Toward an Ethics and Etiquette for Electronic Mail

Preface

Electronic mail and message systems are playing an increasing role in the work we perform. The effects, and side effects, of this new communication medium can be substantial. This report discusses a number of issues related to the use of electronic mail and presents a set of guidelines that should help lead to its effective use.

The report is not an introduction to electronic mail systems, computers, or communication systems. It does not survey existing mail systems or compare and contrast them. Rather, it is a discussion of some important general attributes of such systems, and the effects of those attributes on the quality and appropriateness of communication. The authors discuss the "etiquette" of sending and receiving electronic mail, drawing on personal observation of inappropriate or counterproductive use of these systems. By presenting some initial guidelines for their use, the authors hope to accelerate the process by which social customs and behavior appropriate to electronic mail become established, and thereby to accelerate the effective use of such systems.

The intended audience is persons possessing some familiarity with electronic mail systems, or considering adopting them for individual or institutional use. The guidelines discussed here may ease their transition to, and understanding of, this new and quite fundamentally different communication medium. In addition, the authors hope that the discussion will stimulate reflection by experienced users on their own evolving rules, and thus promote an exchange of views on appropriate electronic mail behavior.

The report was prepared with support from the National Science Foundation and from The RAND Corporation using its own funds.

Summary

Electronic mail and messaging systems, and electronic bulletin boards, are an incredibly powerful and effective means of communication. Because of this, they will grow and become one of the primary means of communication for most of us.

These media are quite different from any other means of communication. Some of the dimensions along which they differ are: speed (of initiating contact, and of transmitting information once contact is established); permanence of the message; cost of distribution, to individuals and to groups; an organization's desire and ability to filter, channel, record, and control messages; experience of both an individual and of our culture in dealing with this new medium.

Perhaps the most important phenomenon in electronic mail systems is the likelihood that the recipient will react negatively or inappropriately in reading material that might well have been misinterpreted. The misinterpretation results from several attributes of the medium that allow casual and formal messages to look superficially the same; that allow near-instantaneous, rather than reasoned, response; that don't permit feedback during the delivery of a message (as in personal conversation); and that require modification to many old traditions of communication. A related phenomenon is "flaming," in which emotions are expressed via electronic mail, sometimes labeled as such, and sometimes not. There is a need, even a greater willingness, to express emotion in electronic mail; if misused (for example, in hastily responding to a misinterpreted message), it impedes or even blocks communication.

A second very important phenomenon is the noncontrollability of who will see a message. Electronic messages seem quite evanescent, but in fact they can live on for years on disk archives, to reappear later in a variety of printed forms, some of which might be much more formal than was ever intended or foreseen.

Old rules of behavior in communicating do not automatically apply to this new medium. Some rules we have found useful for electronic mail (not all of which are unique to this medium) include:

In sending messages

• Create single-subject messages whenever possible
• Assume that any message you send is permanent
• Have in mind a model of your intended audience
• Keep the list of recipients and Cc:s to a minimum
• Separate opinion from non-opinion, and clearly label each
• If you must express emotion in a message, clearly label it
• Other content labels are useful
• Think about the level of formality you put in a message
• Identify yourself and your affiliations clearly
• Be selective in broadcasts for information
• Do not insult or criticize third parties without giving them a chance to respond.

In receiving and responding to messages

• If you receive a message intended for another person, don't just ignore it
• Avoid responding while emotional
• If a message generates emotions, look again
• Assume the honesty and competence of the sender
• Try to separate opinion from non-opinion while reading a message, so you can respond appropriately
• Consider whom you should respond to
• Consider alternative media
• Avoid irrelevancies.

In acting as a coordinator/leader of an interest group

• Perform relevant groupings
• Use uniform packaging, especially in the "Subject:" line
• Exercise reasonable editorship
• Timeliness is important.

Electronic mail is in its infancy, as is our understanding of it. We have collected some guidelines that seem to point in proper directions, and have personally used them in our own use of the medium. Many of them appear to be common sense in a new guise, but they are included because we've seen them violated in practice too often to ignore.

Electronic mail and messaging systems have novel characteristics that will lead toward their becoming a key, even dominant, communication medium in the coming decades. Understanding the unique attributes of this medium, and their effect on users, will help us all to avoid unwanted side effects while obtaining the benefits from this new and important means of communication.

I. An Example

II. What This Report Is About

III. Electronic Mail Is a Fundamentally New Medium

An electronic message often looks very ordinary. What is the difference between receiving:

     Date: Thursday, 19 May 1984 11:45-PDT
     To: anderson at rand-unix
     Subject: telephone message
     From: francine at rand-unix
     Please call Mrs. Cutler.  621-3208

     francine

The most obvious "media" with which to compare electronic mail are face-to-face conversation, telephone calls, notes, interoffice memos, and regular mail. (We could add telegrams, telephone answering machines, etc., but at the risk of making ponderous what we hope is a brisk, readable analysis.)

