I glanced at another committee member and gave what I hoped was an eloquent shrug of my shoulders. As far as I could tell, none of his opinions were reflected in the chart either. He beckoned me with a slight nod of his head, and as I leaned towards him he whispered, "Whoever is the last to touch a view graph holds the power." Belatedly, I realized that the chairman of our committee had finally exercised the power of his position.

During the meetings of our committee the chairman had seemingly abdicated responsibility. He sat quietly and inert as the rest of us foundered leaderlessly. We would take turns grasping the chalk and trying to organize the rudderless conversation, always with a sideways glance at the appointed chairman. But as conversation would ensue, he would remain silent. The only noise from his corner of the table was the clacking of the keys on his laptop computer. What was he writing, we wondered? Was he taking notes? If so, which points was he recording -- in what form -- and which were being ignored? A feeling of helplessness engulfed us and stifled meaningful exchange. Kafka would have approved, I'm sure.

In the best tradition, the chairperson of a committee boards points as they are raised, so you have some idea of where you stand. Of course, this isn't as easy as it sounds. You make some brilliant point, but the chairperson makes no move to write anything down. Then some idiot makes a ridiculous statement that is totally without merit, and the chairperson writes it down as a major conclusion. You try again -- louder, of course, for effect, and faster so as to get in more words before his attention wavers. There is no resultant motion of his writing hand. As I said, at least you know where you stand.

Thinking like an engineer, I often contemplate analogies between the problem of moderating committee discussion and determining protocols for communication. In a committee meeting there is one common channel -- the air -- that must be shared among many users. Issues of admission and flow control, contention, buffering, fairness, and efficiency naturally arise. As the number of participants increases beyond about six, the single conversational channel begins to look like the Harbor Freeway at rush hour. There is no logical way to multiplex the limited channel resource fairly among the potential users.

The traditional communication engineering approach is to have a central authority responsible for scheduling channel usage. Some human leaders use the same strategy, often hiding pompously behind Robert's Rules of Order. You make reservations to talk via a separate control channel; that is, you raise your hand and the chairman puts you on a list to speak. "Let's see," he says, "I have Al, Pat, John, Karen, Bill, and George in that order."

In my humble opinion (IMHO -- the most common expression on computer nets), this is a terrible way to run a meeting. The next speaker, Al, makes a point that you know is wrong. You raise your hand and the chairman schedules you to speak at the end of the queue after George. But by the time George has gotten the floor and made his point, nobody cares anymore about your rebuttal to Al's long forgotten point. However, since you now have the floor, you use the time to make some gratuitous comment about something else. Everyone seems to be speaking in non-sequiturs, and the whole system only seems to generate churn.

At the other extreme are the laissez faire anarchies codified in the protocols for local area networks like ethernet. No chairman in required; if you want to speak, you wait for a pause in the conversation and simply leap in. If someone else also leaps in, you both stop speaking for random intervals. This works fine for small groups, but gets tricky for larger ones, where the pauses in conversation begin to be measured in microseconds. All of your senses need to be tuned to the detection of a hesitation in the torrent of conversation.

With a still larger meeting this protocol tends to become unstable. Now there are no discernible pauses in conversation. Realizing that you will never actually capture the floor by waiting for a pause, you adapt the strategy of interrupting the current speaker whenever you anticipate a pause. (Local area network electronics should be so smart!) The aim is to be the first person to interrupt without being outright discourteous. In some larger groups the constraint implied by courtesy is removed.

Sometimes the chairperson will adopt a token passing protocol. "Let me go around the table and get everyone's opinion," he says. While this approach ensures a certain level of fairness, it also decreases efficiency by forcing a lot of people to talk who have nothing to say. The fairness is not perfect either, since if you sit at the wrong end of the table, all of your points will be used before it is your turn. Naturally, you will speak anyway, dithering somewhere between repetition and irrelevance.

These are all mechanical protocols, requiring no intelligent control by the chairman. Greater efficiency requires active, intelligent intervention. "The chair rules that your argument has no merit, Al, and that further discussion in this direction will not be necessary. Now who else would like to talk?" Of course, this seems like deft chairmanship if your name is George. If your name is Al, you have a different view of your chairman.

Electronic systems could be improved greatly by similar strategies. You would be typing away on your terminal, and suddenly the screen would stop registering your keystrokes. A little message would say, "Further information of this sort will not be accepted by the network -- please use better material in the future."

Most known communications protocols have been shown to be unfair in some way. It is, alas, the same with human leaders. I think that the problem of optimizing the use of conversation time is basically unsolvable, and that we are talking about choosing among suboptimum, even disagreeable, alternatives. There are flavors of chairmanship for every taste. Some are too dictatorial for me, some are too passive. And what pleases me may not be to your taste at all. You may, for example, prefer a chairman who types quietly in a corner, summarizing conclusions that he has crafted from his own imagination. However, I doubt it.

Robert W. Lucky
rlucky@bellcore.com