There is the objective of the task; this has to do with whether you're trying to reach a decision, whether you're trying to explore an area, and you know, this is not new. If you go back there's quite a literature on the facilitation of face to face groups and many people who went through long years of training to be professional facilitators of face to face groups, and a lot of that facilitation had to do with 'how do you structure the group process to accomplish a specific objective'.

In fact, I have borrowed from some of that literature in some of my work and I'm sure others... perhaps more people in the computer field should be doing that. Some of them don't know about it, unfortunately. The nature of how structured the task is, from well structured to what's called 'messy' (Aykoff 1); the group atmosphere, whether it's a cooperating or a conflicting group; the inquiry process: by what methodology do these people agree on the nature of truth or validity, (C. West Churchman 2); and the nature of the group: is it experts, is it learners, is it homogeneous, is it heterogeneous? All these factors can affect the way you decide to structure a system.

We started a number of years back at NJIT to try to develop an approach that would allow for the multitude of structuring that one wanted if one was going to support general populations. So, what I'd like to do is give you a little explanation of the metaphor we developed and sort of the basis of the system we've developed at NJIT, but I'm really here to discuss the ideas of the system, and not the system itself, so I won't go into details on the specific system.

Overhead #5

CMC DESIGN METAPHOR OBJECTS

DIRECTORY

NOTIFICATIONS

MAIL & MESSAGES

CONFERENCES

MEMBERSHIP

COMMENTS

ACTIVITIES

Systems need mail and messages as a very integral part; it is not sufficient to say, 'Oh, you can leave the conference system, and go use the network mail system', if you can figure out how to do that, because mail serves a tremendously complementary purpose to a conference, in that you don't want a conference cluttered up with transient items; there are a lot of things: I can say, 'send a message to all the members of this conference within the conference system'; you want the ability to easily move reference material, so I can show you what's in a conference you're not in by sending you a message that gives you a link back to the material I want to show you, so [there is] a good strong argument that you need mail messages integrated into the conference facilities, particularly for ease of use and some of the things that facilitates.

Then there's a conference which normally has membership so everyone knows who has read what, and it is also desirable to know how much an individual in the membership list has written: their degree of participation. In many conference applications you want to [know this]. I can tell you some examples where it is crucial, in fact.

Overhead #6

CMC METAPHOR I

DIRECTORY

MEMBERS & GROUPS

INTEREST INDEX

CONFERENCES

TOPIC INDEX

GROUP

We added two things that you won't find in most conference systems today: one is the concept of notifications and the other is the concept of activities. And there are some other ways of structuring, but the heart of the structuring of conferencing in our metaphor is the fact that notifications and activities are open-ended; they can be almost anything that we want them to be, and I want to explain both notifications and activities. I'm expecting that most of you have at least seen some sort of conference system, being here at Simon Fraser, and the fact that work has gone on.

Let me just elaborate first, to the second level of this metaphor.

For example, with the directory, you can have things like an interest index, where people can associate themselves with terms, like 'ham-operator', or 'scuba diving', and then you can send messages to terms; so if I send a message to the term in the index, 'scuba diving', it goes to everyone who has decided to associate themselves with that, and similarly, there can be a topic index for conferences. Members in these systems need a personal filing system, or index system. In fact, we've incorporated the ability, where anything you receive, you can index, and you can share your indexing with members of one of your groups. So if you're gathering material for a group as a whole, you can share your personal index with other people, and then people can work together on gathering material.

Roles and privileges and tickets; this gets into a detailed area, but to support roles, you need certain privileges that are not typical in most computer systems, merely to read or write a file, or do whatever: for example, you give a ticket to your secretary to edit one comment that you've written in a conference that she's not a member of. You don't want her to see everything else that's in that conference. The ticket is like extending a privilege to do something, that you are able to do, but the person you're giving it to, is not normally able to do. And if you think about your personal interactions in any project department, or whatever you're working on with other people, you're always extending privileges to do something of some sort, so why not be able to do it in an organized manner electronically. Conferences also have to have indexes

Overhead #7

CMC METAPHOR II

CONFERENCE

MEMBERSHIP LIST

KEY INDEX

COMMENT

ABSTRACT

CONTENT

APPENDAGES

ACTIVITIES

ATTACHMENTS

MAIL

I've been involved in some conferences that have gone as high as ten or fifteen thousand comments; I mean this is a data base... projects that have gone on for a couple of years, and new members come in. Before you bring up an idea you go back and search [to see] if it's already been discussed and things of that sort. I think I'll leave this until I talk about activities, but the point is, that every comment now can have something attached to it, which we tend to break down into the categories of appendages, activities and attachments.