Some of the key dimensions along which electronic mail (EM) should be compared with these other media are:

• Speed (to generate a message, to transmit one, to respond to one)
• Permanence (of the resulting message, with respect to both the sender and the recipient)
• Cost of distribution (to an individual, or a group)
• Organizations' desire and ability to filter, channel, record, and control messages for the perceived good of the organization
• Experience (of an individual with the medium, and of the culture in preparing an individual to use it)
• Accessibility
• Security and privacy
• Accountability and attributability.

Telephone calls are nearly instantaneous, if you get through. Note that there are two aspects of speed that we're concerned with: (1) the time to initiate a communication and (2) the immediacy of the feedback obtained during or after the communication. Only about 30 percent of calls reach the intended recipient on the first try. Sometimes "telephone tag" takes days to reach someone, whereas that same person might be selectively available--at his or her convenience--to read an electronic message before being in the right place at the right time to make a telephone connection. Due to the delays often experienced in establishing a telephone call, EM and telephone calls may be comparably fast, with EM having a slight advantage in convenience, and telephone calls having the advantage if subtlety, humor, or privacy (non- attributability) is required.

Face-to-face conversation is usually fastest and best, except when you have to travel to do it. Then it's one to three orders of magnitude slower. Again, time-to-initiate can dominate time-to-carry-out.

Interoffice memos usually take a day or so; regular mail takes several days or a week. Given that an electronic message is usually received by the recipient (if he or she is an active EM user) within 2 to 12 hours--assume an average of 6 hours--EM is from 4 to 28 times faster.

When it is necessary to conduct a dialog, with repeated to-and- from messages, both parties check for messages more frequently, and thereby exchange many messages in hours, rather than weeks' or months' duration for multiple written exchanges.

The permanence (or conversely, the volatility) of messages varies greatly according to the medium by which they're transmitted. This characteristic of electronic messages is quite unlike any other medium.

U.S. mail and interoffice memos are nonvolatile. Messages sent this way usually have a responsible author and are "part of the record." (For example, they usually survive in various paper files and can be subpoenaed--sometimes years or even decades later--if they are an important part of some transaction.) Through this permanence, the author remains accountable for what is stated in print.

Face-to-face conversations, by contrast, are volatile. They leave no trace. Telephone conversations are similar; although they can be recorded, our society has established a set of legal protections against recording a telephone call without the remote party's knowledge and permission. Even if recorded, that recording often cannot be used as legal evidence.

Electronic messages appear on the surface to be quite volatile. You see them as flickering characters on a green phosphor, as evanescent as fireflies. But consider the following possible attributes of an electronic mail system:

• That message might reside indefinitely on a disk file, and can therefore be recalled.
• The disk file may be backed up each evening onto tape, so that a copy of the message is now buried in an archived tape in the recipient's institution, or on an archived diskette on the recipient's home computer.
• The message can be printed and filed, thereby instantly achieving some of the attributes of a printed memo or letter.
• The message can be altered, then printed, thereby looking like a permanent, authentic copy of the received message, but having altered characteristics.
• The message can be forwarded to third parties (and then fourth, and so on) at the push of a button, without the original author's knowledge. It therefore achieves a form of permanence through its replication in perhaps hundreds of computer systems throughout the country. But nothing indicates whether those are authentic copies or not, even though they might well have the original author's name attached.
• Printed copies of the message, no longer under the control of the author, can be laser-printed or typeset. These may appear much more official and substantial than was ever intended by the author at the keyboard of his PC late one evening.

A theme pervades consideration of the permanence of an electronic message. It is not clear to whom it belongs: the sender? receiver? the organization owning the computer and paying for the service? As multiple copies are made and filed, possibly on different machines, the issue becomes even murkier. Again, we have no answer, but raise the issue for consideration.

Given the strange permanence yet volatility of electronic messages, Colonel David Russell (USA)--when Head of the Information Processing Techniques Office of the Defense Advanced Research Projects Agency and a heavy user of electronic mail to communicate nationwide with various project leaders and institutions--had a simple rule: Never say anything in an electronic message that you wouldn't want appearing, and attributed to you, in tomorrow morning's front-page headline in the New York Times.

One of the authors once violated this rule, and made some rather derogatory comments in an electronic message about someone far away in another organization. Through some path, that message found its way to the subject, causing considerable embarrassment. What was said in haste and in some anger at a particular moment did not disappear into the ether as would a phone conversation; the potential permanence of the remarks in electronic form was overlooked, providing a valuable lesson at considerable cost.

It costs 22 cents to mail a first-class letter, as of this writing. After the letter is produced (perhaps costing several dollars' worth of secretarial time and overhead), the cost of sending 50 copies is about 50 times that of sending one. And it's not a very interactive medium; replies dribble back from within several days to several months.

Interoffice memos can be mass-reproduced and distributed at small cost, because interoffice distribution systems are already in place within an organization. However, interactivity is again poor and cumbersome.

A telephone call distributed to a group (a conference call) is difficult beyond three or four participants. And if the group is geographically distributed, the cost and time to initiate become important factors.

Electronic mail, by comparison, allows communication with a wide, geographically dispersed set of respondents. The communication can be highly interactive, if recipients are expecting mail and frequently check for new messages. If ordinary telephone lines are used, the cost is not burdensome, and can be borne in part by the recipients, not the sender. (For example, the message may be deposited in an information system by the sender, but each recipient dials up and thereby incurs the cost of the call to read the message.) At 1200 baud, a 400-character message (seven or eight lines) can be transmitted or received in 3-seconds' time over a telephone line.

