OrgIntelligence in the Control Room

A control room provides a thin but panoramic view of everything that is going on. The French sociologist Bruno Latour calls this an oligopticon: he describes how Paris is controlled by a collection of separate control rooms, each focused on a different slice of reality; these control rooms may only talk to each other at the margins, or in major emergencies; there is no supreme control room commanding all the others.

"Water, electricity, telephony, traffic, meteorology, geography, town planning: all have their oligopticon, a huge control panel in a closed control room. From there very little can be seen at any one time, but everything appears with great precision owing to a dual network of signs, coming and going, rising and descending, watching over Parisian life night and day. No single control panel or synoptic board brings all these flows together in a single place at any one time." [Invisible Paris, pdf]

Each control room monitors and directs a particular set of systems, and has some responsibility for the smooth, efficient and safe operation of these systems. Except in a fully automated plant, such as a nuclear power plant, the responsibility may be shared with skilled operators and supervisors in the field, such as inspectors and engineers, bus and train drivers, policemen, etc., who not only pass situation reports to the control room (thus acting as the eyes and ears of the control room), but also may have a fair amount of autonomy and initiative to solve local problems, perhaps supported by up-to-date information from the control room or elsewhere. So we may regard the control room as the hub of a larger distributed control system, involving operational people as well as the control room staff.

My interest here is in the collective intelligence of these control systems. As the operational environment becomes more complex and demanding, collective intelligence becomes more and more critical in ensuring smooth, efficient and safe operation. Collective intelligence depends not just on the individual capabilities of the people, but on how the work is organized and how well the various technologies (information systems, screens, dashboards, communication devices) are designed and integrated to support the work. (In other words, we're talking about sociotechnical intelligence - intelligent collaborations of people and technology.)

Organizational intelligence has six constituents, so there are six areas we need to consider.

• Information gathering - what signals and messages are fed into the control room, and are these sufficient to enable critical situations to be quickly recognized or even anticipated?

• Sense-making - how well are complex incidents interpreted, and the possible knock-on effects predicted?

• Decision-making - how well are resources allocated, problems prioritized and solved, operational policies suspended or adjusted?

• Memory - how well are past situations and problems referenced in solving today's problems and anticipating tomorrow's problems?

• Learning - how do we continually improve the performance of the operating environment, as well as improving the effectiveness of the control system?

• Communication - how well do we communicate internally (within the control room), outwards (to people in the field), sideways (to other control rooms) and upwards (to management or other governance bodies)?

A control room or control system may have opportunities to improve in some or all of these areas, and the leverage yielded by such improvements can be very considerable. For example, with a fixed level of intelligence in a traffic control system, it may be impossible to increase traffic volumes without compromising safety. But if we can increase the effective intelligence of the control system, it may be possible to increase traffic volumes: for example, instead of having a standard minimum distance between vehicles, or time between signals, it may be possible to implement a variable rule that is more complicated, more difficult to enforce, but yielding more efficient utilization of resources. The point is that variable rules only work if you have enough sociotechnical intelligence in the control system to manage them properly; the more intelligence you can build into the system, the more variation (and therefore fine-grained and dynamic optimization) the system will be able to cope with.

A control room typically operates on at least three different tempi (speeds).

1. There is a real-time or near-real-time tempo, in which an event triggers an automatic or pre-programmed response, almost like a reflex. These responses are designed according to some pre-established operational model that allows the designer to reason about causes and effects, and should be monitored to make sure that these reflex mechanisms are working.

2. There will be a continuous stream of incidents requiring human intervention. The people in the control room will have to verify what exactly has happened, and then take appropriate action, based on their training and expertise, past experience, as well as practical common sense. The elapsed decision time may be measured in minutes or hours, and the situation as a whole may take days to clear.

3. Then there is a much longer-term learning cycle, where people are constantly looking for more effective ways of controlling the system and improving its performance. This might include analysing patterns of activity and identifying weak signals that would give early warning of possible future incidents, analysing system behaviour to check if the desired outcomes are being consistently met, exploring alternative ways of exercising control, experimenting with design improvements to the technical systems, and so on. The learning cycle may also include occasional crisis management exercises based around a simulated incident, to test the responses to a major emergency. In a rapidly evolving world, this kind of continuous improvement is a vital aspect of collective intelligence, to make sure that the control system maintains its ability to fulfil its responsibilities.