Overhead #8

NOTIFICATIONS

TRANSACTION NOTICES
DIRECT MANIPULATION
PERSONAL TRACKING FILE
CANNED NOTIFICATIONS
WRITTEN NOTIFICATIONS
STATUS CHANGES
TAILORABLE WITH ACTIVITIES
APPEND TO COMMENTS

Notifications are a very short, usually one line notification to a user about something that he or she needs to know about. And you have a list of received notifications so that you can go back and review these notifications; you can go back and reorganize these notifications; there are also handles, direct manipulation handles for retrieval. If I get a notification that Linda received a message I sent and she hasn't been on line for three weeks, I can point to the notification and it will retrieve the message to remind me of what it was that she received; I don't have to go looking for it.

Overhead #9

NOTIFICATIONS OBJECTIVES

BRIEF ONE LINE ALERTS
TRANSACTIONS INFORMATION
QUICK FEEDBACK TO GROUP
HANDLE FOR RETRIEVAL
SORTING & ORGANIZATION FILE
TRACKING STATUS
ACTIVITY PROMPTING

We put notifications in as a way of providing a mechanism to alert people, in the shortest amount of information flow, to things they need to know about. And as a concept, it's open-ended. We can always tailor and add more types of notifications as we need them. Now here are some typical examples.

Overhead #10

NOTIFICATIONS EXAMPLES

VOTE REQUIRED
VOTE VIEWABLE
JOHN AGREES WITH COMMENT
NEW GROUP MEMBER: NAME
TASK RESPONSIBILITY ASSIGNED
REPLY TO REQUEST OF YOURS
BOSS WANTS TO SEE YOU
MEETING TODAY
DATA NEEDED

Someone has created a vote, and it is required that you now vote; somewhere in the system, it tells you where, or you can use the notification to get that voteable issue. If you're doing voting, you don't want to let people see votes, until enough votes are in, otherwise you get all sorts of biases. So you need to notify people when a particular thing they voted on is now viewable, so they don't have to get on, look at it as a viewable, do all that gyration; they just get told when they can see it. Voting, by the way, is a type of activity, so that a lot of notifications are peculiar to the type of activity that you establish. When you're put into a conference you need a notification that tells you someone has put you into conference 'so-and-so'. Now there are other types: you can have a set of what we call 'canned' notifications, and these can be anything you want, but typically in an organization, some nice 'canned' notifications you need are: I agree with that; I disagree with that.

Why write a message? Point to your state of mind, and it is automatically sent to the people who saw that thing; and it may even be appended to the actual content, particularly if it's a notification having to do with the approval of an action, or something of that sort. You can have 'time' notifications that remind everyone that there's a meeting later today, that you set about three weeks ago for that morning. So 'notifications' is a very open-ended concept and it was purposely designed that way.

Linda Harasim: So for example, one of my problems, if you have three or four people in a class of forty people and they're working in small groups, and they get to decide which group they want to work in, as soon as that group is full, they have to go look for another group. And so a notification would say, well, the group on 'apples' is full now, and maybe it would show a list of which ones were still empty.

You would have that; when Roxanne puts in a set of grades, she says, 'send notifications', and then everyone is told what grade they got in the notification; so you don't have to send any messages; it's very tedious. A student says to you, "What's my grade?", in a message, and you say to him, "It's so-and-so", and you're doing that for thirty students. So 'notifications' serve an awful lot of purposes that keep people on top of what's going on, and minimize the flow of information and the amount of interaction that they have to do.

Linda Harasim: Would that be something like where they have an agenda keeper to say, "This group is now finished all their tasks and go on to the next group, go on to the next part", or something?

Well in a sense... I was going to get to that. One type of activity is where you're gathering a list. It's very common. The decision group wants to gather all the criteria by which we should evaluate this decision, or what are the goals, or what are the tasks to get this project... In many group processes, the one key, common process is: here's a group of us and we've got to gather a list of items. And there's someone in control. And that's an activity and you set up an activity where anyone can contribute to that. So you get a notification that says, "This activity is ready to go; we're gathering criteria". The person who is in charge of the process says, "We've got enough, now." He closes the activity and everyone gets notified that this activity has been frozen; you can't add any more. And then he may decide he wants to have that list voted upon. And so everyone is told, "We're voting on that according to a certain voting method." That's another notification, always associated with the idea of a list gathering activity.

So, we go into activities, and activities can be linked or created along with any comment. So normally if you have a comment in a conference, the content of the conference, of the comment, when you go to read it, would describe what the activity is.

Overhead #11

ATTACHMENTS

LINKED TO COMMENT
PROVIDES STORAGE
PASSIVE

e.g. LONG TEXT ITEMS
     OBJECT LINKS (VIRTUAL)
     BINARY FILES
     GRAPHICS

Question: Could you give me an example of what you mean by an activity?