Often, within modern organizations, the infrastructure for electronic mail is already in place, using either central computers with hundreds of terminals attached, leased phone lines, or satellite links, so that the incremental cost of electronic mail is nearly zero.

Many heavy users of electronic mail within the United States, primarily at research institutions, use the ARPAnet. The cost of this important network is heavily subsidized by the U.S. Defense Department, so that the true cost of using it is hidden from the user. In that sense, artificial patterns of use are springing up. On the other hand, an earlier form of communication network was also subsidized in its formative stages by the Defense Department: the Interstate highway system. (And earlier: railroads, telegraph, etc.) So the ARPAnet is not so artificial; rather, it is exploratory. And as the medium begins to mature--if that is not a contradiction for computer-based technologies--self-supporting systems will arise that pay their own costs, but have the same characteristics of ARPAnet pioneering systems. Notable among these latter systems is the UUCP (Unix-to-Unix Communication Protocol) system that links many computers throughout the nation. It permits a loose, heterogeneous system through which thousands of messages flow daily, with costs shared among all the participants as a natural result of its heterogeneity.

Traditionally, organizations have channeled and filtered their message flows along corporate hierarchical lines. For example:

• You do not send a memo to your supervisor's boss without a copy to your supervisor, and usually not without explicit prior permission.
• Secretaries filter incoming mail, telephone calls, and interoffice memos. For senior executives, ALL communications (other than in meetings and conferences) pass through this important filter.

These mechanisms have evolved to support the corporate structure, and to conserve the time and attention of its executives. Comparable mechanisms are not yet in place for electronic mail. Executives working in the evening at personal computers at home can send messages without "copying" their secretaries, resulting in those secretaries being "out of the loop" on matters of which they're normally aware. A junior executive can send a message to a senior executive, bypassing several levels of control. Electronic mail tends to be more "democratic" (some would say "anarchic"). Already, there is the electronic equivalent of junk mail:

     Received: from brl-tgr.arpa by rand-unix.ARPA; Sun, 6 Jan 85
     Received: from usenet by BRL-TGR.ARPA id a008108; 6 Jan 85 5:43 EST
     From: Joan Smith 
     Newsgroups: net.unix-wizards
     Subject: ancient history
     Message-Id: <1078@grant.UUCP>
     Date: 5 Jan 85 15:05:27 GMT
     To: unix-wizards@BRL-TGR

     For those of us who are TOPS-10 alumni, today is a tenth anniversary.
     What were the rest of you doing on January 5, 1975?

     --
     Joan Smith      AT&T Bell Laboratories, Murray Hill
     Phone:          (201) 582-1256
     Internet:       ggs@grant.uucp
     UUCP:           grant!ggs  ( {allegra|ihnp4}!grant!ggs )

Of course, one person's junk mail is another's important message. The above message was sent to a group called "unix-wizards," at least some of whom may have been interested in the message's contents. But for many, it might well be the equivalent of junk mail. Also in this category are notices about cheese buying clubs, upcoming ski trips, and so forth that clog the mailboxes of people who aren't interested in those topics.

Do we need unlisted electronic mailboxes? Will there be authorization lists (electronic files, of course) showing who can send a message to whom within the organization? Should message systems automatically send an information copy to the author's secretary, unless explicitly overridden by the sender? Will "back channels" of information flow, made much easier and in some cases more anonymous by electronic mail, erode the traditional corporate structure? Is this good or bad? For whom?

We don't have answers to all these questions, but as use of electronic mail grows within traditional organizations, their answers will evolve along with the medium itself. The medium is capable of supporting filtering, gateways, permission lists, and other constraints if they are necessary. And yet the explosion of use of, and interest in, electronic mail is certainly tied to some extent to its freedom, its interactive broadcast capabilities, and its democratic nature. All we can say now is that it is a fundamentally new medium with significantly new characteristics, that cannot be treated with the old rules alone.

People have had about 50,000 years' experience in the use of speech and gestures, 5,000 years' experience in writing, and about 100 years' use of the telephone. This cultural history should not be taken lightly; the entire fabric of our society has been shaped in significant part by cultural accommodations to our means of communicating.

As individuals of the species, living within a particular culture, we have a particular messaging history: From birth, we learn speaking roles and rules from conversations. By age 4 or 5, some basic telephone habits are learned (such as: "Say something when you pick up the receiver after it rings--don't just stand there silently"). By age 7, we are writing nontrivial messages. The average adult has accumulated hundreds--perhaps thousands--of rules of behavior regarding telephone and written ethics and etiquette, from practical experiences with these tools since those early years.

We have tried to indicate that electronic mail is different. Part of what we mean by that is that the old telephone or letter-writing rules of behavior do not automatically transfer over to this medium and work. You don't write business letters as electronic messages; messages are usually more informal. And yet electronic messages are not printed telephone conversations either. What we find is that the medium is different enough, and the average user's experience has been short enough, that problems arise. Meanings are misunderstood. Tempers flare and cause ill-conceived responses to be written. Many recipients' time is wasted reading content-free or irrelevant messages.

