96. "An Engineer-Social Science Team at Work," In John C. MacKinney and Richard A. Scribner (eds.) Institutions for the Application of Science to Social Needs (Washington, D.C.: AAAS, 1973), pp. 51-65. Reprinted in Technology Review, Vol. 77, No. 3 (January, 1975), pp. 27-31; in Research Management, Vol. XIX, No. 1 (June 1975), pp. 18-22.
We commonly recognize the complexity of our numerous societal problems by suggesting that their solution requires interdisciplinary teams in which are combined the experience and knowledge of many experts in many fields to face all the main elements of a problem. And we often lament, as James Carrol quoted by Harold Orlans, that the "present compartmentalization of knowledge . . . in the form of academic disciplines is not coincident with social need. However useful this compartmentalization may be for the inner development of knowledge, it is only marginally useful for the resolution of social questions."
But we also often lament that interdisciplinary teams have a hard time working together; as Mr. Orlans writes, "Universities have gad treat difficulty in conducting successful interdisciplinary research.... [And] while research institutes have been more (they could not have been less) successful than universities in organizing interdisciplinary teams, the frequency with which such teams are utilized and the extent of their success is commonly exaggerated." For example, Bruce Smith has reported that interdisciplinary research at Rand Corp., "far from becoming steadily easier and more effective, has actually grown more difficult in recent years." One indication of these difficulties is that such teams are rare, while mono-disciplinary work is common.
I would like to discuss one instance of effective team work which I believe casts some light on the opportunities as well as the issues involved. It concerns a research project on telecommunications being worked on by both social scientists and engineers. The study has not resulted in any sensational breakthroughs; however, the problems of working together were effectively reduced and the results are "products" that seem superior to those which a mono-disciplinary team could have produced.
The Setting-a Multidisciplinary Center
The work reported has been conducted at the Center for Policy Research, a nonprofit research corporation in New York City. The Center's staff includes sociologists, psychologists, economists, engineers, systems analysts, a lawyer, and workers trained in other disciplines. As the research staff has no more than fifty persons, they know each other relatively well. There are no mono-disciplinary departments, chairmen, deans, or reward systems. The Center carries out projects in most domestic policy areas financed chiefly by grants of government agencies, foundations, and private industry, in that order. It does no classified research and considers society as its ultimate client.
Attempts to keep all the 50 members of the staff in dialogue with each other in Center-wide meetings (a desire frequently voiced by staff members) resulted, without exception, in boring, mechanical affairs, which the majority of the staff soon vetoed by staying away. Intensive dialogues take place among individuals whose offices are adjacent or who work in the same problem area in teams of two to nine; or such dialogues occur in "marathons" (one lasted eight hours throughout a Sunday), in which 25 to 35 staff members participate-a research team plus others interested in the subject area. The rest of the Center-wide activities, aside from occasional "business" meetings, are parties which bring out the staff, spouses, and friends. All this had led me to suggest that the well established sociological law which expects that "formal groups" (in which relations are "cooler" and more work-oriented) will be larger than "informal" groups (which are "warmer" and contain friends and peers) needs to be reversed: it might be most effective for research centers, at least, to have small, problem-oriented teams or formal groups and large, informal ones.
Developing Participatory Technology
One major project of the Center is to develop an electronic system which will allow people dispersed in space to conduct a dialogue with each other in groups and to register their views as a group, in the same way a group is able to do when meeting face to face perhaps in a town half. Technology is sought to increase authentic participation, create a democratizing effect, and reduce alienation. The project, named Minerva after the Roman goddess of wisdom, is supported by RANN (Research Applied to National Needs), the new division of N.S.F., the National Science Foundation. Dr. Stephen Unger and I are co-principal investigators; Dr. Unger is Professor of Electrical Engineering and Computer Science at Columbia University, where I am a professor in the Department of Sociology. The Minerva research team includes four social scientists and four engineers. (The study referred to is supported by N.S.F. grant No. GI 29940)
The work on conference calls provides illustrations of the teamwork discussed below. The initial requirements for these calls, in terms of social science, are to provide for a flow of dialogue and for tallying of responses. Here are some or the questions we have had to face: How will requests "for the floor' be transmitted? How will votes-be tallied? How can the "tower of babel" condition (many parties talking at once) be avoided? How can participants be provided with a sense-of-the-group"--how others feel (previous social science studies show that without such cues, people may be reluctant to take stands)?
All these "needs" are served through the use of nonverbal cues when groups meet in face-to-face town hall meetings; for example, a participant raises his hand to request the floor or to vote; he stares (to quiet the interloper), he scans faces as well as nods (io assess the group's reactions), etc.
