Conflict, stability, and a pile of rocks

Continuing a thread, Something’s Happening Here

This comment goes back to a previous thread on social stability with interesting comments by Rick, Bruce, and Brian. A post that Rick sent on February 3 started the thread, and I’ve done a lot of thinking since then about how I could have responded to it. But I write very slowly these days, so the online conversation has moved on.

Nevertheless, the posts in that thread have kept me thinking about some important issues, which are (1) how to define stability, (2) how conflict relates to stability, and (3) whether the concept of a giant virtual controller is helpful for understanding how large-scale stability works. Here are my current thoughts on those three questions.

In his February 9 post in that thread, Rick asked me what my indicator of social stability is, and then he offered several examples of what, in his view, stability isn’t, all of which are examples of conflict.

Rick and I agree about what conflict is from a PCT point of view: an interaction in which two or more controllers try to control the same variable (i.e., the same aspect of a shared environment) but with different references for the state of that variable. And I agree that the changes he lists are an indication of social instability. But I think we have to be careful in thinking about conflict and stability to not assume that one is the cause of the other.

Like conflict, stability has a clear definition in PCT: Stability is what a control system achieves when gets its perception of an environmental variable under control. In ordinary discourse, however, the concept of stability is broader than the PCT definition. It refers to lack of change more generally. Putting together these two ways of thinking about stability, we can say that stability comes in two flavors, broadly speaking: active and passive.

In PCT, we recognize active stability when a control system has some environmental variable under good control. As long as the controller is actively working and has a range of output sufficient to compensate for any disturbances, the controlled variable will show relatively little change. Living things act to stabilize the variables in their environment that are essential to their continued life and reproduction. The indicator of active stability is the relative absence of change in the controlled variable.

Passive stability, in contrast to active stability, occurs when physical objects show relatively little change over time unless some outside force acts upon them. Living things, including people, often make use of passive stability by creating useful objects made from durable materials that are impervious to many disturbances. Birds build nests. Beavers construct dams. Coral polyps secrete calcium carbonate to make the reef structures they live in. Humans create the “built environment” to support their lifestyles.

Some features of the natural world also exhibit passive stability. For instance, the form of a mountain may change very little over the course of centuries, and smaller objects like rocks made of hard metals tend to have a stable configuration, in a human timescale, at least. Whether the stability is active or passive, we regard variables that don’t change or change very slowly as stable.

Notice, however, that stability, whether active or passive, is always temporary. When an active controller stops trying to control an environmental variable, it becomes subject to instability because of the other physical forces in the environment that act upon it.

Due to the random physical forces always at work, passive stability is only temporary, as well . Even mountains gradually erode from the long-term actions of wind and water. Whether stability is active or passive, we can measure the amount of stability by the rate of change. Stable phenomena change slowly and unstable phenomena change more quickly, but stability is never absolute.

Not only is active stability temporary, but it can vary in relation to the “power” of the controller. The degree of stability (the extent to which the variable changes more slowly than it would if it were subject only to other physical forces) depends on the gain of the control loop that is keeping the variable in control.

When the loop gain is low, the control looks loose or lazy, and the actively controlled variable is not especially stable. When loop gain is high, the control looks much tighter, and the controlled variable changes relatively little. But some small amount of change never goes away. If the gain in a control loop is too high, the control becomes unstable again, with output first overshooting and then undershooting in a pattern of oscillations that may either die away or increase catastrophically, depending on how high the gain is.

In short, active control is dynamic. To stay in control of the perception, the controller is always taking action to compensate for the impacts of disturbances on the controlled variable, but the system can never eliminate the effects of disturbances entirely. (Anyone wanting to check these control-system dynamics out for themselves can go to my presentation at the IAPCT conference in 2020, with links to a modeling environment that allows you to play around with control systems by using sliders to adjust their parameters to see what happens.)