What we need is a new set of rules: how to be a constructive, courteous sender and receiver of electronic messages. We certainly do not have this set of rules, all tied up in a tidy package. We do, however, feel it is important to hasten the cultural evolution toward this goal. What follows, then, is a discussion of some of the important guidelines we've observed from experience. They are discussed in separate sections for Sending, Receiving, Responding, and Leading an Electronic Interest Group.

Create single-subject messages whenever possible

You may have three separate things to tell your intended recipient. We argue that three short separate messages are better than one. Some reasons:

• Each of the messages can be filed, retrieved, and forwarded separately by the recipient (and sender), depending on the content.
• Subject lines in each message can be descriptive of the contents of each message, allowing more meaningful scanning of header listings of the messages in one's inbox.
• Replies can be tailored to specific messages, so that the reply's subject line accurately reflects the content that's being replied to. Also, others can be copied on the responses that apply to them, without being burdened by the parts that do not.

Assume that any message you send is permanent

The message will be sitting in someone's private files, or in a tape archive. Through the miracle of computer networking, it can reside on computers elsewhere in the world that you don't know about, forwarded there without your knowledge or consent. It can appear in any form from dot-matrix-printed to typeset at any time in the future. If this has a chilling effect on the content of your informal, chatty messages, that's probably appropriate. At the very least, make a quick assessment of the risks and benefits of what you type, and act accordingly. Have in mind a model of your intended audience

When your message says, "Would you please review the draft document appended to this message, and give me your comments by noon tomorrow?" does that mean only people listed in the "To:" field, and not the "Cc:" recipients? Have you used more computer jargon in your message (lulled into techno-talk by using an electronic medium) than is appropriate to your audience?

Part of the model of your audience are some pertinent details such as their correct electronic mailing address. It is all too easy to send a message to "bob" because that's the one you know best, and ignore the fact that there are eight other "bob"s in the organization. In one company known to the authors, the login name "bob" belongs to the first Bob that joined the company; others are "bobe", "bobw", etc. It is impolite to send electronic mail to unintended recipients, if only because they then feel obligated to take action, like notifying the sender or attempting to forward the message to the right party.

Do you know the recipient well enough, and have other channels of communication with him or her, so that attempts at electronic humor or irony will not be misinterpreted? These attempts usually don't work, and appear quite differently in the cold light of a new day, a new computer, or a new context. Keep the list of recipients and Cc:s to a minimum

In one sense, it is too easy to send electronic mail. Electronic mailboxes fill up with peripheral material that needs to be scanned and continuously culled. If one of your recipients decides that someone else needs to see a message, it can be forwarded at that time.

Receiving messages is easier than sending them. If you want, you can just be a passing observer of the scene. Responding to messages is usually easy, as most systems provide a "reply" function that automatically creates the appropriate header for the response. In addition to the guidelines mentioned here, note that all "sending" guidelines apply while responding to a message.

One rule that we don't provide is: When is it rude not to respond to a message? For example, if you receive the message, "The meeting has been postponed `til 2pm," should you make the sender aware that you've received it on time? It probably depends too much on local context, such as whether the message system provides an automatic "receipt" message to the sender when you access a message. If you receive a message intended for another person, don't just ignore it

It's not good citizenship to ignore a message, or hit the "delete" key. The sender will assume the intended recipient got it, and wonder why he or she didn't get a reply. The intended recipient(s) won't know something they were intended to know. And so on.

If you know from the content of the message (of course you read it, even though you knew after two sentences it was missent) who the correct recipient is, you could forward it with a cover note explaining the error. If you're not sure, use your system's "reply" capability to notify the sender, preferably attaching the errant message to your response.

Avoid responding while emotional

See the following guidelines, which explain why it might be inappropriate to respond in an emotional state of mind. See the section "The Phenomenon of Flaming," below, regarding how to respond when you decide to.

If a message generates emotions, look again

One of the most surprising things about electronic mail is the ease with which misinterpretations arise. People are used to reading "body language," voice intonation, and numerous other cues when interpreting messages delivered in conversation, or even on the telephone. Those cues are missing in electronic mail, and what was meant as a casual comment, or an attempt at humor or irony, is misinterpreted. Even small misinterpretations have a tendency to mushroom. Messages between two correspondents may become more stilted and formal, until what started as a casual exchange of messages becomes a set of diplomatic communiques.

So it is easy to become angered at something in a message ("Boy, that was a dumb thing to say." "How could anyone be so silly.").

We've found that the reader should pause and reread the message. What at first glance was offensive can often be interpreted, on rereading, as merely a poor choice of words in a hasty message--words that might have been casually used, then forgotten, in a face-to-face conversation, but that linger on the printed page (or phosphor screen). It might help to consider the message as a written verbal communication, rather than real writing.

It also helps to consider the source. The sender might be a graduate student (or high school student) hacking away on some remote system, using a colloquial way of expressing him- or herself that is customary in that person's peer group, but that is inappropriate in the recipient's circle of communicants.