At first the social scientists of the Minerva team sought mechanisms that would be as close to the natural group processes as possible. If members of dispersed groups could see each other, for example, say by using picturephones, they might be able to work together as people do in face-to-face groups.
The engineers contemplated various devices to respond to this notion of the social scientists: let the speaker's face be displayed on the conference circuit for all. to see (but we all realized that this would not meet any of the suggested needs except the communication of the speaker's non-verbal cues, such as facial expressions, movements of the hands for emphasis); use a split screen showing all the participants' faces (that might do nicely for the social needs identified above, but we would need a giant screen for a group of 30, or even 12, if the faces of all the participants were to be recognizable); have a scanning device which would display only the face of one participant at a time but which would scan all of them; or have half the screen devoted to the speaker, the other half to such scanning (this suggestion seemed to bring the design closer to what the social scientists thought was necessary).
At the same time we began to wonder whether or not we were correct in our basic approach. We were trying, in effect, to reproduce in another medium all the same effects available in the first one; perhaps instead we should seek electronic equivalents to the social-psychological processes-electronic steps which would answer the same needs but not be identical or even similar in form. Thus when a person wishes "the floor" in a face-to-face group discussion, he cues the chairman, or all present, by some non-verbal cue such as raising his hand. In a picture phone this could be approximated; but why not give each participant a little button with which he could turn on a small light (or flash a number), which would serve the same purpose as raising his hand? Similarly, such cues could be used to assess the group's sentiment about what is being said and where the discussion is moving without any scanning of faces. At this point I suggested equipping each telephone in our laboratory with a cue box that would allow participants to request the floor and to register approval or dissent. Later an integrating unit for tallying the responses was added and a gavel for the chairperson is planned.
At this point, Dr. Unger raised a major consideration which was not alien to the social scientists but which they could not really worry about, let alone solve: if Minerva was to be used on a large scale, the cue-box idea would require modifying a tremendous number of telephones. This would take years, maybe decades, and great expense; it suggested a task which could hardly be welcomed by overburdened telephone companies. On the other hand, if only some telephones were modified, the use of the system would be limited because many groups who would wish to meet on conference circuits would have some members with assisted and some with unassisted telephones.
Two Rules from Technology
Professor Unger suggested that we should try to follow two rules: see how far we can go without modifying the "normal" telephone network; and try to suggest changes; if necessary, whose implementation would be concentrated at telephone exchanges rather than at home terminals. After additional deliberation it was suggested that "flashing" (opening and closing the circuit to signal the operator) in dial telephones and, even better, pushing the buttons of touch-tone telephones could be used for the suggested purposes. The engineering members of the team are now trying to work out the circuits necessary to use existing telephones in this way by modifying telephone exchanges to give them the capacity to "read" and transmit the cues. The team has only been working on this mission for a short while; therefore, it is not yet clear if this approach will pay off. I
Experiments--Problems and Successes
While our discussions were going on, experiments were being run in our laboratories.
The first experimental conference call groups met without any technical assistance. When small (nine members) homogeneous groups were chosen and were given a chairman, and when a "midline" task (not very easy, not very difficult) was assigned, the groups worked quite well and effectively--they communicated easily and reached decisions--without technical assistance. However, when the task was made more difficult the general complaints about the unassisted system rose. Richard Remp who conducts the experiments, reports that the percentage of participants who indicated that they were "always" able to follow the discussion fell from to 48 percent on the easier task to 34 percent on the more demanding task. Similarly, the percentage of participants who indicated that they could get the floor "very easily" fell from 20 percent to 13 percent. While 17 percent of the participants characterized the discussion on the easier task is "very effective" in exploring the topic, only 6 percent of the participants on the more demanding task chose that description of the discussion. It was of particular interest that a fair number of the participants in both tasks spontaneously mentioned their need for a means of more clearly identifying the speaker.
Meanwhile, to test the basic notion that electronic equivalents could better serve our aims, the engineers followed my suggestion by developing a cue box, which is now being used in experimental conference calls in the telecommunication laboratory of the Center. The box might be said to simulate the conditions which would exist if touch-tone phones could be used for cueing. The cue box also provides a model of the technical assist that may be used if efforts to do without special assists fail, and its development has confirmed our main thesis-the value of a close, continuous dialogue across disciplines.
The first cue box was designed to provide a button, a switch, and two lights per participant. The button is to control one light which lets the chairman know that a participant wishes the floor without the request being carried verbally, which would interrupt the dialogue. The switch signals assent if pulled up and dissent if pulled down; the second light is used to register both feelings: short rapid flashes indicate approval (like applause); long flashes--dissent (like a boo). Each participant's cue box panel also has two lights for each other participant; thus each participant can see what all other participants feel and if any has requested the floor.
