Explanation in science

The usual usage of the term 'explanation' broadly includes relating a fact to something else in our experience. As such, the word is used in much the same way as the word 'understanding'.
Scientific explanations represent particular forms of understanding (Verstehen), which vary with the different types of science.
All types of scientific explanation aim to:-
1) give a full and explicit statement of the condition of a phenomenon
2) relate a fact to a system or a combination of facts as a whole.

As such, science aims to be descriptive and not prescriptive (not norm-setting). Thus science avoids trying to answer questions of 'why' something exists or happens by giving an exact and exhaustive description of 'how' a thing is or occurs. Scientific theory avoids, as far as possible, metaphysical assumptions and speculations in favour of empirical observation and generalised description. Descriptions can be generalised where they report on the essential features of events, such as we often term causes and effects and the type of operative conditions that must accompany these. Accidental or irrelevant facts are thus eliminated. Empirical laws are basically only summaries of the essential features of a large number of differing sorts of observation. Their 'explanatory' function lies in the degree of generality and precise abstraction (often mathematical) of such 'laws', which allow one to make accurate predictions about any number of particular instances of the same type as those upon which the law was originally based.

To say that scientific explanation, aims to be descriptive and non-normative is also to regard science 'ideally' as being limited to the pursuit of 'pure' knowledge, without thereby regarding the purposes inherent in this search for knowledge. As such, most science does not itself question 'why' it obtains knowledge, for what utilitarian, social or political purposes it may be pursued. Its 'explanation' excludes any exposition of scientists' own understanding (Verstehen) of the purposes or goals inherent in the researches, though some such purposive motives are always in fact operative, or else the scientist could have no self-understanding, i.e. no understanding of scientific activity as part of the scientific institution with its very great consequences for society, whether for better or worse. Scientific explanation, therefore, always presumes an extra-scientific understanding of its potential, either in technical application ('applied' science) or in influencing attitudes, values and policies in society and life in general. Such understanding is usually peripheral to internal scientific debate itself.

In the following differing types of explanation are distinguished at the outset, before considering the methods involved in more detail.

As made clear in the foregoing section, there is a very distinct difference between using intentional and, extensional frames of reference in interpreting symbolic behaviour (such as language). Applying the same distinction to types of explanation, one may speak of extensional explanations as opposed to intentional ones.

Extensional explanations are those which depend on demonstrating that an event is the result of physical causes using evidence that is always ultimately dependent upon sensory observation. Thus, 'extensional1 explanations are those that are based upon the material world of 'extension' (i.e. of matter in space/time). They are typified in the exact natural sciences, where they take the form of causal theories. Extensional explanations are those that assume the existential primacy of the 'external world' of commonly-observable and 'objective' phenomena as the frame of reference for our common scientific knowledge.

As such, extensional explanations reject as 'subjective' and secondary all the facts of consciousness or the 'inner world', eliminating these where possible (eg. as in the form of observer bias or 'subjective interpretations of meaning' of phenomena). In. short, all intentional explanations are rejected from the strictly physicalistic scientific viewpoint.

Intentional explanations are those which depend upon showing that an event is the result of purposive agency, using evidence that always ultimately depends upon the valid interpretation of the agent's meaning or purpose. Intentional explanations can be about both subjective and objective phenomena, thus they include both facts of sensory experience and of consciousness in their sphere of study. Such intentional explanations cannot simply be judged true or false by reference to observation because 'meaning' is not an objectively-observable fact, but a 'subjective' or psychic fact accessible only to interpretation upon the basis of evidence, testimony, reasonable argument etc.

As such, intentional explanations are those whose explanatory force ultimately lie in the 'inner' experience of meaning as evaluated or 'intuited' by the interpreter of the act or event in question. Their focus is less the outer 'substantive' world than the inner world of intuited meanings... whether this so-called 'inner world' is purely individual or is also collectively-shared.

Intentional explanations are typified by the historical sciences wherever events or acts are explained by reference to such intentions as motives, purposes, guiding ideas etc. Historical events that come about as the consequence of a number of persons' actions (possibly combined with given objective circumstances, prevailing conditions etc.) are explained by reference to intentional meanings (possibly combined with empirical studies of the existing conditions) so that the end result is an intentional explanation supported by reference to objective conditions. The validity of such an explanation cannot in principle be tested by empirical means for it will rely mostly upon testimony (interpretative reports made by 'subjective' observers) and the researcher's own interpretation of the evidence as a whole. Whether such a theory gains wide scientific recognition or not will often depend much more upon a variety of other factors than the degree of demonstrability of its truth. (These factors may be infra-scientific (such as its rationality, its 'formal' acceptability as regards such things as the reliability of sources quoted) or extra-scientific (such as whether the theory 'catches interest' or is controversial, whether its propagation is heavily funded or not, even whether it opens up possibilities of further researcher employment or not etc.).