As we've mentioned, the most likely explanation is that an attempt at humor or irony went awry. Try to interpret the message that way before firing off your own cynical reply, thereby escalating the process to higher levels of misinterpretation. There is danger that your response can cause a chain reaction of emotional responses, or "flaming," that is very counterproductive. Assume the honesty and competence of the sender

Giving someone the benefit of the doubt isn't a bad rule of thumb, especially when they're 3000 miles and three time zones away. (This rule may seem too obvious to mention, but we have observed numerous examples of replies to messages that appear to assume the original sender is an idiot, whereas the recipient might well have misunderstood the context or intent of the message.) Try to separate opinion from non-opinion while reading a message, so you can respond appropriately

The sender, of course, should have labeled opinion as such. In case he hasn't, it is worthwhile trying to unravel opinion from fact, since your reply will benefit from making the distinction. Consider whom you should respond to

If the message was sent to a distribution list, do you really want your answer to go out to that same list? Wouldn't it keep the electronic clutter down to respond only to the sender, even if that means editing out the name of the distribution list in the "Cc:" field your text editor so helpfully supplied?

There's a nice compromise: Send an answer only to the sender, with the P.S.: "If you think this response merits wider distribution, feel free to do so." That way, the original sender can batch together responses received, and provide a coherent update to the issue (giving you credit, of course, for your insightful contribution to the debate). Consider alternative media

Can you walk down the hall, or pick up the phone, and respond better? Was there something in the original message that needs clarification, so that a real conversation might well cut through to the heart of the issue, rather than starting a string of messages and responses?

Recall our earlier caution that a string of messages and responses tends to become brittle and may lead to misunderstandings in a way that personal conversations do not--a fertile field for behavioral analysis. The lack of additional cues, in our experience, almost invariably leads to misinterpretation, especially between parties who do not know each other very well. So consider reaching out and really touching someone. Avoid irrelevancies

We've mentioned the growing role of special interest groups in electronic mail networks. They perform several very useful functions: (1) focus on one subject matter, so that there is continuity and coherence to the dialog; (2) bring together diverse individuals or institutions interested in a common subject matter; (3) provide a repository of expertise in an area that can occasionally be tapped by others.

The activity and diversity of these groups are illustrated by a list of special interest groups within the USENET community (as of November 1984). (The first 20 groups are listed here; a complete list is given in the Appendix.)

     net.abortion                 All sorts of discussions on abortion.
     net.ai                       Artificial intelligence discussions.
     net.analog                   Analog design developments, ideas,
                                    and components.
     net.announce                 Moderated, general announcements
                                    of interest to all.
       net.announce.newusers      Moderated, explanatory postings
                                    for new users.
     net.arch                     Computer architecture.
     net.astro                    Astronomy discussions and information.
       net.astro.expert           Discussion by experts in astronomy.
     net.audio                    High fidelity audio.
     net.auto                     Automobiles, automotive products and
                                    laws.
     net.aviation                 Aviation rules, means, and methods.
     net.bicycle                  Bicycles, related products and laws.
     net.bio                      Biology and related sciences.
     net.books                    Books of all genres, shapes, and sizes.
     net.bugs                     General bug reports and fixes.
       net.bugs.2bsd              Reports of UNIX version 2BSD related
                                    bugs.
       net.bugs.4bsd              Reports of UNIX version 4BSD related
                                    bugs.
       net.bugs.usg               Reports of USG (System III, V, etc.)
                                    bugs.
       net.bugs.uucp              Reports of UUCP related bugs.
       net.bugs.v7                Reports of UNIX V7 related bugs.

The success of these groups is often highly correlated with there being a coordinator or leader who takes responsibility for group communications. It's not a simple or easy job, but it is a valuable service. (For example, the ARPAnet IBM-PC interest group now has three different editors, on a rotating basis, to handle the volume of messages.)

We've listed below some key guidelines for the performance of this coordinator/leader job. It's a role that will be even more commonplace and important as the volume of electronic communication increases. Perform relevant groupings

It is helpful to readers when messages received on a common topic from diverse sources are grouped together in a "packet" message. Readers may well detect common threads or issues that would otherwise have remained obscure. Also, the packet can be filed by subject matter as one unit, not many. Use uniform packaging

If some part of the message header of messages routed within an interest group have some key word or phrase in common, these messages can be filtered out and organized by recipients using "scan" and "file" functions common in many message systems. Perhaps this can be as simple as the "To:" line containing the name of the interest group.

Exercise reasonable editorship

Perhaps a world without censorship would be nice, but we're not there yet. Messages that are not relevant should be excluded, as should ones that are sufficiently tasteless to be offensive. But it is important that opinions (preferably labeled as such) be given a hearing.

There also tends to be much redundancy of messages and questions in these interest groups. New people are joining all the time, and asking questions that have been answered before. The group coordinator provides a very useful function by excluding these messages from continued widespread distribution, and pointing the sender (individually) to the group archives for the answer. If it is a topic that appears to be of extreme continuing interest, periodic broadcast messages can alert new participants to the relevant archives. Timeliness is important

Perhaps the attribute of electronic mail systems that most distinguishes them from other forms of communication is their propensity to evoke emotion in the recipient--very likely because of misinterpretation of some portion of the form or content of the message-- and the likelihood that the recipient will then fire off a response that exacerbates the situation.