The relationship between 2. and 3. is an interesting one. Sometimes the people responsible for 3. don't actually sit in the control room, and may even report into a different part of the management hierarchy. But although we certainly cannot ignore the formal management structure, the real question here is about the effectiveness of feedback and learning, and in providing as many people as possible with the opportunity to contribute to the learning process, and therefore to the intelligence of the whole system.

Lots of interesting issues here then, both in terms of organizational change and technological change, with the possibility of producing large improvements at relatively small cost.

---

For more on organizational intelligence, please visit the Organizational Intelligence website.

Organizational Intelligence and Gender

A new study by a team of researchers (three male, two female) indicates a statistical correlation between the collective intelligence of a team and the proportion of women. They attribute this correlation to a factor they call "social sensitivity", which they describe in terms of the ability of group members to perceive each other's emotions. They acknowledge that this is not an exclusively female ability, but suggest that females tend to possess this ability to a greater extent than males.

Social sensitivity is one element of emotional intelligence. It would be interesting to know how other elements of emotional intelligence (e.g. confidence, determination, self-awareness, self-control) affected the performance of these groups on these tasks. See my earlier post on Emotional Intelligence.


By the way, we probably shouldn't think of "social sensitivity" as something that belongs to an individual woman in isolation - obviously it would be pretty meaningless and impossible to observe outside some social context - we could instead think of it as a group phenomenon that happens to emerges more readily in the presence of many women.


Another relevant factor appears to be the conversational dynamic of the group - taking equal turns rather than allowing one person to dominate.

I suspect that the relationship between conversational dynamic and gender balance is a complex one - men may respond in various ways to the presence of women and vice versa - and of course this relationship depends on cultural context. And by culture I don't just mean macroculture (Americans versus Japanese, Generation X versus Generation Y) but also microculture (the style and identity of this particular organization). This relationship is not elaborated in the extracts I have seen.

What doesn't appear to be a relevant factor, at least for the set of group tasks included in the experiment, is the average or maximum intelligence of the group members.

This confirms something I have long asserted, that collective intelligence is not determined by individual intelligence, but emerges from the interactions of the group or organization.


While these findings are undoubtedly interesting, it is important to qualify them with the observation that these were fairly short-term tasks from groups that were apparently put together for the purposes of the experiment. So it would be useful to have some research that looked at the performance of mixed gender groups over longer periods. If women are better at reading emotions quickly, is this advantage eroded over time as team members become more familiar with one another, or does this represent a persistent source of advantage?

If we accept the idea that the presence of females in the boardroom would make the board of directors more intelligent, this raises some further interesting questions. Instead of talking abstractly about glass ceilings, we need to understand specifically what power structures allow male-dominated organizations to survive against the potentially superior intelligence of female-dominated organizations. Alternatively, we might have to explore the idea that the intelligence at board level is a lot less relevant to corporate success than the pay and self-importance of senior management might suggest.

Another way of interpreting the results is to say that gender represents an important mode of diversity. In this particular experiment, gender may have been the most significant mode of diversity, especially if all the participants were drawn from a relatively homogeneous population, and were of similar age and educational background. In an experiment with a more heterogeneous population, other kinds of diversity might turn out to be more significant, which would lead to the explanation that it is diversity in general rather than gender in particular that produces these outcomes.  However, this research happens to provide data about gender, and I have not seen any data about the comparative value of different kinds of diversity.

For example, Hillary Clinton and David Miliband have much in common, and would be grouped together in many classification schemes.

---

Anita Williams Woolley, Christopher F. Chabris, Alexander Pentland, Nada Hashmi, Thomas W. Malone Evidence for a Collective Intelligence Factor in the Performance of Human Groups Science DOI: 10.1126/science.1193147

Collective Intelligence: Number of Women in Group Linked to Effectiveness in Solving Difficult Problems ScienceDaily (Sep. 30, 2010)

Study finds small groups demonstrate distinctive 'collective intelligence' when facing difficult tasks PhysOrg (Sep. 30, 2010)

---

Places are still available for my Organizational Intelligence Workshop on December 8th.