Overhead #12

ACTIVITIES OBJECTIVES

LARGE DOCUMENTS
FORMS TO FILL OUT
EXECUTABLE PROGRAMS
COLLECTIONS PROCESSES
RESOURCE CONTROL
TRACKING PROCESSES
COMMUNICATION PROCESSES
COLLABORATIVE MODELS

The comment might be a statement of a task and the voting on it is how desirable it is for us to do. So what you do, you then ask to supply a vote on a scale of desirability. Another activity is a document. You don't want to get, in a conference, a 400 line comment. If any of you have experienced this, you realize that's not what you want. What you want is an abstract; a paragraph that describes what's in those 400 lines and the 400 lines is attached as a passive activity; it may not be completely passive, but it's attached, and when you say, "do" the document, the activity, it then might give you a table of contents and you can choose pieces of that document to look at, or it might just give you the 400 lines; it depends how the author wanted to set it up.

Overhead #13

ACTIVITIES

LINKED TO COMMENT
EXECUTABLE PROGRAMS
ATTACHED TO COMMENTS
REQUIRED/ALLOWED
DONE/UNDONE
TRACKED BY MEMBERSHIP
GENERATE NOTIFICATIONS

e.g. DOCUMENT
     QUESTION
     VOTE
     FORM
     LIST

Overhead #14

ACTIVITIES EXAMPLES

BUDGET ESTIMATES
VOTING CASE
TRACKING TASK SELECTION
DATABASE QUERIES
APPROVAL PROCESSES
GROUP CALENDAR
ALTERNATIVE COLLECTION
NOMINAL GROUP PROCESSES

So when I say that it can be any piece of executable code in the computer, I mean it. That activity can be anything, any computer resource, any information resource, a virtual window to another system; it's completely open-ended. So whatever structures, facilities, you want for a particular group... a whole interface facility for this is done in Smalltalk, so any programmer who knows Smalltalk, can set up any interface in an activity to any computer facility available in the network the conference system is on. And you can then integrate anything you want into a conference environment. Is that sufficient? Questioner: yes.

So I really mean it when I said we set up the concepts of activities and notifications to allow completely open-ended tailoring of a given conference. When a person sets up a conference for their group, they choose what activities out of those available, they want to allow, and who they want to allow to create those activities. And that is a major method in which the tailoring is done, so that the rest of the process is taking the basic system and adding activities that are appropriate to specific applications. We've got a group of Ph.D. students who, as part of each of their Ph.D.s, they are adding activities in certain application areas that they want to experiment with that application and so we also have the motive of setting our system up as a research support vehicle for Ph.D. students and our own research of course. And I guess the two areas we've been working the most in, has been the educational area and the decision support area... has been our primary concentration where a lot of the work is being done right now.

Overhead #15

LIST GATHERING ACTIVITY

EXAMPLES

CONTROL

NOTIFICATIONS

List Gathering Activity. You can have alternatives, decisions, criterias to evaluate decisions, factors to be considered, different stakeholders, but almost any group process has some sort of list people gather. Control, by the person who is running the group or is facilitating the group, has to do with opening and closing the list, how many levels of comments he wants, whether they want to be divided up into pro and con comments, when to start a voting phase for the list. And notifications would be generated when you have new entries, status changes, when votes are required, when votes are reviewable. So this just shows one activity and the fact that there is a need for a human to act in the role of controlling the activity, there is a need for notifications to keep the people informed of the fact of what's happening and what transactions have occurred.

Overhead #16

CONFERENCING STRUCTURING

MEMBERSHIP AND ROLES
REPLY LEVELS
COMMENT SIZE
WRITE AUTHORIZATION
INDEX METHODS & FILTERS
ROOT COMMENT ASSIGNMENTS
TYPES OF ALLOWED:

One could go into activity examples in areas like planning and estimating budgets, something you find in the university, a lot of energy is expended upon. Maybe Simon Fraser is better run. Voting, tracking cases in an organization that have to be treated, selecting tasks to do; collecting data base queries that are of interest to a group, the approval process; a group calendar is another type of activity, and there are various group processes for brainstorming and so forth, that can be incorporated as activities.

Question: Announcements, are they an activity or a notification; if you want to make any kind of announcement of a change...

That would be a notification; in other words, any time you want to announce something in one line to people, you use a notification, and there are a lot of 'canned' notifications and when you design an activity, you design notifications that go with the activity; so there's usually always a 'creation announcement'... that says, 'so-and-so' has created a voteable item in conference 'xyz'. It will go out and let everyone in that conference know that this voteable item is there to be voted on. The other thing about notifications is they come into your area and you have a stored list of these so you can re-sort these and organize them. Let me put it this way; one of the problems is that when you think about how you use communications, any individual communication, someone on the phone, someone comes into your office, is not an isolated instance.

It is usually part of some bigger picture or task, and that task is characterized by a long series of communications. With the notifications and the indexing, it becomes possible for you to reorganize things so that everything in that task is collected from you as an individual, regardless of whether it occurs in different conferences and different messages. So one of the ways of trying to overcome information overload is trying to give people the ability to capture the total communication pattern from the perspective that they have as individuals, of the tasks they have to accomplish. You might be taking three or four courses and the way each teacher organizes the individual course, is not the way you organize the way you go about doing your homework... if you can find a paper that works for both courses, you know, that's your privilege.

