According to my copy (2nd edition, MIT Press, Cambridge, Mass, 1961), Norbert Wiener tells us of "Cybernetics, or control and communication in the animal and the machine". So control is central to cybernetics; yet strangely, Wiener does not list it in his index. (Maybe it does not appear in Wiener's index because it seemed obvious at the time).
I am not sure how well the notion of control is understood: is it really so obvious as to need no examination? How do we come to need control, and what do we need that we may have it?
I shall start from definition - not because I have great faith in either defining or lexicography, but because definition provides a good starting point.
My Shorter Oxford Dictionary gives:
Control (substantive) 1590
1. The fact of controlling, or af checking and directing action; domination,
command, sway
2. Restraint, check (1594)
3. A method or means of restraint; a check (1752)
4. A person who acts as a check; a controller (1786)
These definitions are imbued with the idea of directionality: the words domination and restraint are perhaps the mose telling, suggesting a power that permits the issuing of orders that are unlikely to be disobeyed. Cotrol is of one (possibly more physically powerful) thing by another, to the (not necessarily exclusive) benefit of that other: dictators control populations; managers companies; computers suites of machinery in automated factories; switches heating systems. There is often a sense of amplification, of the more powerful acting for (under the control of) the less powerful. And, always, the controller controls with the intention of benefitting.
The central notion of control has been extended - as happens in any field - so that its meaning may no longer be transparent. Thus, in second order cybernetics, we are familiar with the circularity (bi-directionality) of control, an understanding that may be complete anathema to the engineer concerned with designing a computer system for a factory. (I have a friend, an engineer turned systems scientist, who will not accept even the possivility that the computer system controlling the machinery in the automated factory may be seen, itself, as being controlled in turn by the machinery).
Since the notion of control has been extended while remaining central to Cybernetics, it merits examination.
So why do we need control?
In the first instance, an answer implies two types of response.
Firstly, we need control because the "world" (in the form of a particular system or Object - of which more in a future issue) is not to be trusted.
This may be because the world is not describable or predictable (therefore, not modellable) with either theoretical or experiential exactitude: either the implication of Goedel's Theorem, that we cannot, ever, make perfect models, or Wittgenstein's (for example) observation that just because something has always appeared to be the case does not mean it must go on being so, holds.
Or it may be because the world, itself, is taken not to be quite as regular as we have believed the (mechanistic) classical physics of Newton and Descartes indicates: as manifested (eg) in the indeterminacy principle, catastrophe and chaos theories.
Thus, there may ve (although we cannot necessarily say this with any great cetainty) an irregularity that results from either the world being a less perfect place than we would believe it to be (being full of randomness, perturbations and surprises), or from our inability to make adequate (complete, consistent and repeatable) descriptions of this presumed-to-be-regular world: and we have to allow for such irregularities if we are to be able to control (yes!) our actions so that they may have an effect of the sort we intend, so we can, in turn, gain some sort of stability and, hence, predictability - for stability need not be limited to the static, but can be dynamic as, for instance, with living systems1.
Secondly, our need for control is based in our intention. When we set up a "control-situation" (in the classical sense) we are trying to get something else to do what we want it to do (regardless of whether it can function thus in itself). This highlights several salient points.
There is a goal: a desired state or condition for the system which determines that it be treated as goal-oriented and which motivates its performance and response to the situation it is in, while satisfying the controller's intentions.
Notice that this goalis exterior to the controlled system itself, and satisfies the purpose of the controlling system: the aim set by the controlling system, which chooses it for the controlled system. The controlled system has no part in this choice. Thus, "exterior" is exterior in both the senses that it is neither within the system and nor within the system's "jurisdiction".
The success of the controlled system in achieving the aims of the controlling system as indicated by the goal set2 is also assessed by the controlling system (and is therefore assessed externally to the controlled system). This is achieved through an act of comparison: the state desired is compared to the state of being (the state achieved) and the difference computed. This difference is aplied in a corrective manner to the controlled system so as to achieve a change in behaviour and censequent state, with the purpose of achieving a change such that the state of being more closely approaches the desired state, that is, the goal. We know this process as feedback, and it is one of the great gifts of Cybernetics (in the older sense of a dowry).