This finally gets me back to Rick’s comments about stability and instability. It seems to me, from the way he asked the question about my definition of stability, that he was thinking of stability in absolute terms as the complete absence of change. He seems to have reasoned further that because conflict can lead to change, stability is impossible to achieve except in the absence of conflict.

What I am arguing here is that stability is a variable, not an absolute, and that the degree of stability can be indexed by the rate of change. Furthermore, in my view conflict and stability are not necessarily incompatible. I see lots of empirical situations in human society in which some degree of conflict occurs, but the situation remains relatively stable. And, I want to argue, we have good theoretical reasons to account for the co-occurrence of conflict and stability.

Basically, the concept conflict refers to an interactive relationship between two or more people or, more generally, two or more control systems attempting to control the same variable in their shared physical environment. The degree of conflict varies with the degree of difference in the reference values that the parties bring to their attempts to control the same variable. The concept stability refers to change or lack of it in the physical environment that the controllers share.

Conflict and stability are concepts that refer to different levels of analysis, and there is no a priori reason to think that one should be associated with the other. Empirically, as I will argue below, conflict between large numbers of people may often contribute to the stability of their social arrangements, rather than the reverse.

I agree Rick that conflict has been increasing in American society over recent years, particularly since the beginning of the second Trump administration. And I agree with him that things have become much less stable than they were before Trump’s second inauguration. For sure, the political situation in this country is alarming. But I don’t see the situation (yet) as a total loss of stability, and, in my view, giant virtual controllers are the reason that some measure of stability still remains.

At the end of Rick’s February 9 post, he accuses me of being a sociologist (guilty as charged), which seems to be why, in his view, I have clung to this very confusing (to him) theory of collective control and, in particular, to the concept of a giant virtual controller.

I think you like the idea of a Giant Virtual Controller (GVC) because it seems to legitimize Sociology as a discipline independent of Psychology; the GVC is a social rather than an individual behavioral phenomenon. I think, therefore, that the concept of a GVC is very important to you, as a Sociologist, and, therefore, it will be hard (if not impossible) for you to give it up. But I wish you could give it up because I believe you could do great PCT-based research with all that great modeling skill you’ve developed.
[/quote]

Sociology and psychology are certainly two different disciplines, and Rick is right that they examine human behavioral phenomena at two different levels of analysis: the social and the individual. But even though their subject matters differ, I don’t buy his conclusion that one discipline is legitimate and the other isn’t. (Where did he get that idea? Is psychology the queen of the disciplines? I don’t think so.)

Furthermore, I do not think that sociology as a discipline is independent of psychology. On the contrary, I have worked for 35 years to show how the two disciplines are linked, and I have attempted to construct an approach to sociology firmly based on the facts of human psychology, as shown by Bill Powers.

The problem, as I see it, is that Rick’s view of sociology and of society more generally lacks “sociological imagination,” as sociologists call it. Sociological imagination is the ability to understand individual experiences in the context of larger social patterns. The term was coined by C. Wright Mills in his 1959 book, The Sociological Imagination. In my view, Rick needs to do some reorganization of his perceptions of how larger social patterns work.

Because Rick probably never took a sociology course and has never read much sociology, his perceptions of the discipline of sociology are extremely simplistic. For example, to think about conflict, Rick has often used the images of a tug-of-war or an arm-wrestling contest. These images are fundamentally misleading ways to think about social conflict, because the imagined interactions between people take place in a single location and concern a single physical variable. The large-scale phenomena that sociologists seek to explain take place simultaneously in multiple locations and involve multiple physical variables spread out over large geographic areas.

Let me suggest a different image that might be more helpful in thinking about conflict on a sociological scale. Imagine a big pile of rocks at one end of a large field, like a football field. There are two teams. Let’s call one the Conservatives, and their goal is to keep the pile of rocks exactly where it is. Let’s call the other team the Liberals, and their goal is to move the rocks down the field and spread them out nearer the other end.