We have touched upon various possible causes for this phenomenon. They are:

• It is difficult to tell the level of formality of a message from its appearance; to a considerable degree, they all look the same. The cues are more subtle than telling the difference between a scrawled note and a formal memorandum.
• Partly because of the lack of cues to the level of formality, because of the nature of writing, and because most participants are not professional writers, attempts at humor, irony, sarcasm, and wit are often misinterpreted.
• Immediate feedback from body language, interruptions, or other cues we have developed as a society to aid the intercommunication process is lacking in this medium.
• Normally in written communications, time intervenes to blunt the edge of a response, or to allow reconsideration. A written letter that is received may lie on the desk several days or weeks before it is responded to. In contrast, the ease of creating an immediate "reply" to an electronic message (often easiest to do immediately upon viewing the message) biases the EM user to respond immediately, "off the top of his head."
• Telephone calls and personal conversations that have involved hasty or ill-chosen words fade with time. Electronic messages containing similar infelicities have a permanence to them: They sit around in electronic inboxes, or are printed out and remain tangible, and can even be printed in a manner (inkjet, laser, or typesetting) that gives them an aura of formality and importance that was never intended.

One additional factor often mentioned is anonymity. It would appear that persons sending electronic mail to others over a network who are not known in person might be freer in communicating feelings than to friends or associates. If no one knows who "fritz at cmu-ca" is, fritz can say almost anything. In fact, we have not observed significant difference in "flaming" between remote correspondents who don't know each other personally, compared with communication among people who know each other. The anonymity factor does not appear to be an important one.

What can be done to minimize the problems of escalating emotions that arise? A number of the guidelines and suggestions we have listed earlier are relevant to this issue. To summarize:

• A phenomenon called "flaming" has appeared on electronic mail networks, in which messages are sent having a deliberate emotional content, but usually carefully labeled as such. Sometimes just the annotation "Flame! Flame!" alerts the reader to the fact that the writer knows he or she is being emotional. The intent is that the reader should take that into account and not assume this is a carefully reasoned statement (although it might be; the two are not mutually exclusive).
• Resist the temptation to fire off a response. Go ahead and write the response, but file it away instead, and wait 24 hours. Reconsider the response later, in the light of a new day (and perhaps a rereading and reinterpretation of the original message).
• Use alternative media to break the cycle of message-and- response. A telephone call or personal conversation can do wonders, when body language, eye contact, and all the other cues we've developed can take effect. This is especially important if electronic communications seem to be becoming more formal and stilted than seems natural.

Much of the problem seems to stem from the paucity of cues that electronic mail affords its readers. Perhaps the technology that spawned electronic mail will go further to help with the misunderstandings it creates. One can imagine message systems in which the boldness of the characters displayed is a function of the force with which the keys are hit; in which the speed at which it is typed is reflected in the character spacing (or color, or size, etc.). Or providing a set of standard forms to be selected, ranging from "Note from the desk of . . ." to "Corporate Memorandum" to give additional cues to the level of formality intended. Perhaps the most informal messages will be displayed in the handwriting of the sender (even though keyboarded for convenience) as an additional cue to its informality. More certainly (because the systems are in prototype form already) there will be systems in which the cold green (or amber, or whatever) characters will be accompanied by voice annotations, so that the humanity and state of the sender will be retained and "read" by the recipient.

In the meantime, caution, awareness, and an evolving ethics and etiquette of electronic communication will certainly help.

V. In Conclusion

Electronic mail and messaging systems, and electronic bulletin boards, are an incredibly powerful and effective means of communication. Because of this, they will grow and become one of the primary means of communication for most of us.

These media are quite different from any other means of communication. Many of the old rules do not apply.

This discussion does not supply a new set of rules for this new medium. Electronic mail is in its infancy, as is our understanding of it. We have collected some guidelines that seem to point in proper directions, and have personally used them in our own use of the medium. Many of them appear to be common sense in a new guise, but they are included because we've seen them violated in practice too often to ignore. These guidelines are suggestions, intended to generate reflection and stimulate discussion.

With the new power of electronic mail comes the need for responsibility in using that power. We can all enjoy the power and benefit from it if we find new forms of behavior--even etiquette--that are appropriate. The alternative is a rising tide of irrelevant messages and electronic junk mail that will turn off most thoughtful users. By evolving a set of guidelines such as those presented here, we can all use the incredible power of the medium and benefit from it.