So, in effect, the controlling system determines everything for and about the controlled system as far as what we may think of as the control interaction is concerned.
Thus we come to need control. And behind this need, behind the assumptions or pre-suppositions that are involved - such as stability, state, behaviour, goal etc - lies the second great gift of Cybernetics: error. That is, the ability of some system or Object to behave in a manner transcending the predicted (the regular) and deviating from (its progress towards) the aim.
Previous to Cybernetics, error had been taken as an unfortunate flaw, to be excused away, and was generally assumed to be caused by an essential failure deriving from limitations in our abilities and our tools and caused by anomalous behaviour/inadequacy in the model/instrumental (observational) failure/a lack of research. This is, essentially, a theological notion that can be tied in with that of perfection being the remit of God. Turkish (and other Islamic) carpet weavers, for instance, reflect his ethic when they intentionally wear errors into the carpets. To aim for perfection would be impertinent to Allah3.
Cybernetics, in contrast, accepts error as a basic fact of life: it is not incidental to, nor an embarrassing flaw in the subject and in ourselves, nor an indication of our unGodliness: rather, it is at the heart of the subject, at the centre of the motivation of the subject's matter. Cybernetics is remarkable (unique, at least until recently?) in that it not only accepts, it welcomes error, taling error to be a desirable centre in its subject matter rather than merely excusing it away. In effect, cybernetics claims that error is, for whatever reason, in the "nature of things", and must be accepted as integral in our understanding.
And it is true that, with Wiener, we talk of control and the need for control just because he was concerned with, amongst other things, servo mechanisms and other forms of amplification of the controller's power in the controlled, and also because of delay, imprecision, time and the difficulty of calculating with infallible certainty for the future. Error, the failure of the world to ve as we like to insist it is, the inaccuracy (and resulting discomfort) of prediction, our inability to act correctly is central to our experience and is central to cybernetics as it attempts to help us create and maintain some sort of stability in our interactions. Our knowledge is that, while things continue to proceed as we would like, they will not continue forever: the one quasi-fact we know of our machines (including our scientific descriptions) is that, in the end, they are not perfect: the fail. Thus, error always lurks. Thus control (to counteract error when it occurs) is an essential tool to help us live at one with the worlds we make and inhabit.
Control is taken as central in Cybernetics. But there are many pre-suppositions behind it, necessary for it to take its central place, some of which have been introduced. We have considered the distinction between controller and controlled, aim/goal/purpose/intention, feed-back, error; model and description, stability and externality; and Goedel and Wittgenstein, Newton and Descartes, mechanism and classical. Later, we will return to these. But, having located control within Cybernetics, I shall, in the next issue, look at control in contrast to governance and regulation; to control and cause; and to what causes the controller to behave as it does. And to ask about control AND communication.
Notes
1. Forgive me for this sort of description. It is a shorthand. I do not believe that what I describe here is quite "what is". The world that I inhabit is, in my view, deeply rooted in my descriptions. It does not exist apart from me and from this describing, at least in any way. I can usefully know about or inhabit. But I cannot always extend this full argument, or I would get nowhere. At a later date I will make this argument. For the moment, I use the short-cut of a more conventional (more realist) presentation, knowing it does not really represent what I wish to say, but knowing it allows me to try to make a point relatively tersely.
I shall presume that this short-cut is to be taken as understood and is acceptable under the circumstances, and I shall assume not to have to excuse myself too often for using this convenience!
2. The goal does not, of course, have to be static. It can be moving. The notion static is, anyhow, similar to that of goal in the static is (normally) determined in reference to some external fixed point (itself recursively static!). Or it is interal and, therefore, an aspect of a self-referrential system. There are several such notions in cybernetics: what it means to be static will also be examined later.
3. It should be remembered that Newton's great purpose was to glorify God and show the immaculateness of His Order and the frailty of the human except through discovering the Truth of God's own Laws. Newton was an Anglican (Episcopalian) priest.
Return to the content of this issue
Return to the Cybernetics and Human Knowing Homepage