The rocks are heavy enough that an individual player can only carry one at a time, and the player can only carry a rock a few meters before having to put it down and take a rest. However, when one player gets tired and puts a rock down, a player from the other team can swoop in to pick it up and carry it in the other direction. The rules specify that players can’t physically attack each other. (No rock throwing!) All the players can do is carry rocks. (But maybe some of them cheat?)

Is there conflict going on between the two teams? Definitely. How stable is the position of the pile of rocks that they are fighting over? Well, as the conflict continues, rocks may gradually get dispersed from their initial position at one end of the field, and some rocks may make it all the way to other end, but, because the teams are working at cross purposes, the overall change in the position of the pile of rocks is likely to be slow. In other words, the center of the distribution of the rocks is likely to remain pretty stable.

In this example, the conflict does not create instability. Conflict, in fact, contributes to stability in the position of the rocks. The more conflict, the less change in their overall position. Thus, conflict promotes stability in the position of the rocks, and rapid change could happen only if the teams cooperated.

Lots of societal phenomena are like this. We see high levels of conflict, but, despite the strenuous efforts of reformers, nothing seems to change. Some people get very frustrated when they don’t see anything changing, while others worry that any change will mean the end of their whole way of life. And anything the first group tries to do is counteracted by the second.

In short, where possibly contentious issues are involved, conflict tends to slow down the rate of social change rather than speeding it up. That’s just the way it works with collective control (lots of players on multiple teams acting independently) and sociological phenomena (like broad social patterns) spread out over time and space.

Of course, the rate of change of the pile of rocks in this example would depend on how many people were on each team, and how many were just sitting on the sidelines to watch the action. Big teams are likely to be more powerful and better able to create rapid change. Social power matters. Organization could matter, as well. Teams that are well organized might be able to set up something like a bucket brigade to move rocks efficiently down the field. Access to resources might also affect the situation. One could imagine that some players in the game might be really rich and able to afford heavy equipment to scoop up and carry off buckets full of rocks.

However, the bottom line in this example is that conflict slows down change and thus increases stability of outcome, as it does in many empirical examples of social phenomena, like changes in racial and gender norms, the switch to cleaner sources of energy, patterns of resource use, patterns of income and wealth distribution, etc.

A final comment: Giant virtual controllers involve widely distributed patterns of active control by individuals acting independently but using somewhat similar references for controlling their perceptions. To the individual encountering a GVC, the active stability produced by the GVC might feel more like passive stability. Anyplace you go, if you try to disturb the customary pattern, someone is likely to resist the changes you want to make. It might seem like “society” has dictated the way things should be. But in effect, the GVC is active control on a collective scale masquerading as passive control–something that just seems to be out of human control, a fact of life.

And a postscript: I don’t have any great investment in the label “giant virtual controller.” This was an idea that Martin Taylor and I dreamed up when we were collaborating on our chapters for the first PCT handbook, and my memory is that Martin came up with the wording. As in the case of our even more unwieldy label, ATENFEL, we deliberately chose to use unfamiliar lingo in talking about these phenomena just to emphasize that we were looking at them from a distinctly PCT point of view.

In PCT, stability, S, is a measure of quality of control. It can be measured as S = 1- Vo/Ve, where Vo is the observed and Ve the expected variance of the CV (expected based on knowledge of how disturbances would affect the CV if there were no control at all). When control is good, Vo << Ve and S approaches 1.0. When there is no control, Vo = Ve and S = 0. So, the closer the stability (S) of a CV is to 1.0, the higher the quality of control.

“Active stability” is a synonym for the phenomenon of control. PCT recognizes the fact of control and measures the quality of the observed controlling with a measure of stability, S.

“Passive stability” is stability achieved without a control system. It’s a measure of variation in a variable that is expected from physical law. It corresponds to Vo in the PCT measure of stability. For example, one measure of “passive stability” is wind produced variation in the lateral position of the roadbed hanging from the cables of a suspension bridge. The bridge is designed to produce the smallest such variation (minimum Vo) given the known wind conditions. The golden gate, which I walked across last week, has been very stable for nearly 100 years. The original Tacoma Narrows bridge was not so lucky.

