodes and networks, whatever the nature of the connection.

The terminology of Usenet participation is similarly imprecise. Relying heavily on the metaphor of news broadcasting, the preferred usage refers to the discussion forums as ‘newsgroups’, to contributing messages as ‘articles’, and to the act of contributing as ‘posting’. In actual practice, however, ‘message’, ‘article’, and ‘post’ are used interchangeably.

Information Overload

Anyone accessing Usenet newsgroups for the first time may be forgiven for assuming that he or she had stumbled into a madhouse. The experience can compared to that of blundering into a sports stadium in which several thousand people are all talking to each other at the same time.

Computer networks present the user with far more information than he or she can comfortably absorb or digest. This information overload is caused by a number of factors, the most significant being volume, flow, and organization.

Two years ago there were approximately three thousand newsgroups, or discussion forums, in existence. It is now not uncommon for Internet providers to carry over twelve thousand.

In the same two years, the number of network users has risen from approximately three million to over fifteen.

The volume of network traffic has increased accordingly. Two years ago, approximately twenty-five million characters (equivalent to six or seven volumes of the Encyclopedia Britannica) were being fed into the system every day. The figure is now closer to fifty million.

The three factors of volume, flow and organization do not operate in isolation. Flow amplifies the effect of volume, and lack of organization further aggravates the effect of overload.

To understand this third factor of organization, some familiarity with the structure of newsgroups, and the nature of newsgroup messages, is required.

Message Organization

In a small BBS with only a few dozen members, there is little need for organization of messages. Only a few messages are posted each day, and it is relatively easy for users to keep track of the various discussions being conducted.

The volume of messages on larger BBSs makes it more difficult to follow the ‘threads’ of discussions, and messages are usually organized into topic-specific categories. Often it is necessary to further subdivide these categories into sections, resulting in a hierarchical organization similar to that used in libraries.

For example, a board may have a forum[1] devoted to a discussion of sport. This forum may be subdivided into categories dealing with various sports such as basketball or soccer, and these categories may themselves be subdivided into, say, topics dealing with individual teams or individual players.

For all this neat arrangement, discussions still have a tendency to fragment and drift. A lively discussion of next season’s prospects for a specific American basketball team may generate messages about the hiring of new players. This discussion may, in turn, lead to a discussion of the merits of the present team manager. Some members of the discussion group may continue discussing next season’s prospects, others may continue discussing new players. In this way several related or unrelated threads may run concurrently, and it is not unusual for users to participate in multiple threads.

On multiple-node networks such as Usenet, the basic unit of discussion is the newsgroup.

The name of the newsgroup defines the subject of discussion, and to some extent this naming follows a hierarchical system.

Staying with the example of basketball, as of November 1995 there were 37 basketball-related newsgroups on Usenet:

The dots or periods between words identify different levels of a specific hierarchy, descending from left to right.

Most of the groups listed above belong to the ‘alt’ hierarchy, so-called because it provides a more easy-going, alternative forum for discussion of a wide range of topics. Seven groups belong to the mainstream ‘rec’ (recreation) hierarchy, and the remaining two belong to local hierarchies.

Within a specific newsgroup, however, there is no organization of messages, and the only way to identify related posts is through internal evidence—information contained within the messages themselves.

Message Format

It might be useful at this point to take a look at a typical Usenet article, and identify the main features.

From: Bill Scorzelli <>
Subject: Re: The dog ate my Trancenet
Organization: none
Lines: 22
Message-ID: <47dju0$>
References: <jmknapp-1310951219180001@>
<> <46atsj$797@jefferson.fairfield.c>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 1.12 (X11; I; HP-UX A.09.05 9000/712)
Xref: cfi.waseda alt.meditation.transcendental:14628 alt.meditation:11391 alt.journalism:26443 sci.skeptic7 (Korrinn Fu) wrote:

:Have you seen the book Combatting Cult Mind Control by Steve Hassan?
:Thought Reform and the Psychology of Totalism by Robert Lifton?  And
:the book (?) by Margaret Singer and Janja Lalic.  Keep an open mind.
:Take a look.  Browse through them in the book store.  You don't even
:have to buy it if you don't want to.  Then people will respect your
:opinion as an educated one.

