I think that’s a weird comment. Because a highly competent PCT researcher does not specify something well known to just about every competent person in the PCT community, you simply assume, or perhaps assert, that he makes a stupid assumption that doesn’t conform to the usual PCT expectation.
The researcher (Kent) didn’t make any stupid assumptions. He built a model that demonstrates how a variable that is being controlled relative to different reference levels by two or more control systems can end up being kept in a virtual reference state, protected from the effects of disturbance, so that the variable appears to be controlled by the collective – a virtual controlled variable.
The modeling is done extremely well. My only complaint about it is that the behavior of the model has not been compared to that of a collective of real people. Doing that might lead to some surprising discrepancies between model and data, and maybe not. But I can’t evaluate the usefulness of the model as an explanation of social phenomena until I see how well it handles real data.
As it sits, Kent’s is a perfectly competent model. It is also clearly consistent with the “free market” concept of how people should be able to operate in an economy. If there are higher level systems operating that are trying to set references cooperatively, this should be shown explicitly in the model. But as it sits, there is no cooperation in the model; there is only conflict. Maybe conflict is what the higher level systems in all the individuals in the collective want but, again, if that is so, it should be shown in the model. But I remember Kent saying in some of the early publications on this that his aim was to show that stability can emerge from conflict (which, I believe, is a tenet of free market economics as well) and his model demonstrates this admirably. Now the question, for me anyway, is whether or not this is actually the case in real life.
Rick, Let’s leave it there. You don’t seem to want to learn about collective control, but if you ever decide you do, I suggested an easy lead-in for you. Your complaints always fail to deal with my responses to your earlier complaints, which makes for difficult debate, and I really can’t be bothered unless you show some interest in learning about the many forms of collective control, even at the most basic level. I did try to help, but was rebuffed, so I don’t see any point in continuing.
What is the context in which you are envisioning the emergence of a negative feedback loop? Some possible contexts are
- The first emergence in evolution, as Bill sketched in Powers 1995: The origins of purpose: the first metasystem transitions
- The first emergence in muticellular ‘colonies’ of e.g. slime molds
- The first emergence in the DNA-structured development of an individual organism (cytogenesis, embryology, …)
Or are you saying that it is the same in any context?
In the 1995 paper, Bill says
When a negative feedback system is set up so as to resist disturbances of a controlled variable, it tends strongly to maintain that variable near the zero-disturbance state, the “natural” state of that variable.
This seems to define the ‘natural’ state as the zero-disturbance state. That ‘natural’ state is a particular value of the CV affecting the relevant input function of the molecule. If zero disturbance is what makes it a ‘natural’ state, then we already have that ‘natural state’ value as a reference value, the appeal to nature is empty.
Bill avoids falling into a circular argument by recourse to a rather vague definition of ‘natural state’ as “its lowest-energy or otherwise natural configuration”. But this is from the point of view of a chemist or physicist. That analytical viewpoint is admissable only under transformation to the subjective viewpoint of the control system. From the point of view of the molecule, in order to have a reference value absent a higher-level loop setting it, some kind of memory is necessary. Given the stipulation (in this imagined proto-scenario) that there are disturbances which the molecule resists, we know that the ambient value of the [environmental input to the] CV is not static, it varies. If as observers we stipulate that the ‘natural’ state is that which obtains most of the time, then we have a statistical basis for establishing the reference value as the value established in memory by the frequency of its occurrence.
Bill then goes on to propose that
In a population of molecules" there would emerge "variations in the effective reference signal […] not at the natural zero-disturbance level but at a variety of levels distributed around that point.
He does not say how this variation arises. Here’s a plausible way. The ambient level varies in different parts of the environment, so individuals in different places establish different reference values in memory. Put the other way around, it is implausible that variation in the ambient level would be synchronized throughout the environment populated by these molecules. Synchronization, either static or varying, would be exceptional and would require special explanation (natural cycle, e.g. diurnal or lunar, or being a CV of another control system).
From the point of view of the control system, the reference value is subjectively zero, the value from which any departure is either plus or minus some amount. Of the ‘variety of levels’ that the molecules have established, each as its subjective ‘natural’ zero-point reference level, “some would result in better replication than others. The result: the appearance and propagation of non-zero reference signals as intrinsic parts of the control systems.”
Now hold on. Prior to an optimum value for replication, there is an optimal value for survival long enough to replicate. We are talking about
a kind of molecule that is not only stable, but superstable—its interactions with its environment create enzymes that repair damage to the molecule. This is active rather than passive stability—stability that results from counteracting the destructive forces of nature.
It is here that we must look for the first intrinsic variables, and, just as in complex organisms, they are controlled in the somatic branch (or its equivalent). A reference value for some environmental input is established only insofar as some values of that input disturb the structural integrity and functioning of the molecule. The deleterious values thereby become ‘nonzero’ values for those individual molecules (but the degree of departure from zero may vary across the population). Those members of the somewhat diverse population which can e.g. produce an appropriate enzyme in appropriate quantity to neutralize the deleterious effects of those values are consequently more able to survive long enough for their capacity to replicate actually to perform.