And in the last 100 years or so people have also made use of “active stability” (i.e. control) to create very useful objects (e.g. thermostats, airplanes).

If by “natural” you mean “physical” or “non-living” then ALL features of the natural world exhibit passive stability. Passive stability is a result of the operation of open-loop physical causality. Active stability is a result of the operation of closed-loop negative feedback control.

Actually, the now uncontrolled variable becomes passively stable – subject to the physical forces in the environment that act upon it just like the effect of wind on an uncontrolled suspension bridge.

The quality of active stability (control) can vary and that variation can be measured by S. But if active stability goes away, as the term “temporary” implies – and it does go away, for example, when we die – then the controller no longer exists and what was the controller just becomes part of the physical world of passive stability.

I think of stability as a variable. How we measure that variable depends on whether we are dealing with a CV or just a V; that is, we must know whether a control system is involved in determining the stability of the variable. If a control system is involved, then we are dealing with active stability, and the measure of stability is proportional to Vo/ Ve. If a control system is NOT involved, then we are dealing with passive stability, and the measure of stability is proportional to Vo.

There can obviously be stability when there is conflict – that’s what the virtual reference level of a CV is all about. Since you know that control systems are involved in producing the stability then the appropriate measure of stability is S = 1-Vo/Ve. The problem is that this measure will give you different results depending on the amplitude of the disturbance. If the disturbance amplitude is greater than the dead zone, then S will be close to 1.0 – it’s looks like the CV is being kept stable; there is good control. But if the disturbance amplitude is smaller than the dead zone, S goes to 0 – now it looks like the CV is not being kept stable; there is no control at all. So the “virtual control system” that seems to be created by conflict exists only when the disturbance to the CV is large relative to the width of the dead zone. When the disturbance varies within the dead zone the virtual control system no longer exists.

A closer look at the situation in terms of PCT shows that the CV is not really under control when there is conflict unless the dead zone is very small. If the dead zone is large, then the observed stability of the CV comes at the cost of large errors in the systems involved. If the dead zone is small, then the systems have nearly the same references for the state of the CV so that the systems are functionally cooperating rather than being in conflict.

I agree. But, as I noted above, how we measure that variable depends on whether a control system is involved (in which case stability is measured by S = 1- Vo/Ve) or not (in which case stability is measured by Vo alone).

Great. How about showing how a control model accounts for this phenomenon?

I would like to know why you have this view. What is the evidence?

I think they are both “legitimate”. I just guessed that you liked the idea of a GVC because it is a uniquely social phenomenon. I think sociology will be even more legitimate without it.

Sociology was my undergraduate minor and one of my favorite classes was one given by Harvey Sacks, who was (and still is, I presume, even though he died young in a car crash) a major figure in the area. The class was on conversation analysis and in retrospect I see that in many ways it foreshadowed my interest in PCT. Like PCT, Sacks’ work was “phenomena first”. He had tons of data on conversations between kids and adults and his theories about what was going on were based on observable consistencies in that data. And his theories were rather PCT-like. As I recall, one of his basic ideas was that kids learn conversational conventions as the means of accomplishing their conversational goals. Sacks himself was quite a character. His lectures involved thinking out loud while he paced back and forth smoking one cigarette after another. He was like a combination of two of my favorite geniuses: Bill Powers and Bob Dylan. I got an A in the class!

I don’t think the tug of war is any more misleading about conflict on a sociological scale than balls rolling down inclined planes was about motion on a solar system scale. And I don’t think your big pile of rocks view is any more illuminating about the role of conflict in social behavior than the biblical idea of a flat, immovable earth enclosed by a solid, transparent dome was about the role of the planets and stars in the universe.

And I don’t have any great investment in some forms of social stability not being the result of conflict. I just haven’t seen any evidence of it. Evidence being data that can be accounted for by a model that includes conflict.