Overhead #17

TOPICS

INQUIRIES

SHORT ITEMS

ROLE FOR INDEXING

RESPONSES

SELF SELECTED TRACKING

ROLE FOR INDEXING

SIZE CONTROL

TYPES:

FACTS

REFERENCES

LEADS

OPINIONS

BRIEFS

ROLE FOR EDITOR

DATABASE FORMULATIONS

LARGE GROUPS AND BROAD SUBJECTS

Okay, just to give a little flavour, [pointing to the list on the overhead], one could break down almost any one; there are activities that are just appendages, that are passive; you can allow on an item, delta edit, so I can suggest a re-wording of some part of the conference, and it is attached as a delta edit, to Linda's comment when she looks at it; it doesn't destroy what she already has there, it shows her exactly what my substitute is. Marginal notes is just a different version of delta edits. Then there are conditionals, documents, graphics, data sets, documents where you get the table of contents and you can choose what you want, graphics where you may have to download it to your PC before you can view it; spreadsheets, binary files, hypertext, which once again, may be a document that has a little different structure for browsing.

There are some basic issues, by the way, related to hypertext and conferencing; we've tried to do some work on 'what is a taxonomy of knowledge that's appropriate in a conferencing environment in a hypertext structure for knowledge collection'. That's more of an expert system type question related to collaboration. [Referring to the overhead list] Response types where the person has to respond; I can't talk about exams, Roxanne will talk about that; can't talk about assignments; a lot of these are Roxanne's types of attachments. Voting of course, and surveys, there are an awful lot of options, types of voting, that are different, that are more appropriate for different decisions.

It's very logical to consider that if you're dealing with decision processes you want an extensive tool box and voting capabilities. You can even have collecting of rankings that are processed by a multi-dimensional scaling package in the computer, fairly sophisticated voting analysis; the fact that the communications are on the computer provides a whole different level of analysis of what a group voting process is.

Overhead #18

DELPHI CONFERENCE

POLICY ISSUE

ALTERNATIVE SOLUTION

DESIRABILITY VOTE

FEASIBILITY VOTE

PRO ARGUMENT
CON ARGUMENT

IMPORTANCE VOTE

SIGNIFICANCE VOTE

ANONYMITY

AT ANY TIME:

ENTER OR CHANGE VOTE

ENTER ALTERNATIVE OR ARGUMENT

Overhead #19

DELPHI EXAMPLE

TREND ANALYSIS & PLANNING I

PHASE ONE: GROUP INPUTS

PHASE ONE: ANALYSIS

Overhead #20

DELPHI EXAMPLE

TREND ANALYSIS & PLANNING II

PHASE ONE: GROUP INPUTS

RATE TENTATIVE ASSUMPTIONS FOR:

VALIDITY

CONTROLLABLE

CATEGORIZE & RANK ASSUMPTIONS BY:

TRUE, MAYBE, FALSE

CONTROLLABLE, UNCONTROLLABLE

DELPHI EXAMPLE

TREND ANALYSIS & PLANNING III

PHASE ONE: GROUP INPUTS

PHASE TWO: ANALYSIS & RESULTS

Let me try to make the illustration of why you need facilities or tools to structure the communication process. Let me show you one type of Delphi process. I do a lot of work in the Delphi area and I do a lot of consulting in the Delphi area. This is one of hundreds of Delphi designs. This one I've used with a number of consulting operations and companies and it's one example of a specific design that accomplishes a certain end, and usually it's done with a group of experts. You take a set of variables that are important to the company...

Linda Harasim: Do you want to define what a Delphi is, maybe not everyone knows...

Okay, Delphi is the design of communication processes, on paper and pencil, using paper and pencil, so you send a survey to people; they fill it out; you summarize the result, give it back to them with an opportunity to change their mind about anything, and with a new survey, and you keep doing this for two, three, four rounds, sometimes even five. And it is a way, for example, in a company, where you have a lot of planning decisions, where, instead of thirty people trying to get into a room, none of them have the time to do it; they're all over the company; you've got to gather this data together, so a group of people becomes the conferencing system for the wider group of people.

Traditionally it's done in paper and pencil. A lot of the designs still require graphics so that's why you haven't seen too many of them in a computer environment. You have a variable; the variable might be the number of sales of this product that the company made over the past ten years. You send it out to a whole group of people involved in the development, the R & D, and the sales, and say, "Tell us what you think sales will be for the next ten years." The company would like to know that. And you say, "Okay, while you're drawing this line, or curve, or whatever, tell us what assumptions you're making and what uncertainties you have." An uncertainty is something you don't believe will happen, but if it happens, it would change your curve.