You are not suggesting that these books are above objective studies are you?
If you are... 
       "people won't respect your opinion as an educated one" 
no matter what books you read.  :^)

"Most people involved in cults replace this need with another cult" -unknown

The first thing to note is that the message itself begins with ‘….’ The eighteen lines of information preceding the body of the message were generated automatically by both the author’s message-posting software and the news server that delivered the message. This information, usually referred to as the ‘header’, is crucial in identifying messages, and helping to connect related postings.

The article itself begins with a quotation from an earlier article, and an attribution to identify the author of that article. The author is referred to by his Net address, and by the name he has registered. This name, by the way, does not have to be a real name. Many users register nicknames or handles for reasons of privacy, or as part of a network ‘persona’ they are cultivating.

For a number of reasons, newsgroup quotations adopt a format far different from that found in reports and academic papers. Most quotations are identified by a series of symbols at the beginning of each line, and quotations-within-quotations by a complicated system of nesting. A more detailed description of quotation practice can be found in ‘Quotations in Network Messages’ (Buda 1995).

The article itself consists of a short comment about the quotation, and is followed by a ‘signature’. This can consist of a name, a nickname, or indeed by several lines of information, such as contact addresses or telephone numbers. It can also contain a short joke or quotation that the author wishes to share with his or her readers.

Message Distribution

Depending on the popularity of the newsgroup, anything from two or three to three or four hundred such messages might be posted in any twenty-four hour period. These messages will, of course, be posted at different times and in different places, and will take different routes through the network system. These routes may differ from day to day, or from hour to hour, depending on the availability and capacity of connections between news servers.

As a result, articles can arrive at particular sites in no apparent order, and it is common for the response to a query to arrive before the query itself.

Indeed, it is not unknown for messages to become stalled or lost in the system, and it is sometimes necessary to reconstruct such lost messages from the quotations contained in follow-up responses.

This brief and rather superficial description of Usenet newsgroups and messages may suggest some of the complexity of the system. Without some kind of newsreader (software designed to access the articles stored on the news server computer, and present them to the reader in a manageable form), it would be almost impossible to handle the thousands of messages arriving each day. Even with the help of such a newsreader, posting a message to a newsgroup (as opposed to reading incoming traffic) demands a significant degree of familiarity with several computing and word-processing skills, as well as an understanding of the overall system itself.

The size, complexity, and flexibility of this system of message exchange enables millions of users to communicate with one another, and for individual users to have access to audiences of hundreds of thousands. Unlike other mass-media such as radio, television and the printed word, multi-node networks allow almost anyone, at almost any time, to communicate their ideas to a global community of network users.

At the same time, the sheer complexity of the system imposes a number of constraints and limitations on the act of communication itself. Compared with other forms of remote communication, newsgroup discussions would seem to exhibit a greater incidence of misunderstanding, argument, and acrimony.

Such negative elements hinder or stifle the free flow of ideas and information, and network users and administrators have devoted much time to developing counter-measures.

There exist, for example, a number of widely-accepted guidelines for participation in newsgroups, and these guidelines are frequently publicized in both the newsgroups themselves, and in introductory publications aimed a familiarizing new users with the network medium.

One such guideline recommends that new participants monitor the messages in a particular newsgroup for at least two weeks before posting an article. This is partly to enable new participants to familiarize themselves with both the current threads in the group and the current participants in those threads. It is also to enable newcomers to acquaint themselves with any unwritten ground rules that might exist in that group.

Another guideline suggests that new users first read the relevant Frequently-Asked Question (FAQ) sheets, or at least ask experienced users to point them to the location of such sheets.

As the name suggests, such FAQ sheets contain a list of questions frequently asked by newcomers to the group, and a corresponding list of answers.

