Diagram2

                   Within the context of Garden Management we can visualize three main actors: the Garden, the Gardener and the Observer. Singular is used for convenience
although each term stands for " the Set of n members".   We propose to include two of the three namely Garden and Gardener within the System and to leave the Observer in the Surroundings. In this way our System's main objectives will be stated in terms of the
interaction with the Observer or, the raison d'etre of the system is the Observer or, in very plain language: no Observer, no System.( the analogies with script -actor-public are self evident)
                    In Diagram 1 below the grey frame delimits the system so that Garden and Gardener are included as sub-systems which interact with each other; the left-pointing arrow stands for the actions or operations of the Gardener on the Garden and the right-pointing one for the feedback from the latter to the former. The down-pointing arrow
stands for the actions of the System on the Observer. In this first approximation model,
there is no feed-back from the Observer (Alas! this is very often the case ... no hand clapping or booz for the gardeners)

        We will now consider briefly Churchman's five points for thinking about the meaning of the system. The five points, as enumerated in the Introduction, were:

                                                  Objectives and performance measures.
                                           Environment and constraints
                                           Resources
                                           Components
                                           Management

1.- Objectives and performance measures:
           Regarding the main system's objective or main function we could tentatively propose following statement:

               The system's main function is to please the Observer.

     Tentatively because, within the Hard system approach, if we don't manage to find quantitative performance measures for the main function, we'll have to go back to square one and replace the main function or objective for a more amenable one.
Accordingly, the keywords to please will have to be analyzed in depth, firstly in search of an unambigous definition and secondly through the conceptual chain:

    garden unit--> visual perception--> sensation--> emotion.

    This sequence will be explored later on in this work since it is of paramount importance regarding the effect on the Observer. However it is obvious from the start
that in the present state of knowledge we cannot quantify sensation or even less emotion Since to please is at those levels, the main objective stated tentatively above is not suitable in the frame of systems analysis. The maximum we can aim now in the sequence is for quantification at the level of visual perception. Accordingly, we propose to state the systems objectives at a lower level, in terms of visual signals and visual noise,
more precisely in terms of the ratio of signal to noise. This is a common procedure in
systems analysis, the alternative one is called a PROXY Objective and is chosen because it is measurable and can be quantitatively discussed. See PCP/objective.
So, again tentatively, we propose as the system's main function:

               The main function of the system is to maximize the overall Signal to Noise Ratio, within the environment's constraints.

               The Signal to Noise Ratio (SNR) as applied to gardening is discussed in detail in the Section Communication Models

                                                        I.- THE SYSTEM

          Provisionally we will define system as " a unit of organization which has some functional integrity". By functional integrity we mean that, once the main function of
the system is stated, its various parts or components play a part in the accomplishment of such a function. For more about systems see   Principia Cybernetica/system.
           The contents of the subystems Garden and Gardener are discussed in the Sections Resources and Components.
            The system illustrated in Diagram 1 may be considered as a Control System
including two subsystems: a controlling one, the Gardener and a controlled one, the Garden. (the reader will be wiser if he goes on to read the node on Control at PCP/ctrl
because in the foll. I am relying heavily, almost subserviently, on its general concepts)
The interaction bet. the two susbsystems is not symmetrical in the sense that the action
of Gardener on Garden is not of the same kind as the reverse one. The Gardener may
change the state of the (controlled) Garden while the reverse action is the formation of
a perception of the Garden on the Gardener. The controlled system is unable to change the state of the controlling one (the case of a gardener driven mad by the obstinacy of a garden that
refuses to be controlled is not yet included in the theory).
               As illustrated on Diagram 2, below, the subsystem Gardener may be thought to
be made of two parts a representation and an agent. A given state of the subsystem Garden is represented through perceptions (later on to be identified with our Signal to
Noise ratios) and this representation is sent to the agent part as a flow of information.
The agent, who is responsible for his actions, acts upon the Garden according to this info
flow. At a given level of abstraction we could think of two compartments in the Gardener's brain: one receives the perceptions form the Garden, forms a representation of the state and sends the information as an statement to the other compartment, which then issues a command for an action to be performed.( See further Control in Terms of Statements and Commands at PCP/ctrlsc.
          The following example may serve to illustrate this discussion: a lawn area (controlled subsystem) has a green yellowish color (visual perception), the paired
yellowing/water deficit(representation) forms a message that is transmitted to the gardener(agent) who issues a command to irrigate the lawn(action). Let's complicate the example a bit and assume that the controlled subsystem is a set of n lawn areas, each of them producing a different perception, from them a complex representation is formed at
the controlling subsystem; assume further that because of internal or external constraints
the agent cannot act on the n lawns simultaneously. In this case the agent has to design a program or plan of actions to deal with the situation but, even if no inmediate action is taken, we still have a control system as long as the program is carried according to schedule. The implication of this being that the simple model of Diagram 2 can be adapted to situations involving a number of subsystems, a complex information flow
and a multifuncional agent capable of designing and executing programmed actions
(any computer software being included into agent).

A control system closer to an actual situation for gardening may be one in which not all the observed variables of the controlled subsystem can be affected by the
agent, at least during a given time period. This is illustrated in Fig. 3 of PCP/control, also
there the notion of goal is included. We can think of the goal or target in this context
as an ideal representation that can be approached by performing certain actions. In our case the main objectve, comprising a number of goals and targets is discussed in 1.- below. The agent compares the ideal with the actual representation and takes actions designed to minimize the difference between them, in other words he steers so as to keep
the state for the system as close as possible to the desired one.
An alert reader( I mean one who has not fallen asleep yet) may have realized that by introducing a goal we are getting close to control by steering in the sense used
by R. Wiener and since we already have a feedback loop (garden-->gardener) we may propose that our system can be treated according to cybernetics. (which was my reason to get involved in the above lengthy discussion).