So from each of thirty individuals you get back these curve projections and you get about ten or fifteen assumptions and uncertainties. And Linda's assumption is Lucio's uncertainty. So what you do is, and you want to eliminate duplication, so on this one curve, thirty people, half duplication, you get about 150 potential assumptions, 150 potential assumptions, and a human still has to eliminate the duplications and fix up the wording on this. Then you present them all back to the group as potential assumptions and you ask every member of the group to vote on how valid they think it is; is this certain to occur, or is this complete nonsense and it will never occur, or "I don't know", "maybe". And they vote on these things and you then give them back the uncertainty projection which is at the fifty percentile boundaries for projecting a curve, and you give them all these assumptions, now voted on; here are the ones everyone agrees are true, and here are the ones in the middle that no one seems to know if they're true or false, and here are the ones at the bottom that everyone agrees are false.

Now at this point, you say, well, I could have produced a projection curve by a statistical routine, but what I could not have produced is thirty intelligent people, forming a collective model of all the factors that may affect sales of this product in the future; statistical routines won't do that. So actually getting this collective model is far more important than getting the statistical projection.

By the way, this communication process then can go on, but it might go on with different people. You view all the assumptions that are 'maybe' as the things that are affecting the uncertainty of the projection. Then you ask yourself, "Which assumptions can our organization influence; what actions have to be taken to influence those assumptions?" Now that might be with a different category of people, so the information from this first set of people in the organization is then fodder for a new Delphi process with other people in the organization.

So there's an example of a very specific communication structure and there's no reason, even though that one is done quite often today in paper and pencil, why that one can't be done in a computer-mediated communications environment. Clearly, you can't eliminate the human effort, so there is information that is flowing between different groups, and different people are synthesizing it and you have different ways of organizing it. And I guess I want to make a final point. And I guess I'll have to do that a bit fast.

Overhead #22

GROUP PROBLEM SOLVING

PHASES

PROBLEM RECOGNITION

PROBLEM SPACE

ISSUES

STRATEGY FORMULATION

GOALS & OBJECTIVES

POLICIES

GENERATION

CRITERIA

ALTERNATIVES

BENEFITS & RISKS

EVALUATION

ESTIMATIONS

VOTING & MODELS

DISAGREEMENT EXPLORATION

Overhead #23

META PROCESSES

REGULATION

FACILITATION

Then there are all sorts of things they are concerned with: problems, issues, questions, goals, strategies, policies, agendas; you'll find papers on 'how do we set agendas'...

Overhead #24

OBJECTS OF DISCOURSE

PROBLEMS, ISSUES, QUESTIONS

GOALS, OBJECTIVES, PLANS
STRATEGIES, POLICIES, AGENDAS

CONCERNS, CRITERIA, ARGUMENTS
ASSUMPTIONS, VIEWPOINTS
OPINIONS, VALUES, INTERESTS

CONSEQUENCES, SCENARIOS, IMPACTS
TRADEOFFS, COMPROMISE, PROPOSALS

SOLUTIONS, DECISIONS, PROJECTS
TASKS, ALLOCATIONS, POSSIBILITIES

What happens when a face-to-face group gets together? We know that there are different individuals that have different psychological abilities. There are people who have to deal with gory detail; they aren't happy unless we deal with specifics in this discussion. There are other people who can only think of broad goals; we've got to discuss the goals; how can we bring up decisions before we've set what the goals are? You've heard this. Many different people need to approach a problem from a different direction, yet we cram all these people in a face-to-face room and say, "Here's a process, and we're going to go through it; now we're going to discuss goals". And maybe you can't stand it so you're just going to have to sit there and listen to us, and you've got to find something to say even though you don't like to think about the problem in this way. I think this is one of the problems in face-to-face meetings.

Now what happens when you put the same process on the computer?

Overhead #25

ASYNCHRONOUS OPPORTUNITIES

INDEPENDENCE OF:

META PROCESS & SYNCHRONIZATION

BACKTRACKING

CHANGING VIEWS

INDIVIDUAL CONTROL

EQUAL PARTICIPATION

MIXED COGNITIVE STYLES

BOTTOM/UP vs. TOP/DOWN

DATA vs ABSTRACTION

If you design carefully and you have the proper structure, that exposes the types of considerations and the directions from which considerations can be made, every individual can all of a sudden, work asynchronously. The one who wants to concentrate on coming up with specific alternatives can start listing those immediately, and the one who wants to come up with goals can start coming up with those and the computer can sort of put these where they belong and organize them. And the moderator has to play with 'when do we decide we can synchronize this to get a vote on goals', or whatever, so we're not going to eliminate the human in here.

Overhead #26

GROUP DECISION SUPPORT

EXAMPLES

My claim is that there is a potential in this technology, to do things better than we do in face-to-face. Most people say, "Gee, computer conferencing, that's neat. Maybe we can do things almost like we could in face-to-face". No, forget it; there is the potential we can do things better than face-to-face, and that's really the way we should think about the technology and think about how we're designing for groups to be supported with this technology. If we think the other way, we've already cut off our ability.

