Resources are allocated between agents by actions, appropriating, consuming, giving, trading, sacrificing, extracting, etc. These actions can be the result of conscious decisions, habits, customs, or involuntary responses (breathing is seldom a conscious decision, but not breathing may be). Habits may be the results of decisions made a long time ago, and customs the results of long term societal decision making processes. To the extent that some of these actions are the result of our decisions, and in theory controllable, we need to know the outcomes of our actions and what other factors are relevant.
Accounting systems can aid our decision making by providing information relevant to the decision and to the decision makers (Arnold & Hope 1990, Drury 1990). In this context they can aid us in allocating environmental resources. But the crux of accounting is its provision of relevant information. We must understand the process of decision making in order to design an effective accounting system for this problem. Accounting systems also provide a check for the validity of that information through the process of auditing and accountability (Gray et. al 1996).
The decision making process can be looked at as a group of people (decision makers, agents, actors) trying to decide on the correct actions to achieve goals using information about the results of past actions. The whole process can get very convoluted by meta-decisions: deciding on a goal for the decision process, deciding on the form of decision process, deciding on the form of the decision process for deciding on a goal for the decision process, etc. Two particular areas make decision making necessary, conflicts of interest between multiple actors, and uncertainty as to the results of actions.
If society decides upon a goal, be it environment or economic, another problem rears its head: what actions should we take to achieve that goal? We can prescribe an action but how can we tell if that action will produce the result that we want. How can we distinguish between an action that has achieved the goal, and an action that may have more chance of achieving that goal but was unlucky. The issue of uncertainty is perhaps the most important in decision making. This further complicated by the fact that different types of uncertainty require different methods of response.
Uncertainty can be loosely split into two sections, external and internal uncertainty. External uncertainty is where all uncertainty is generated by sources external to the actions of an agent. With internal uncertainty, agents cannot take an external perspective to the uncertainty, because their own actions are a source of uncertainty (McGlade 1993). There is likely to be shading between the two sections, as the size of the agent's effects on a system increases, the more internal the uncertainty becomes.
In this set of definitions, the result of tossing an unbiased dice or dealing cards involves risk. All the outcomes of throwing a dice are known (1,2,3,4,5,6) and the probabilities of those outcomes would also be known (all turn up, on average, one sixth of the time). However, if you were presented with the opportunity to engage in a friendly game of cards with some people you meet on a train this becomes 'uncertainty'. We know that a deck of cards contains 52 cards, but (from useful information provided Hollywood movies) we would have to deal with the possibility that the shuffling and dealing process was not entirely random. This would make the probabilities of cards turning up unknowable.
In this definition chaos means that you don't even know the possible state of nature. Chaos in this sense is intractable, but it is at this point where the usefulness of this set of definitions falls down. We may not know all possible out comes or their relative frequencies, but we do know some of the possible outcomes and we do know something about their past frequency of occurrence. How far we take this information is another thing. We can treat it as being the actual outcome and probabilities while taking account of the limitations of the data (the Bayesian approach) Or we can treat it as intractable chaos, or something in between.
The variability of the outcomes as they are encountered gives us another piece of information. If every outcome is different from last then its not very helpful, but if a percentage are similar to previous outcomes it is some help. Of course we have to deal with uncertainty in variability of the outcomes as well, but this is all interesting (i.e. useful) structure to the data.
Figure 2.1, How a decision affects the result of an action.
The current state of the environment is a consequence of all past actions upon it, as well as by its own internal processes. We can view the some of the internal processes of the environment as being external to our actions. Uncertainty in these cases can be considered as external uncertainty. Figure 2.1 shows how internal uncertainty can be generated by the results of our actions affecting the environment in which the result are produced.
The fact that an action changes its local environment means that the action is potentially contextual. This is because the causative action and needs to interact with the environment in order to change it. The degree to which the result is dependent on the either the action or its environment allows a large variation in potential outcomes.
The usual nature of decision making in some respects amplifies this. Essentially, decision making is not the rational homo economicus model, who assesses every action all the time and selects the best, but more crisis management. Keep all actions the same until something drastic happens, panic, explore lots of options and change things until the problem goes away for a while. Of course these are both generalisations, but in the crisis management model, actions are likely to be applied over a long time. The cumulative effects of the action its environment can change its result, (so spawning the crisis). Decision making is invoked to change actions because we are constantly making actions, and the results of some of these actions change over time.
The link between a causative action and its environmental result is not always amenable to total elucidation. Even if the process is entirely deterministic, the result can be Chaotic (in its dynamic sense). Chaotic action however results in levels of certainty based on our ability to map the attractor. Chaotic results are not random but possess a high degree of colour and structure. However the high fractal dimensions and fine structure of some attractors makes their mapping difficult.