They provide, in fact, the minimum background information necessary for a meaningful contribution to the discussion within a particular newsgroup, and forestall the need to answer the same questions over and over again.

Recommendations such as these are, however, little more than general guidelines concerning newsgroup participation per se. The following section offers a number of more specific guidelines for article composition and article posting, and analyses the problems and difficulties the guidelines attempt to address.

Message Guidelines


This first and perhaps most important recommendation has little to do with aesthetic considerations. The newsgroup system is complex enough without the added complication of long and involved messages. The volume and flow of network messages mentioned earlier impose a premium on both space and time: the space occupied by individual messages, and the time needed to read them. Long, intricate messages are at best an indulgence, and at worst a hindrance to the smooth flow of newsgroup discussion.

A simple message exhibits three characteristic features: short paragraphs, short quotations, and simple formatting.

Short Paragraphs

The preferred formatting of network messages is non-indented paragraphs separated by blank lines. This format provides the highest legibility for mono-spaced non-justified text displayed on video monitors. The eye of the reader has far more trouble following the lines of such text than it has following the lines of traditional printed matter, in which the counterstrokes of serif typefaces, the rhythm of ascenders and descenders, the finely-tuned kerning of letter pairs, and the leading between successive lines, all help the eye to track across the page from left to right, and from there to the beginning of the next line.

On a video monitor, the blank lines between paragraphs serve as visual reference points for the eye of the reader, and long paragraphs tend to nullify this effect.

Different computer terminals will display different typefaces and different line lengths, and the absolute size of each character will also depend on the resolution of the monitor being used. It is difficult, therefore, to give a hard-and-fast recommendation for a maximum paragraph length. Anything over ten lines, however, becomes increasingly difficult to read smoothly, and anything over fifteen lines tends to appear as a daunting block of solid text.

In the same way that the eye of the reader tends to scan ahead, looking for the blank lines that indicate the ends of paragraphs, the eye also scans even further ahead, looking for one of several markers (signatures, etc.,) that indicate the end of a message.

Without such indicators, reading a message for the first time becomes a voyage of exploration, a venture into uncharted territory. The reader must progress slowly, building up his or her own internal ‘map’ of the message and of the information it contains. A clearly demarcated message, following all the conventions of newsgroup posting, is familiar territory for the reader, and the message can be read and understood at high speed.

A useful parallel can be drawn with the act of driving along an unfamiliar road. If the road is unmarked and unposted, the driver must progress slowly, carefully ‘reading’ the road ahead and making decisions about what is or is not significant: possible danger points, for example, or areas likely to pose problems.

If the road is clearly marked with traffic lanes, on the other hand, and if all possible dangers and obstacles are clearly signposted, the driver can proceed far more quickly and confidently.

Short messages

If a newsgroup message contains paragraphs of ten lines or less, at least two end-of-paragraph blank lines should always be visible on-screen.

An earlier study (Buda 1995) found that the average length of messages in a corpus of 1,000 Usenet articles was a little over 350 words. If Usenet formatting conventions are followed, this length would correspond to approximately one screenful of text.

In other words, the markers indicating the ends of most newsgroup articles should also be visible at the bottom of the first screen.

If an article is considerably longer than average, the reader is forced to scroll the screen upwards, and in long posts of several hundred lines, scrolling may have to be repeated a number of times.

Such repeated scrolling has a deleterious effect on speed and facility of reading. The eye must jump from the bottom of the screen to the top, and if the number of lines scrolled at one time is not exactly the same as the number of lines displayed in the software window, the eye must scan the beginnings and ends of several lines to pick up the point at which it stopped reading.

The same study suggested that the two factors of screen size and message length were not unconnected: that most users consciously or unconsciously kept the size of their messages to within one screenful in a recognition of the maximum legibility of this size.

Short quotations

The average length of article mentioned above included quotations from other articles. The same study found that a typical Usenet article of 350 words contained a quotation of approximately 70 words.