Overhead #27

ASSUMPTIONS and GOALS of CMC

COLLECTIVE INTELLIGENCE
SUPPORT FOR HUMAN ROLES
TAILORED COMMUNICATION STRUCTURES
INTEGRATION WITH OTHER RESOURCES
SELF TAILORING BY USERS AND GROUPS
CONTENT AS THE ADDRESS
COMMUNICATIONS AS A GENERAL INTERFACE
LIST PROCESSING OF COMMUNICATIONS
DESIGN OF A SOCIAL SYSTEM
ASYNCHRONOUS OPPORTUNITY

Overhead #28

CMC DESIGN CHALLENGES

INTEGRATION
LARGE GROUPS
SELF TAILORING
LARGE POPULATIONS
HUMAN ROLE SUPPORT
EVOLUTION & EXTENSIBILITY
SYNCHRONIZATION OF ACTIVITIES
SIGNALING, CUEING, & TRACKING
INFORMATION OVERLOAD REDUCTION
TRADEOFFS:

And I guess Roxanne and I have always thought that the objective of these systems is collective intelligence. Can the group do better than any individual would have done acting alone? And most experiments on face-to-face ... it hasn't been typical of face-to-face meetings.

Overhead #29

COMPUTER MEDIATED COMMUNICATIONS OBJECTIVES

FACILITATION OF GROUP ACTIVITIES

TAILORING COMMUNICATION STRUCTURES
AND PROTOCOLS AROUND THE
APPLICATION AND THE GROUP

COLLECTIVE INTELLIGENCE

Overhead #30

COLLECTIVE INTELLIGENCE

CAN THE GROUP DO BETTER THAN ANY
INDIVIDUAL MEMBER ACTING ALONG?

NOT TYPICAL OF FACE TO FACE MEETING

EVIDENCE IN CONTROLLED EXPERIMENTS
WITH FACE TO FACE, DELPHI,
NOMINAL GROUP PROCESS, & CMC

Question: I'd like to know...what about the affective area, motivating people who look at a problem, and I think that's kind of interesting that that's the one area where I wonder if it would sort of fall apart, in the affective domain?

In the motivation. Okay. We know from all our experiences that the medium itself will not cause a group that doesn't want to communicate, to communicate. In other words, the motivation always has to be there. Fortunately, in real organizations, working with companies, there's a lot at stake. They have to communicate. So sometimes you don't have to worry about motivation. If they aren't there, someone else may get their resources. There's a lot of motivation in that. That whole thing depends very much on the situation. There's a lot more sensitivity there in the educational process. The motivation there I think, is a lot more key than maybe in an organizational file.

Linda Harasim: From the research that I do at OISE, the students reported that they never got to know their peers to such a degree that they never had a friendlier environment, because the nature of face-to-face is so time limited that people go away and they don't actually get to know the ideas of other people or their feelings, because online, the classroom is on seven days a week; people socialize and begin to move around, and start a group space. [Directed to Roxanne:] Is that your experience? We don't want to pre-empt next week, but ...

Roxanne Hiltz: Some students and some computer conferencing users are more emotional and more personal in this medium than they are in any other; other people do treat it like a business memo and they never let down their formality online. It seems to be some kind of personality construct.

The same as face-to-face, you mean.

Roxanne Hiltz: Yes, with different people.

Typically, there are some people who just blossom in the written word, where they are quiet and shy in face-to-face; and just the reverse: some people who are very good in face-to-face, feel very uncomfortable in the written word. Some of that is a matter of learning, because I think to be an effective communicator you have to learn a new writing style which involves things like paralinguistic cues and so forth. You begin to see among the words, 'ha', and 'smile', and 'I agree'; you can't nod your head; you've got to be very explicit about these things. What startled me over the years is how do people become awfully frank in the written environment...when a student gives you a complete history of a suicide attempt to the class as a whole; I don't expect that in my face-to-face classes and it's not in anyone's computer science course.

Audience member: There is a factor that I discovered too, and that is if you ... at least have some idea of who the online networkers are and have some opportunity to meet them, there may be some reticence in some people but if you're dealing with international communications bulletin boards some of those ... will leave and new ones will come in where they are intimidated with the idea of talking to the entire world on some issue ...

Another questioner: Briefly, what do you mean by hypertext and why doesn't it fit into everything else you've been talking about?

You can view just about every conference system that's ever been designed as a specific instance of a hyper-like system, because it has its own inherent hypertext structure, conferences, comments, links between comments, flags to comments...very explicit. And most hypertext systems today are very specific in that they are designed largely to fit the particular application. Some of them are also starting to build in communications so a lot of hypertext systems are now obviously conference systems. The problem there is a little deeper one, at least in my view.