Appendix

Network Interest Groups

       Newsgroup                     Description

net.abortion             All sorts of discussions on abortion.
net.ai                   Artificial intelligence discussions.
net.analog               Analog design developments, ideas, and
                           components.
net.announce             Moderated, general announcements of
                           interest to all.
  net.announce.newusers  Moderated, explanatory postings for new
                           users.
net.arch                 Computer architecture.
net.astro                Astronomy discussions and information.
  net.astro.expert       Discussion by experts in astronomy.
net.audio                High fidelity audio.
net.auto                 Automobiles, automotive products and laws.
net.aviation             Aviation rules, means, and methods.
net.bicycle              Bicycles, related products and laws.
net.bio                  Biology and related sciences.
net.books                Books of all genres, shapes, and sizes.
net.bugs                 General bug reports and fixes.
  net.bugs.2bsd          Reports of UNIX version 2BSD related bugs.
  net.bugs.4bsd          Reports of UNIX version 4BSD related bugs.
  net.bugs.usg           Reports of USG (System III, V, etc.) bugs.
  net.bugs.uucp          Reports of UUCP related bugs.
  net.bugs.v7            Reports of UNIX V7 related bugs.
net.chess                Chess and computer chess.
net.cog-eng              Cognitive engineering.
net.college              College, college activities, campus life, etc.
net.columbia             The space shuttle and the STS program.
net.comics               The funnies, old and new.
net.consumers            Consumer interests, product reviews, etc.
net.cooks                Food, cooking, cookbooks, and recipes.
net.crypt                Different methods of data en/decryption.
net.cse                  Computer science education.
net.cycle                Motorcycles and related products and laws.
net.dcom                 Data communications hardware and software.
net.decus                DEC Users' Society newsgroup.
net.emacs                EMACS editors of different flavors.
net.eunice               The SRI Eunice system.
net.flame                For flaming on any topic.
net.followup             Followups to articles in net.general.
net.games                Games and computer games.
  net.games.emp          Discussion and hints about Empire.
  net.games.frp          Discussion about Fantasy Role Playing games.
  net.games.go           Discussion about Go.
  net.games.pbm          Discussion about Play by Mail games.
  net.games.rogue        Discussion and hints about Rogue.
  net.games.trivia       Discussion about trivia.
  net.games.video        Discussion about video games.
net.garden               Gardening, methods and results.
net.general              *Important* and timely announcements of interest
                           to all.  (Note the description of net.misc.)
net.graphics             Computer graphics, art, and animation.
net.ham-radio            Amateur Radio practices, contests, events, rules.
net.info-terms           All sorts of terminals.
net.invest               Investments and the handling of money.
net.jobs                 Job announcements, requests, etc.
net.jokes                Jokes and the like.  May be somewhat offensive.
  net.jokes.d            Discussions on the content of net.jokes articles.
net.kids                 Children, their behavior and activities.
net.lan                  Local area network hardware and software.
net.lang                 Different computer languages.
  net.lang.ada           Discussion about Ada.
  net.lang.apl           Discussion about APL.
  net.lang.c             Discussion about C.
  net.lang.f77           Discussion about FORTRAN.
  net.lang.forth         Discussion about Forth.
  net.lang.lisp          Discussion about LISP.
  net.lang.mod2          Discussion about Modula-2.
  net.lang.pascal        Discussion about Pascal.
  net.lang.prolog        Discussion about PROLOG.
  net.lang.st80          Discussion about Smalltalk 80.
net.legal                Legalities and the ethics of law.
net.lsi                  Large scale integrated circuits.
net.mag                  Magazine summaries, tables of contents, etc.
net.mail                 Proposed new mail/network standards.
  net.mail.headers       Gatewayed from the ARPA header-people list.
  net.mail.msggroup      Gatewayed from the ARPA MsgGroup list.
net.math                 Mathematical discussions and puzzles.
  net.math.stat          Statistics discussion.
  net.math.symbolic      Symbolic algebra discussion.
net.med                  Medicine and its related products and regulations.
net.micro                Micro computers of all kinds.
  net.micro.16k          National Semiconductor 32000 series chips.
  net.micro.432          Discussion about Intel 432's.
  net.micro.6809         Discussion about 6809's.
  net.micro.68k          Discussion about 68k's.
  net.micro.apple        Discussion about Apples.
  net.micro.atari        Discussion about Ataris.
  net.micro.cbm          Discussion about Commodores.
  net.micro.cpm          Discussion about the CP/M operating system.
  net.micro.hp           Discussion about Hewlett/Packards.
  net.micro.mac          Material about the Apple MacIntosh and Lisa.
  net.micro.pc           Discussion about IBM personal computers.
  net.micro.ti           Discussion about Texas Instruments.
  net.micro.trs-80       Discussion about TRS-80's.
  net.micro.zx           Discussion about zx's.
net.misc                 Various discussions too short lived for other
                           groups.  Also items of a general nature not
                           important enough for net.general or
                           net.announce.
net.motss                Issues pertaining to homosexuality.
net.movies               Reviews and discussions of movies.
  net.movies.sw          Discussions about the Star Wars saga(s).
net.music                Music lovers' group.
  net.music.classical    Discussion about classical music.
net.net-people           Announcements, requests, etc. about people on
                           the net.
net.news                 Discussions of USENET itself.
  net.news.adm           Comments directed to news administrators.
  net.news.b             Discussion about B news software.