At this point, the social scientists started to wonder about the social consequences of a technology which actually accelerated the natural process it merely sought to reproduce. In a real-life group one rarely if ever knows at all times exactly where each member "is" with reference to the issue under discussion; with our cue box every member receives every other member's signals constantly. We hypothesized that greater individual and aggregate visibility might inhibit participants who are clearly in a minority from expressing their views lead to decisions before people have had a chance to fully air their ambivalences, or reduce the chances for give and take (individuals, once they have identified with a position, may be reluctant to shift).
The engineers suggested two technical adaptations to test for these postulated undesirable influences (do they really exist?) and to partly eliminate their effects. Other devices developed by the team go beyond that. But one thing at a time.
The "request-the-floor" button, at first, was to be fixed, that is, once you push it your request is registered, and to undo it, you have to push it again. Now the engineers suggested to wire the "assent-or-dissent" signal so that it erases if you stop pushing it. In this way technology encourages" people not to freeze their position. (Who wants to hold a switch on "dissent" for an hour, unless he really is very strongly opposed?) Of course, when the signal is used for voting purposes and responses are to be tallied, a "freeze" procedure is used, but this occurs only at a stage when the group--or at least the chairman--is ready to vote, which is one form of registering views for which the cue box is incontestably useful.
Second, the engineers gave the chairman a central switch by which he or she may cancel all visibility (except each participants own signal). Thus the chairman can eliminate each participant's feedback from others, withholding individual and aggregate visibility for whatever period is desired either by the chairman or the group. This device can be used, for example, to be sure that discussion is uninhibited during the first half of the meeting or the first meeting of a series, or until the group votes that it has had enough dialogue and is ready for a first "straw" vote.
Testing of the cue boxes revealed different behavior pattern, which approximates that of a natural group, when the number of participants increased. While it was easy for members of small groups (of, say, seven members) to recall what each person had said via their signalling lights, when the group grows to say 25 to 30, each member can see only group patterns (lots of dissent lights, lots of assents, many switches, etc.); he is not able to identify a position with each member of the group. He could keep track of how X or Y felt, if these were viewed as particularly important persons, but he could not keep track of everyone.
One day an engineer brought to the laboratory a device which translates electronic cues into optical bars. Seeing it flash led me to realize that it could serve for a more radical alteration: the individual lights could be replaced by a summary indicator giving a group visibility somewhat like the figures on an election-night television screen. However, our experience suggests that while this approach has certain virtues, it also prevents individual and subgroup recognition and may project a more monolithic indication than the situation warrants. For example, if many members of a group are undecided, 10 out of 30 shifting their views from "yes" to "no" and 10 from "no" to "yes," the summary indicator would continue to read "20 favor," giving no indication of indecisiveness to the participants.
What about automating the chairman? If indeed large numbers of groups are to be in dialogue, including people who have not met before and who may be assigned to a group by a computer (on the basis of their indication of the kind of people they want to talk with), there will be difficulty in electing or otherwise choosing a chairman.
Professor Unger suggested one day that a mini-computer at the telephone exchange could replace the chairman by receiving the "I-wish-the-floor-next" cues and assigning the floor in the order of the requests. The airlines are now using such a device when lines are overloaded to tell callers that all agents are busy and that callers will be assigned an agent when their turn comes. This could be adapted here. Moreover, Professor Unger pointed out, the computer could inform callers of how much time they would be allotted, the amount of time depending on the size of the backlog stacked up.
Well, I wondered, such an automated chairman sounded quite insensitive. A real-life chairman often does not allot the floor mechanically, on a first-come, first-served basis. A good chairman takes many things into account: whether or not the person had the floor before; if he tends to be representative; if be tends to be verbose and repetitive; if his side has just been heard three times ("let someone speak against it'"); whether or not others can be drawn into conversation (e.g., people who are shy); etc.
Fine, replied Professor Unger, we could give the automated chairman most of these guidelines. For example, no one would be chosen twice before every person who sought it had the floor once; the floor would be assigned alternately to those who cue assent and dissent, or whatever.
I was skeptical because I remembered that a chairman often follows these rules implicitly and intuitively; to make them explicit or visible may cause many strains and conflicts. Professor Unger then suggested that one need not announce the guidelines followed by the chairman either, although there might be some social merit in doing so--or at least in being able to do so. We decided to try an open and an opaque automated chairman.
What I seek to highlight in reporting this process is how the interaction between the engineers and social scientists identified several technological options which clearly differ in their social consequences. Without close give-and-take the engineers might well have settled for one of the options, having had no reason to identify others, and the social scientists might have felt there was no way to deal with the side-effects of the one available. To put it simply, we are all richer for the dialogue we have shared.