I do have a paper on it; if you give me a card I can send it to you. What we want in a collaborative system where people are trying to accumulate knowledge, even in an educational environment where students are trying to accumulate knowledge about a certain field, you want an underlying hypertext structure that will allow you to link knowledge about that field in such a way that you can find the material by some very fundamental knowledge structure, because then you can build an expert system maze for the people wading through what is a classic information retrieval problem.

Hypertext is a delight to neophytes because you manipulate a little piece and somewhere in this mass of material you can branch out, but you very quickly can also get lost amongst the tangled web. Hypertext itself, most of the people have never dealt or wrestled with the classic information retrieval problem; they don't even talk to the information retrieval people. My ideas are that the solution is in the structure of hypertext. And one approach one of my Ph.D. students is taking is to go to Guilford's Theory of the Intellect and use the Theory of the Intellect Model as the model of the hypertext structure. That might work in this environment...without starting a lecture...

By the way, if you come back next week, Roxanne doesn't agree with any of the things I've just said. [laughter]

Question: In the beginning you said that there were specific issues, like some for individuals and for also the group. Could you give me an example?

Oh, very easily. A lot of individuals who usually start out complaining in our system in that when you receive a message the individual who sent it gets confirmation of when you received it. You would be surprised how many individuals say that's a violation of their privacy. They don't want other people to know when you've received it. Furthermore, if you choose to zip through it, it means you take it off your waiting cue without displaying the text, they get a different confirmation so the people that received it know that you didn't redisplay the text. It's very easy in a group if people don't know you're ignoring what they're sending you, then people can go and ignore everyone. But if in fact there are indications that people aren't reading what you're telling to them, the group has to take some sort of action, or individuals have to say, "Hey, Joe, why don't you read when I send messages?". "Well, you're too long."

Questioner: You're in teleconferencing?

I'm in computer-mediated communications. I'm saying by how you do things, even like notifications, you can get signals back to members that let people know whether they're using or doing 'what' with the material. The system gives notifications when we copy something to someone else, to the original author. And people said, "That's violating my individual rights to send that information around", well, you're an individual; we're interested in group performance.

(Tape Change) ...it's hard to say. It depends on the group and what their attitude is to what they're doing. We have a lot of indications where you have solved the motivation problem ... you really want to improve what they're doing so that they can do better given this medium than with the other tools that they have.

Question: My second question is would you think that computer-based communications within a group changes the very nature of the group's communication and dynamics?

In one of our early studies we did actual ... Bale's Interaction Process Analysis of the content and we had matched face-to-face and conferencing groups. The content distributions were completely different.

Roxanne Hiltz: And the participation patterns were changed.

Yes. There has not been enough evaluation from a linguistic point of view. I know of no good linguists who ever have gotten into this field.

Questioner: The reason I suggested this bit of business organization, a large one that's national and international... face-to-face meetings often serve totally different purposes than the origin or the reason for the meeting in the first place. The reason for the meeting may be an excuse for all kinds of other reasons for people to get together. I see that communicating through the computer affects that dramatically, the relationships among the employees ...

Let me just mention the benefit of having Ph.D. students who know the whole area of working with this from a linguistic standpoint, of doing a linguistics evaluation of these mediums; it is just wide open; nothing has been done. There's so much in the way of writing style and different nature of contents, that we feel for many years, exists in many application group uses of this medium that have really not been examined from a linguistic standpoint.

Audience member: The meaning of jargon and shorthand and that sort of thing?

Do you want to elaborate, Roxanne? You're the evaluator; I'm the designer.

Roxanne Hiltz: The nature of language, the organization of presentation of ideas, the way in which ideas are linked together and follow on each other or don't. There's been a lot of content analysis of say, T.V. or face-to-face meetings, but no sophisticated linguistically oriented content analysis of computer conferences. A few amateurs have looked at three transcripts and written something; I'm calling them amateurs in the sense that they weren't trained in linguistics; they came across this from another field and saw it. John Carey3 did something about ten years ago on paralinguistic cues in conferencing, where he pointed out that people turned their non-verbal cues into verbal cues: they said, "huh?", "smile", you know, "ha, ha", they made words out. So that was one of the few articles I saw that showed that there was something linguistic going on; that people made written equivalents of non-verbal or paralinguistic cues, and I haven't seen much since then.

Even in our very first experiment, which was done what year... [Roxanne Hiltz:] 1978. ... where we did the Bale's Interaction Analysis, we found something that should have triggered research, but we're doing it in the computer field, so none of the people saw it. It was very significant; we found what seemed to be a basic thing going on, this group problem solving, that hadn't been identified in previous literature. There was a shift in the types of content between face-to-face and the types of content in computer conferencing. And the shift was such that the distributions, when say, joking is higher and so forth, the groups are getting pushed to quicker consensus, but they're getting pushed to poorer quality solutions. When you have more opinions given, you're getting pushed to higher quality solutions, but much pushed away from consensus. But what was happening in the conferencing, you're being pushed to more opinions and less joking, and in face-to-face, more joking and less opinions, so it is much easier to get consensus in face-to-face, but poorer solutions, and much harder to get consensus in the conferencing, or solutions, and this seemed to be a fundamental property of communication as opposed to the fact that we were just looking at this medium. But that work, no one would have gone anywhere with, I think, some potential in the new Mac is starting...