The logic of newsgroup quotations is a little different from that of quotations found in books, articles, and reports. Newsgroups quotations are, of course, used to present evidence and identify sources. Their primary purpose, however, is to establish continuity with preceding articles.

Without accurately attributed quotations, it would often be impossible to follow the threads of newsgroup discussions.

To give an extreme example, what would most newsgroup participants make of the following terse message?

Yes, it has.

Perhaps some participants might recall another message posted several days earlier, one that contained the query:

Is it true that Precision Engineering has gone bankrupt?

Most participants would, however, not associate the two articles, particularly if the message containing the original query had still not arrived at their site.

On the other hand, if the response followed newsgroup conventions, and contained an attributed quotation, all would be clear, even to a casual visitor to the group: <Tom Alton> wrote:
> Is it true that Precision Engineering has gone bankrupt?
Yes, it has.

If the attribution also contained the ID of the original message, any reader so inclined would be able to track down that message and read it for him- or herself.

Quotations become even more important in threads involving several participants. Without quotations, it would be almost impossible to determine who was responding to whom, or which point was being addressed.

Quotations are indispensable to follow-up articles. At the same time, they add to overall article length. Keeping an article short presupposes keeping the quotation (or quotations) correspondingly brief. If the average size of a newsgroup article is 350 words, and if the average size of a quotation is 70 words, it follows that approximately 80% of a message is deleted in the act of creating a quotation. This deletion is important for two reasons: it reduces the size of the quotation, and thus keeps the size of the follow-up message within reason; it also helps to focus on the point that is being addressed.

Short, relevant quotations are thus an essential feature of newsgroup messages, but the generation of such quotations is a skill exhibited by a relatively small proportion of newsgroup participants.

One reason is judgment: the ability to determine which part of a message is relevant and which is not. Another reason is the technical difficulty of generating quotations. This difficulty has been treated in depth elsewhere (Buda 1995), but suffice to say that selecting text from a message and editing for clarity and continuity require a fair amount of familiarity and facility with the limited word-processing functions of most news-posting systems.

Clumsy, illogical quotation can lead to misunderstanding, confusion, and dispute. A survey of newsgroup messages reveals that the two most common problems are misattribution and misquotation. The following two recommendations address these two problems.

Accurate attributions

In reading and responding to newsgroup articles, the overwhelming majority of participants will make use of newsreader software. In addition to displaying articles in a logical, orderly form, newsreaders also automate the task of composing follow-up messages.

The newsreader will address the follow-up to the relevant newsgroup or newsgroups, and will usually paste the text of the original message, formatted as a quotation, into the text editor being used to create the response. It will also generate an attribution by locating the name of the author in the header, and inserting it into a pre-registered template such as:

In article <Message-ID>, (User ID) (User Name) wrote:

This system works smoothly and efficiently as long as the quotation is a relatively simple one. In an ongoing newsgroup discussion, however, it is often insufficient to quote from the preceding article only . The preceding article may be a response to an earlier message, which in turn may be a response to an even older post. Without including selected quotations from all three preceding articles, it may be difficult to clarify the relationship between a specific follow-up article and the articles before it.

Each time a quotation is pasted into a message, it will have quotation indicators added to it, and will move one step to the right in a nest of compound quotations. A simplified example of such a compound quotation might look something like this:

In article <2464>, (Thomas Hicks) wrote:
 > In article <3921>, (“Sabertooth”) wrote:
 >> In article <1359>, (Harold Evans) wrote:
 >>> Could someone please recommend a good, reliable printer in the
 >>> $400-$450 price range? Thank you.
 >> Well, I bought an RZ650 last month for $420, and I must say I’m
 >> very pleased with it. The quality of the output is excellent, and it
 >> handles low-grade paper far better than the SpeedWriter 150c I
 >> had before.
 > It’s strange that you should say that. I bought a RZ650 in April, and
 > I’ve had nothing but trouble with it. The print quality is just fine, but
 > it chokes on large files containing complex graphics.