In all subsequent evolutionary developments the same principles apply, except that there is an accumulation of inherited intrinsic variables, with perhaps some falling away, though it is conceivable that those primitive molecules and their chemical defenses are still integral to our cellular and metabolic composition.
In sum, emergence of control requires inputs representing aspects of the structural integrity and functioning of the control system itself. These ‘intrinsic variables’ are the variables that must be held in their ‘natural’ state or “zero-disturbance state”. Inputs from the environment are controlled as secondary means of controlling those intrinsic variables, and thereby surviving long enough to replicate. The ‘natural states’ of environment variables never determines the control system’s reference values for them.
Bill is inchoately describing a bootstrap process that first involves setting references to the statistical norm as the predominant determiner of the value in memory (or the ‘natural state’ somehow defined and sensed). Only those survive with deviations from that value that make survival more likely. His explanation of the deviations is as thin as his explanation of the norm, but this bootstrap concept avoids requiring the molecule to sense its own structural integrity and functioning.
Bruce, I would like to know what you think of my take on the same problem (PPC Chapters II.1 and II.2 ). I look at it from a rather different viewpoint, taking replication as secondary and being concerned with the origin of control in basic chemical processes.
Martin, you actually introduced the term:
The term is not inherently ‘bad’, but the potential for confusion is. Of course social interactions are structured, but not in the same way that control-system models are structured. In the early '90s (maybe later) Bill pushed back on discussion of social structure, saying in effect show me the input and output functions, the reference input and comparator.
Bill agreed with you that the neural signals in the model are theoretical entities and that as soon as we look into the biological foundations of complex living control systems we find that what we abstract as a signal is a result of collective control by sets of cells. Likewise, when we look at a level of complexity ‘above’ the individual autonomous control systems we abstract to what Kent picturesquely described as ‘gossamer threads’ passing through aspects of the environment. These theoretical entities in the model of collective control are not signals in a superordinate control system*. For each participating CS I think of them as lines of attention, intention, and influence.
You’ve indicated small steps toward understanding the emergent properties of autonomous systems interacting collectively, varying their number and vectors of influence on a single variable. I suspect that more frequently ‘in the wild’ a single variable is collectively controlled as a consequence of CSs controlling perceptions of more than one aspect of the environment, not all the same.
The Matthew Effect is a family of behavioral examples.
- Or if they were, we would be unaware of the signals as such, just as an individual neuron does not perceive its rate of firing as such. Necessarily so, as what is perceived may come under control, which would set up conflict which, if generalized, would destroy the system of which the individuals are constituents. When I posted that observation on CSGnet, Bill said “I feel like my floor and ceiling disappeared” or words to that effect, and there was no further discussion.
Ok, my memory is obviously faulty, besides which my collective control structure idea is not of the structure of an individual collective controller with its virtual reference, perception, and action components, but of interactions among multiple controllers (collective or unitary) of different variables by different (perhaps overlapping) collectives. Those are the structures I think of in the case of collective control.
In PPC III.1.7 I offered a “Short Taxonomy of Collective Control” written some years ago. FWIW, here is what I wrote then. There should also be a general partition between continuous and stochastic, and between whether the collective control is associated in each case primarily with Roles or with individuals.
We thus have at least three types of Collective Control in which all the members act on the same CCEV as a means of controlling their own perceptions.
1.1. Conflicted Control: The participants have independent reference values for perceptions whose CEVs are closely related to the CCEV. The CCEV remains as if it corresponds to a controlled perception, but the outputs of the individual controllers tend to increase as in any conflict. Several people push on a rock, all wanting it in a different place.
- Collaborative Control: The participants control a higher level set of perceptions of belonging and being seen to belong to “the group”, bringing toward a common value their references for their perceptions of the CEVs that combine to form the CCEV, eliminating the conflict while maintaining strong control. Several people push on a rock trying to move it to a place on which they agree.
1.3. Coordinated Control: All members who are controlling for being perceived and perceiving themselves as belonging to the group accept reference values provided by an agreed leader. Several people push on a rock trying to get it to a place chosen by the leader.
In addition, there are at least three forms of Collective Control in which the participants act on different aspects of the environment in order to achieve a common higher-level purpose — a reference value for a higher-level CCEV — that all have in common, rather than all trying to influence the common CCEV in the same way. We will consider some of them in more detail later.
1.4. Guided Control: A plan, with or without a specific planner, determines who does what (I’ll hold the pole if you hammer it into the ground; I’ll get the supplies if you guys get the the tents put up.) The similarity should be clear between this form of collective control and a two-level hierarchy.
1.5. Giant Real Control Unit: Different people or groups of people use protocols in ways that mean that some play the roles of the different units of a control unit (Sensors, Perceptual Function, Reference Function, Comparator, Output function, Effectors), so that the whole social structure acts as a controller. This concept is elaborated in Chapter IV.1, Chapter IV.2, and Chapter IV.3, especially Section IV.2.1.
1 6. Hierarchy of Collective Control Units: Same as 5, with different levels of controller interacting as they do in the Powers hierarchy of control units within an organism.