In reverse order, I would disagree with that idea too, if anybody had said it. I think what you’re referring to is the observed reference value of an observed collectively controlled variable. To talk of its ‘stability’ is merely a paraphrase of saying that it is controlled. But because the continuous control of it cannot be attributed to the control actions of any one individual, but rather is the net result of control by many individuals at idiosyncratic individual reference levels, the observations (a variable maintained at a reference level despite disturbances) have an appearance as though a single control system were controlling it at the observed level. Even in an apparently simple dyadic interaction, e.g. at the point of sale/purchase of some good, it is prohibitively complex to include in a model every individual with a control loop that is closed through aspects of the transaction, the price of the good, the value (‘buying power’) of the money, the tamper-resistant packaging, and most immediately the expectations as to what comes next after each move in the transaction.
Begin with an observed phenomenon, a controlled variable C and its reference state. The phenomenon requires explanation. The familiar explanation is by a model of a control system. But further observe that the variable C is not maintained at its reference value by any single control system. Observe that its state is affected by the control outputs of many control systems in a population of control systems.
Another perspective on the problem: There are perceptions (environmental phenomena) which it is difficult or impossible for an individual to change. Moving a mountain, or controlling a perception that the sun rises in the east. Among configuration perceptions there are mountains that no one can reach out and move because of limitations of their output capacity and characteristics of the environmental feedback path (laws of physics).
There are perceptions C which it is difficult for an individual to change because many individuals are controlling C at once. One individual disturbing its manifest level or value cannot overcome resistance by the many who are also controlling it, the individual has only slight effect. They need not be controlling C at the same value for their separate outputs to amount to a net resistance to the disturbance. These are observable phenomena which require explanation. It is not hard to find examples. It is impossible to do anything coherent concerning the (proposed) Principle or System Concept levels without acknowledging that those perceptual variables and their reference levels are social in nature and social in origin. Anything that came to be in your environment by being manufactured, sold, and purchased or given has been overtly subject to collective control, and collective control of it may become evident at any time, as e.g. in what you do when your keyboard fails fully to serve as your means of controlling other perceptions.
Why is this not a source of continuous conflict and reorganization? Well, some folks are rankled by a great many collectively controlled variables. In the extreme, they are rankled at any social standards, expectations, rules, laws, conventions, protocols for polite or civil interaction, etc. But most of us adapt to them in the same way we go around mountains rather than trying to move them.
Collectively controlled perceptions often (maybe usually) are controlled so as to function as means of controlling other perceptions. The reference value for C is “I can use C as means to control x”." A disturbance to the collective reference level for C affects the individual only when its utility is degraded, that is, when disturbance to C indirectly disturbs the individual’s control of other perceptions that more immediately matter. For a particular individual at a particular time the reference value for the bus schedule is does it get me to the dentist on time. If the granularity of the schedule gets me there either late or 55 minutes early that is a disturbance to controlling efficient use one’s time, tolerable by bringing a book or a laptop or scheduling in some shopping, etc. If the bus driver’s control of being on schedule is disturbed by traffic slowed by a fender bender the 55 minute lag is a good thing, tolerance built into that loop. Or if you’re late it affects your participation in the collective control of the schedule of the dentist’s office. A phone call might enable them to accommodate your changed participation in the collective control of their schedule. If the driver’s control of being on schedule is lackadaisical, one or more of the regular riders might make a phone call or write a letter.
In the Crowd demo, the agents do have in common a perception and a reference value for it, but they do not control the state of that perception (the ‘attractive’ individual). They control their individual relations of proximity to it, simultaneously with their proximity to one another, perceptions which in a relative and somewhat abstract sense they have in common, but these are not collectively controlled perceptions, they are individually controlled with collective side effects. There is a collective phenomenon as a byproduct of individual control, but there is no collective control. Not the same at all. The observed social phenomenon (rings and arcs) is not perceived as such. But as I pointed out it would be possible to add to the model one or more individuals which perceive the rings and arcs as such and employ them as means of controlling other perceptions, e.g. in the observed situation in a park a speaker with contrary views might step up on a bench and control to draw individuals from that ring to this ring, or a vendor might push their ice cream cart conveniently near and ring a bell.