You too, eh? I thought I’d just hit on a bad unit, but now it sounds as if the
machine has some basic defects. I wonder if I’m covered by the warranty?

Such nested quotations are almost unreadable, and short of counting the number of quotation indicators at the beginning of each line, it is almost impossible to ascertain who said what to whom.

Knowing this, some users will attempt to simplify the nest of quotations by deleting the older attributions without, however, deleting the relevant quotations. Used judiciously, such deletion can be effective, but any mistake in the selection of lines for removal can result in misattribution.

A participant may assume that he or she is responding to a remark made by, say, ‘Thomas Hicks’, when if fact the response is to something posted earlier by a user with the nickname of ‘Sabertooth’. If the response is a negative one—strong disagreement or criticism, for example—it might bring an angry riposte from the offended party.

Mistaken quotations are another feature certain to stimulate angry reactions. As with attributions, misquotation is highly unlikely unless the author of the message has edited the quotation. Lines thought irrelevant by the author may appear far from irrelevant to the person who composed the original message from which the quotation is taken. A common component of network flames (described later) is the charge that someone has deliberately misrepresented another person’s position by deleting important lines of the messages.

Such complications can be avoided if quotations are kept short and to a minimum. An ideal message would consist of a single response to a single quotation, with indirect reference to preceding articles in either the body of the message, or in a short preamble to the quotation.

Such simplified quotation would also avoid another undesirable by-product of nested quotations, a problem addressed by the next recommendation.

Correct line length

Most newsreaders support line lengths of 80 characters. Any lines of text longer than this will either run off the right margin of the screen, or will be wrapped onto the next line. The carriage return included at the end of every line of newsgroup text will be preserved, however, resulting in an ugly sawtooth pattern:

Had she done so, Ms. Cleere would have quickly discovered
astrologers are well aware of the precessional movement of
equinoxes. She would also have discovered that a planet’s
significance is based not on the visible constellations (as she
purports), but instead solely on the planet’s cyclic movement
respect to the equinox.

It is rare (but not unknown) for network users to post original articles with line lengths of over 80 characters, but even text formatted to less than this length can run over the limit if a quotation indicator is inserted at the beginning of each line. Once this sawtooth pattern is caught and preserved in a quotation, it will be picked up and reproduced in all subsequent quotations, and if such subsequent quotation should push even more lines over the 80-character limit, a truly ugly and almost illegible textual morass will result.

It is highly advisable, therefore, to set text editors to line lengths of well under 80 characters in the assumption that the message might be included, as an indented quotation, in subsequent newsgroup articles.

Simple quote indicators

It is also advisable to choose the simplest (and shortest) quotation indicators possible, to prevent one’s newsreader from contributing to the problem of line overflow by inserting indicators that push quotations several spaces to the right.

By far the most common indicator is the ‘greater-than’ symbol, used in approximately 60% of all newsgroup quotations, and there is little reason, other than personal fancy, to choose other symbols. Especially to be avoided are indicators that include multiple symbols or characters, or unnecessary spaces.

There remain two other quotation-related recommendations: one concerning orphan quotations, and one concerning message signatures.

No orphans

Orphan quotations are created when a user creates a response to a newsgroup message but forgets to delete irrelevant parts of that message. This happens when the user inserts text into a previous message already formatted as a quotation, and the remaining part of the quotation is pushed off-screen. Off-screen, the remaining part of the quotation is invisible, and it is all too easy to post the response without being aware of the unattributed and irrelevant quotation at the end.

Such an orphan quotation, coming as it usually does after the signature indicating the end of the article, greatly confuses the reader and if the orphan quotation itself contains a signature (as is usually the case) can cause confusion over the identity of the author.

The only reliable way to eliminate orphan quotations is to review the entire text of a message before posting it.

Simple signatures

Strictly speaking, signatures at the ends of messages are not essential, as the header of a message will identify the author of the message. The use of signatures is, however, almost universal, adding, as it does, a personal touch to an impersonal medium. Most newsgroup participants keep their signature on file, and it is inserted automatically at the end of any message they compose. Complex signatures extending over several lines, and containing jokes, literary quotations and ornamentation, are to be discouraged.