Finding the "Dialogue Barrier"
We already know from these experiments and from those conducted by a New York enterprise called Tele-Session that groups of 30 can conduct an effective dialogue on conference circuits. The question we ask, as we seek to approximate town hall meetings, is what is the upper limit; where is the "dialogue barrier"? Reference here is to the number of people "hooked" to the same channel, who all can hear each other and can effectively participate as if they were in one hall, where the audience reaction--restlessness, cheering, booing, careful attention--affects spectators and is part of the process in which the group makes up its mind.
This dialogue barrier becomes crucial when we plan for the larger groups for which two-way cable television is to be used. One strategy for studying dialogue barriers would be to equip two-way CATV participants with a return audio capacity; that is, a microphone could be placed in each home so participants could speak to each other about the issues being discussed on CATV. But the size of the dialogue barrier will be affected by the "communication-manners" of the participants; the better they are, the more people can participate. Could we do something to hinder any one participant from disturbing the whole meeting--a kind of electronic equivalent to staring down disruptors in a public meeting?
An engineer suggested that to discourage these obstructors we could build a system in which people could hear, see, and receive cue feedback; but their microphones would be activated only when they are given the floor by the chairman. We quickly came to call this the "harsh gavel."
The social scientists suggested that we need a microphone to pick up audible reactions from each participant--such as grunts of approval or disapproval; but no participant should be able to obstruct the meeting by shouting or tapping his mike. At the same time, it would be quite appropriate for 50 or so participants to be able to make the kind of blocking noise they would make in a town hall.
The engineers met this suggestion with the idea of a meek gavel;" each person would have a weakly amplifying microphone, and his microphone would be turned up when he had the floor. (We are now looking for such a microphone; while the concept is clear, we do not know if the equipment is available.)
The Minerva project is still in full swing. Among the topics engineers and social scientists are now grappling with are alternative methods for tallying responses, arranging conference circuits, switching CATV viewers, activating television screens, and formulating rules of access to the new technology. But these topics are still being evolved; while it is premature to decide which technology is best, or best under what conditions, it is clear that the inter-disciplinary team is working hard to find the optimal solution. This discussion may suffice to illustrate the close interaction between the two disciplines.
The Forces at Work
We believe that cross-disciplinary work is appropriate at the Center for Policy Research for several reasons.
Some of these lie in the organizational arrangement and the participants' orientations. The Center is unlike a university department or professional school; it is a cross-disciplinary unit, and these tend to be regarded as "second class citizens" in terms of budgets, rewards, and status. The Center for Policy Research is wholly interdisciplinary and considers such work a virtue--indeed, essential for handling our societal problems.
Second, the Center and the Minerva team are more committed to serving the society than to enriching basic research knowledge. They are quite happy to add to the coffers of our collective knowledge; but they care most strongly about being of value to the "real" world. This attitude makes the group problem- rather than discipline-oriented; in turn, this brings members of the different disciplines together.
Closely related is our focus on a concrete problem rather than on abstractions, models, or measurements. People working with applied or policy problems within each discipline are closer to workers from other disciplines than to the theoreticians or measurement builders in their own fields. Perhaps even more important is continuous, frequent, intimate interaction. There is almost daily give-and-take within a fairly small group; without this, the disciplinary blinders are difficult to overcome, and team members tend to talk by each other rather than to each other. We find that each member of the group must become aware of the implicit assumptions which each makes about the others' disciplines (many engineers tend to assume that their colleagues are rational men, that social scientists are non-rational ones). Basic concepts have to be shared; thus as I learned to talk about frequency-division vs. switched cable systems, Professor Unger began referring to "primary" groups.
The importance of such teamwork lies in the fact that most problems we face today have social and natural science elements. If the two elements are dealt with separately, the implicit hope that someone will later be able to synthesize the mono-disciplinary findings is often frustrated. The development of true cross-disciplinary work may not be an essential prerequisite for defense and space missions; but it seems to us essential--and possible--for most domestic missions.
"Minerva: An Electronic Town Hall" by Amitai Etzioni in Policy Sciences, December, 1972.
"The Efficacy of Electronic Group Meetings" by Richard Remp forthcoming in Policy Sciences.
"The Secret Power of Telephone Conference Calls" by Amitai Etzioni in Telecommunications, October, 1973.
"Two-Way Communication: A Design Analysis of a Home Terminal" by Noam Lemelshtrich, forthcoming by Sage Publications.
"Economic and Legal Foundations of Cable Television" by Leonard Ross, forthcoming by Sage Publications.
"Policy Research" by Amitai Etzioni in The American Sociologist, June, 1971.
Amitai Etzioni was born in Germany and studied at Hebrew University (Jerusalem) and the University of California; he has been at Columbia since 1958. His research and writings span a wide range of subjects related to societal organization and social change, including issues through which science and technology affect human institutions.
Technology Review, January, 1975.