You have been saying “I won’t believe such phenomena exist until you show me data”. How to obtain data for such phenomena Is one problem. Not just any data, but data relevant and appropriate for building a model. So the prior problem is how to model concurrent control by a population. The problem then is how to deal with the complexity in an orderly way, working with generalizations over aggregates while ensuring that the generalizations are in principle reducible to control by the individual agents in the population. For discussion and design, the generalizations have names. Because PCT has not previously dealt with these problems, these names are new to PCT.
You say that Martin tried to add to and deduct from PCT.
As additions, you object to concepts that Kent and Martin developed to grapple with the enormous complexity of social phenomena in which a given variable has an observed reference level which is influenced by tens, hundreds, thousands, or millions of people controlling that variable whenever its departure from the observed reference value interferes with their (diverse) control of other perceptions. Of course this must all be reducible to control by individuals, in principle. But to begin verifying how it happens it is necessary first to recognize that it happens, and to have a coherent and consistent way of talking about it.
What ‘deductions’ did he propose? I don’t think there are any.
Don’t get too huffy. Everything that has been tested is also not necessarily true. Theories are provisional, never final. It is not possible to test a model of a given phenomenon of collective control until you have a model of collective control to test.
Again, not quite. Martin investigated the properties of control systems, viewed as systems rather than from within the system. He was trained in engineering physics, then operations research (which concerns the functioning of systems), and then experimental psychology, followed by a long research career in assuredly non-S/R psychology. Beginning as an undergraduate and throughout his career he was most interested in communication interactions among autonomous agents, and he developed a theoretical apparatus for this (layered protocol theory, LPT) which Bill agreed is a specialization or subset within PCT. Looking at PCT analytically, he applied mathematical and conceptual tools of physics, such as thermodynamics, which have been exhaustively tested and demonstrated to apply to all physical systems, control systems included. This analytical point of view is not your interest.
This is what he said he was doing in his last work, Powers of Perceptual control. He said it
builds on the roots Powers planted and nurtured for so many years, developing in directions he often suggested but seldom explored as well as in some directions of which he did not approve. It speculates and offers tentative implications of PCT in many different domains of ordinarily ‘siloed’ research. It offers an organic structure with branches that will certainly not be as strong as the rooted trunk provided by Bill Powers. Nevertheless, I hope that its rambling branches may perhaps show some of the power of Bill’s vision.
A more specific and ‘concrete’ example
Sometimes PCT investigations suggest that commonly believed ‘good things’ may not be so good at all. One example is the commonly held belief that national governments should strive to attain and maintain a balanced budget. A PCT analysis (following the lead of an IEEE conference paper by Samuel Bagno (Bagno 1955), suggests that if an economy is not to stagnate and run down, with increasingly tall and narrow islands of extreme wealth among oceans of poverty, governments should instead aim for an average annual deficit of perhaps 2% to 3% of GDP with a similar rate of inflation. A PCT analysis of why these ‘islands of wealth’ matter for social stability is developed in Chapter IV.7.
The increasing tide of autocratic populism in the developed world may perhaps be largely attributed to the fact that Bagno’s analysis has been totally ignored by those economists to whom politicians listen. Economic advisors to government leaders continue to claim, I think falsely, that balanced budgets, or even government surpluses, are targets to be aimed at on average. Both Bagno’s and the PCT analysis argue that such economists are dangerously wrong. The very survival of Democracy worldwide may depend on politicians ceasing to listen to them. We might, perhaps, relate the political rise of Donald Trump to the much approved budgetary surpluses produced by his two-decade-earlier predecessor as US President, Bill Clinton.
I have no illusions that I might persuade you of any of this. But none of us are obligated to persuade you. You are not the gatekeeper of PCT. I think that you will not ever read Martin’s book. How could I possibly object to that? However, I do object to misrepresenting what he said and did. Martin is safely beyond any felt need to respond. I hope that Kent will continue his careful exposition here. I hope that he and others will take up the challenges of modeling the ways in which humans have learned to live together humanely. The need is great.