The final three recommendations relate to the process of article posting in general, as opposed to the composition of the articles themselves.

Minimum crossposting

When a network user posts the first article in a thread, he or she must decide to which newsgroup or newsgroups it should be sent. Sometimes the decision is not an easy one. A user seeking information might feel that posting to several related groups will increase the chances of a response. If the groups are not chosen with care, however, such cross-posting might generate condemnation from group participants who feel that the subject matter of the post is not appropriate for their newsgroup.

Care must also be taken when posting follow-up articles. Newsreaders will usually preserve the distribution list found in the header of the article being responded to. If an article has, for instance, been distributed to six newsgroups, the automatic reply function of a newsreader program will send any follow-up message to the same six groups.

As a discussion thread develops and drifts, it could be that the original set of newsgroups is no longer appropriate. Before posting messages, newsgroup participants should scan the list of groups displayed in the header, and delete any that are no longer suitable. Some considerate users make a point of announcing such deletions before carrying them out, giving those participants who might still wish to follow the thread the option of doing so by subscribing to one or other of the groups to which it is still being distributed.

Prompt follow-ups

In the early days of computer networking, almost all newsgroup messages were read and composed on-line by users sitting in front of terminals with direct connections to the Net. Such institutional users did not have to worry about connection charges, and were able to make use of the relatively sophisticated newsreader programs lodged on their local computer systems.

The proliferation of commercial network providers offering Net access to non-institutional users has meant that ever more newsgroup participants are accessing from their homes via telephone links, and are having to pay both access and phone charges. For such users, reading and writing messages on-line can be an expensive luxury, and many prefer to do both off-line — after they have disconnected from their Net provider.

Access and phone charges are not the only consideration. Even if users enjoy cheap Net access and toll-free local calls, they may not have the time available to read several dozen messages, and compose the necessary replies, all in one sitting.

Still other users may prefer to take their time writing messages, and may be reluctant to post hastily composed articles.

The problem with off-line reading and writing is that it introduces a time-lag into the process of newsgroup discussion.

This lag can be one of a few minutes or hours; it can be one of days or even weeks.

Given the tremendous volume and flow of newsgroup messages, a week can be a very long time, and a user posting a tardy follow-up article might find that several hundred other follow-up messages had been posted in the meantime, that his or her point or comment had already been made by other users, and that the thread had moved on to a consideration of other matters.

No bunching

Off-line reading and writing also contributes to another undesirable phenomenon of newsgroup discussions: article bunching. Users who prefer to read messages at leisure, and compose replies in a similar manner, often post a week’s worth (or sometimes more) of messages at one time. This will result in a string of message by the same author addressing a large number of other messages, and a large variety of topics.

If the responses contain accurate attributions and carefully edited quotations, there should, in principle, be nothing unacceptable about such bunched follow-up articles. In practice, however, they are strongly disliked by most newsgroup participants. Few users enjoy reading more than two or three consecutive articles from a single author, and if the articles happen to address issues brought up (and resolved) much earlier in the discussion, the negative effect will be further compounded.

No flaming

One of the least commendable characteristics of network communication is the phenomenon of flaming. A flame can be loosely defined as a unwarranted personal attack on another newsgroup participant. Such attacks are extremely common, and the frequency with which they occur may be related to the sensitivity some users feel with regard to their network image or persona.

Whatever the cause, flames have a profoundly detrimental effect on newsgroup discussion, and in extreme cases have resulted in a major exodus of participants from specific groups.

Although some flames are started deliberately by irresponsible and aggressive users, most flames have their origin in spontaneous over-reaction to remarks that are taken as personal attacks. Almost any remark can be taken in such a manner, and there is no guarantee that even the most innocuous of comments will not be the start of a flame.

Some categories of remarks do, however, stimulate more than their share of flames, and such remarks are to be avoided wherever possible.