Audience member: Well, didn't you find that the more people are able to be informal, and to trust and to let go, and treat one another as friends, and kid about it, that the more they can do things in the way of shared projects internationally together, because that's been my direct experience. The more I goof about it and pretend to be an absolute idiot, the more people will open up and start to co-work with me, and then I can present really heavy ideas and they love it and they work with me.

Yes. That's a phase thing. There is the importance of the socialization leading to later phases, but what I was talking about is where you are measuring a single, limited time, problem solving exercise in an hour and a half, where everything was together. But certainly we have seen even one of the earlier... socializing before the group exercise definitely correlated with higher satisfaction ...

Audience member: Can you import graphics over online... can you do graphics directly instead of ...

We're sort of at the point of only being able to begin to work with graphics, because within the facility one of your attachments will be a complete binary file. But then everyone using the PC has to have the same graphics package. We can do that with the NJIT students because they all get a PC with certain packages...

Audience member: I have a notification and a question. The notification is that there has been some very recent work by statistically sophisticated people in this area and much of it has been across the Atlantic, ... Leavy and associates from Manchester, George McCarthy and associates from the University of York, and on this side of the Atlantic, in fact on this coast, H.H. Clarke and company at Stanford, have been looking into computer-mediated communications.

I'd much appreciate it if you could get us some publication references or pass them on to Linda.

The question is... I'd like to know what you think of the suggestion by Joe [?McBrock?] in a recent book chapter that you have a variety of communication tools available to you, and that some of these tools might be more appropriate at different phases of the group task. You might want to have or produce a richer medium at the start of the group project and then work towards a more ventilated media such as CMC of the variety we have been discussing this afternoon, later.

I have nothing against multi-media, although for the point of being devil's advocate, I guess my emphasis is to try to see if everything can be done in this medium. I do know in the educational environment, I still prefer to give a certain number of video seminars, because so much material can be delivered by a professor giving a good lecture; as a part of any course, I wouldn't want to eliminate that from a lot of courses; you still need those lectures in a course. Video tape is the cheapest and best way to get them out to the students. Maybe three or four percent of the face-to-face meetings I am currently forced to participate in either at my university or some companies I work with that I wouldn't give up. That means there's probably a 95% that I would give up gladly if I could reach most of those people through a conferencing environment.

Linda Harasim: But within a task, let's say there's a task, that's what you're saying; there are different phases. Isn't there a moment, I've found with my students, there's a moment when they need some synchronous, when they're coming to decisions. Some of the conferencing systems don't have a synchronous environment where we've got a split screen. Ours has it but most of the ones...

I think you have to look... we don't have a good group process model and I didn't make that point, because if you think about all the group process models in the literature, there's evaluation and so forth, they are all derived from studying face-to-face groups. And then these same people turn around and say, "We have these process models we derived from studying face-to-face groups", applying it to study an asynchronous group, where the individuals can work independently. It doesn't make sense. It's like they are violating all the premises in the foundation of how they evolved it. In the asynchronous group we need a model that integrates individual problem solving as an explicit part of the model with group problem solving. And no one has come up with and been able to validate that sort of model yet. I keep sending my students out to think about it and they keep throwing up their hands... if we had that sort of underlying model, we might be able to understand what processes might be better in what medium, when we're trying to compare the conferencing with face-to-face.

The discussion period was closed and Linda Harasim thanked Murray for his presentation.

1. Akoff (spelling may be different) wrote about structured problems, unstructured problems, and problems that are neither structured or unstructured and therefore are referred to as 'messy' problems. For an explicit reference, see Linstone & Turoff, The Delphi Method, Addison-Wesley, 1975.

2. Churchman is the author of The Design of Inquiry Systems and his writings provide the philosophical foundation for Turoff's work. For an explicit reference, see Linstone & Turoff, The Delphi Method, Addison-Wesley, 1975.

3. John Carey: author of Online Communities, in Transactions of Sociology for Computational Linguistics, Ablex Press, Norwood, N.J.

4. The three best sources for the material in this lecture are the papers:

[Turoff, 1991]. Turoff, Murray. Computer-Mediated Communication Requirements for Group Support, Journal of Organizational Computing, 1(1), 85-113, 1991.

[Turoff, et. al., 1993]. Turoff, Murray, S. R. Hiltz, A. N. F. Bahgat, and Ajaz Rana. Distributed Group Support Systems, MIS Quarterly; December 1993, 399-417.

and the new edition of the book:

The Network Nation: Human Communication via Computer, Roxanne Hiltz and Murray Turoff, 1993. MIT Press