  net.news.config        Postings of system down times and interruptions.
  net.news.group         Discussions and lists of newsgroups.
  net.news.map           Postings of maps.
  net.news.newsite       Postings of new site announcements.
  net.news.sa            Comments directed to system administrators.
net.nlang                Natural languages, cultures, heritages, etc.
  net.nlang.celts        Group about Celtics.
  net.nlang.greek        Group about Greeks.
net.notes                Notesfile software from the University of Illinois.
net.origins              Evolution versus creationism (sometimes hot!).
net.periphs              Peripheral devices.
net.pets                 Pets, pet care, and household animals in general.
net.philosophy           Philosophical discussions.
net.physics              Physical laws, properties, etc.
net.poems                For the posting of poems.
net.politics             Political discussions.  Could get hot.
net.puzzle               Puzzles, problems, and quizzes.
net.railroad             Real and model train fans' newsgroup.
net.rec                  Recreational/participant sports.
  net.rec.birds          Hobbyists interested in bird watching.
  net.rec.boat           Hobbyists interested in boating.
  net.rec.bridge         Hobbyists interested in bridge.
  net.rec.coins          Hobbyists interested in coin collecting.
  net.rec.disc           Hobbyists interested in disc activities.
  net.rec.nude           Hobbyists interested in naturalist/nudist
                           activities.
  net.rec.photo          Hobbyists interested in photography.
  net.rec.scuba          Hobbyists interested in SCUBA diving.
  net.rec.ski            Hobbyists interested in skiing.
  net.rec.skydive        Hobbyists interested in skydiving.
  net.rec.wood           Hobbyists interested in woodworking.
net.religion             Religious, ethical, and moral implications of
                           actions.
  net.religion.jewish    Group about Judaism.
net.research             Research and computer research.
net.roots                Genealogical matters.
net.rumor                For the posting of rumors.
net.sci                  General purpose scientific discussions.
net.sf-lovers            Science fiction lovers' newsgroup.
net.singles              Newsgroup for single people, their activities, etc.
net.social               Like net.singles, but for everyone.
net.sources              For the posting of software packages and
                           documentation (cf. net.wanted.sources).
  net.sources.bugs       For bug fixes and features discussion
                           pertaining to items in net.sources.
net.space                Space, space programs, space related research, etc.
net.sport                Spectator sports.
  net.sport.baseball     Discussion about baseball.
  net.sport.football     Discussion about football.
  net.sport.hockey       Discussion about hockey.
  net.sport.hoops        Discussion about basketball.
net.startrek             Star Trek, the TV show and the movies.
net.std                  All sorts of standards (e.g., ANSI, IEEE).
net.suicide              Suicide, laws, ethics, and its causes and effects.
net.taxes                Tax laws and advice.
net.test                 For testing of network software.  Very boring.
net.text                 Text processing.
net.travel               Traveling all over the world.
net.tv                   The boob tube, its history, and past and current
                           shows.
  net.tv.drwho           Discussion about Dr. Who.
  net.tv.soaps           Postings about soap operas.
net.unix                 UNIX neophytes group.
net.unix-wizards         Discussions, bug reports, and fixes on and for
                           UNIX.  Not for the weak of heart.
net.usenix               USENIX Association events and announcements.
net.usoft                Universal (public domain) software packages.
net.veg                  Vegetarians.
net.video                Video and video components.
net.wanted               Requests for things that are needed.
  net.wanted.sources     Requests for software, termcap entries, etc.
net.wines                Wines and spirits.
net.wobegon              "The Prairie Home Companion" radio show.
net.women                Women's rights, discrimination, etc.
  net.women.only         Postings by women only (read by all).
net.works                Assorted workstations.
  net.works.apollo       Discussion about Apollo workstations.
fa.arms-d                Arms discussion digest.
fa.arpa-bboard           ARPAnet bulletin board.
fa.bitgraph              The BBN bitgraph terminal.
fa.digest-p              Digest-people digest.
fa.editor-p              Editor-people digest.
fa.energy                Energy programs, conservation, etc.
fa.human-nets            Computer aided communications digest.
fa.info-mac              Apple MacIntosh micros.
fa.info-terms            All sorts of terminals.
fa.info-vax              DEC's VAX line of computers.
fa.info-vlsi             Very large scale integrated circuits.
fa.laser-lovers          Laser printers, hardware and software.
fa.poli-sci              Politics and/versus science.
fa.railroad              Real and model train fans' newsgroup.
fa.tcp-ip                TCP and IP network protocols.
fa.telecom               Telecommunications digest.
fa.teletext              Teletext digest.
mod.map                  Announcements and software concerning maps and
                           routing.
  mod.map.news           Maps of the USENET network of news sites.
  mod.map.uucp           Maps from the UUCP mapping project.
mod.movies               Moderated reviews and discussion of movies.
mod.motss                Moderated newsgroup on gay issues and topics.
mod.music                Moderated reviews and discussion of things musical.
mod.newslists            Postings of news-related statistics and lists.
mod.singles              Moderated version of net.singles.
mod.sources              Moderated postings of public domain sources.
mod.test                 Testing of moderated newsgroups--no moderator.
mod.unix                 Moderated discussion of UNIX features and bugs.
mod.std                  Moderated discussion about various standards.
  mod.std.c              Discussion about C language standards.
  mod.std.mumps          Discussion about standards for MUMPS.

Glossary