One such category is gratuitous criticism. Negative comments about the spelling, vocabulary, or grammar of messages contribute nothing to the discussion, and invariably cause great offense.

Another category of common flame-starters is personal comment. Any remark prefaced by ‘You are wrong…’ or ‘You are mistaken…’ is far more likely to spark off a flame than an impersonal comment which points out perceived inaccuracies or, better still, which simply states the author’s differing point of view.

Once a flame has started, however, the main consideration is damage control.

The quickest way to snuff out a flame is to refuse to respond to one. This is, unfortunately, easier said than done. It requires extreme self-control to let a public insult or criticism go unchallenged. In the hectic thrust-and-parry of network discussion, there are very few clear-cut ‘hits’, and debates are often perceived to have been ‘won’ by whichever party to the debate exhibits the greatest stamina. Getting in the last word is considered equivalent to winning.

Withdrawing from a discussion is often tantamount to withdrawing from a newsgroup, and this is another reason why many users do not see it as a viable option. As long as one party to a flame remains active on a group, it is more than likely that the flame will be rekindled the next time the withdrawing party posts a message, even if that message is addressed to someone else, or is concerned with a completely different thread.

A less certain, but more elegant, way of terminating a flame is by offering an unconditional apology to the person who feels hurt or offended.

Another option is to disregard the flame component of a response, and only address the parts relevant to the thread in question. This tactic can be effective in blurring the focus of an attack, but it can also produce an effect opposite to that intended: it can add fuel to the flame by making the other person feel that their (presumably valid) criticisms were being ignored.

The above guidelines are intended to address many of the difficulties of newsgroup communication, but even in the unlikely event that a majority of participants were to follow these guidelines, there would still remain many unresolved problems.

The first study of this series (Buda 1991) suggested that technical improvements in the medium would someday overcome many or all of the physical constraints and limitations of the present system, leaving only the dual problems of a pervasive lack of basic communicative skills and an over-sensitivity to network personae.

In the five years covered by the present series of studies, the Usenet network of newsgroups has grown exponentially, but there has been little or no change in the technical features of the system itself, and the quality of the interaction between participants has, if anything, deteriorated still further.

In the same period of time, a number of other paradigms of computer-mediated communication have appeared and developed. Perhaps the most significant of these is the World-Wide Web (WWW). For many people, indeed, the ‘Web’ has become synonymous with both the Internet and computer networking in general.

The Web provides access to linked multimedia data stored at multiple sites, and Web users are free not only to access such data but also to submit data of their own. They do this by setting up their own Web ‘sites’ in the form of home pages, and other Web users can then visit these sites and access the information stored therein.

Although most Web sites are still read-only, more and more are becoming interactive, allowing readers to fill in questionnaires, leave messages, or place orders for goods and services. Some sites are, indeed, offering capabilities indistinguishable from those of Bulletin Board Systems, supporting the reading and posting of messages on hierarchically-arranged discussion forums.

At present, the creation and support of such sites requires resources far beyond the reach of most individual net users, but if the rapid propagation of sophisticated individual home pages is anything to go by, it will not be long before the technology required to create and maintain such interactive sites diffuses down to the level of the average Web user. When that happens, each and every Web user will be in a position to create his or her own virtual single-node network, and links between such networks will create countless alternative multi-node systems.

It is difficult to imagine what kind of communication will take place on such virtual ‘alternative’ networks—whether it will adopt one of the existing models, or whether it will create an entirely new model of its own.

If the latter should be the case, it may become necessary to create a completely new set of quidelines for network communication.


Buda, J. K. (1991). “Electronic Network Communication.” Otsuma Women's University Annual Report: Humanities and Social Sciences XXIII.

Buda, J. (1995). “Quotations in Network Messages.” Waseda University School of Commerce Cultural Review 6.


1 The naming of hierarchical levels varies from board to board, and from network to network. Examples of alternative names are: SIG (Special-Interest Group), Conference Room, and Roundtable.

Source: Waseda University Sanken Series, No. 28, 1996