Some sciences have as their subject matter both objective physical and subjective psychical phenomena (i.e. both extensional and intentional phenomena), for example, sociology studies the consequences of human actions and interactions upon society. Society is not simply a physical reality, it is also something which exists by virtue of the minds of its social members and which also derives its particular nature or organisation at any epoch or place from the way it is conceived and 'run by its partakers. A study of society will therefore necessarily involve a researcher in accounting for both the objective conditions that prevail at any time (physical, climatic, geographic, economic, demographic etc.) and the subjective conditioners too (such as the actions of individual policy-makers and the concerted meaningful action of groups to alter physical conditions or any other conditions than may arise in matter of the non-natural sciences has been succinctly described by Jean-Paul Sartre as the "practical-inert field". Society is the result of historical human action (praxis) working upon the 'inert' matter of nature. Thus, buildings have various functions for the members of a society and, as such, are social as well as physical facts. Their inert matter is given meaningful form by the 'subjective' intentionality of praxis (i.e. the purposeful, meaningful action of individuals and groups).

The same may be said of any object with social significance. A banknote, for example, is a physical thing with symbolic value. Only if others agree to interpret the piece of paper you offer them as a banknote representing genuine currency with convertible value does it become a banknote. Otherwise it is merely a physical thing. Its full social significance is complex indeed and is symbolic, signifying some form of human praxis (which includes the act of interpreting and finding or giving something a meaning). Similarly, all use of meaningful symbols exemplifies the practical-inert field. In writing a book, the inner 'intentional' meaning is expressed in material symbols, becoming an .inert physical thing, a book.

Functional Explanation

Where one is concerned only with the physical aspect of things, ' extensional' causal explanations are sought. Likewise, v/here one is dealing only with the intellectual, psychic or socially-meaningful aspects of life, intentional explanations apply. In the area of the practical-inert, however, one is often concerned with a complex interweaving of the physical and the psychical, or say of the successive 'reactions' of the natural environment to the 'actions' (and interactions) of humans... and vice-versa. In this situation the social sciences often attempt explanation by reference to the function of some phenomenon towards attaining some goal. The goal may be of a collective nature - or possibly of a biological or some other supra-rational sort. For example, for a society the assumed goal may be political stability through avoidance of revolts and revolution. More abstractly or generally it is often formulated as 'social equilibrium' or 'balance between counteracting social forces and pressure groups 'etc. Important social facts are attempted explained by showing their role in bringing about this 'social balance'. The hard suppression of criminal violence, for example, may be functional under given conditions, non-functional under others. (Consider, for example, the death penalty as a means towards or away from social equilibrium).

The chief difficulty of sociological and anthropological functional theories lies in justifying the goal assumed to be given. It is often simple presumed (from 'common sense') or taken to be either 'biologically' or collectively pre-determined as an optimal goal. This seems to be taking a political standpoint without admitting that it is political and not scientifically-founded in the fullest possible empirical observation.

In principle, functional explanations are logically contrary to the model of causal explanation in that they explain - not by reference to a proceeding event - but by reference to a subsequent possible future event. Though logically contrary, however, they need not contradict and exclude one another.

While causal explanations are invariably nomothetic in that they extract what is essential or universal in the relations between physical phenomena, so-called 'functional' explanations are more or less ideographic in that they would show the advantage or disadvantage of some phenomenon in relation to a specific system in which that phenomenon occurs. For example, in the evolution of living forms each specific feature of any plant, animal etc. can be regarded as functional or dysfunctional (i.e. disadvantageous) within the ecological system wherein the particular plant or animal has come to be). The giraffe's long neck is thus regarded as functional (advantageous) for the giraffe in an environment where forage is otherwise scarce.

Functional explanations are thus in a sense 'holistic' for they always presume some view of a 'whole' or a system in relation to which one regards various phenomena (such as the specific characteristics of an organism regarded in relation to its natural environment say, as advantageous for the survival of that organism in that particular environment).

In the human or historical sciences, functional explanations have not generally achieved a secure status as conclusive scientific explanations. Though functional theories (such as that of Robert K. Merton in sociology) are based in empirical studies, the coals or ends which they assume (and for which the various phenomena are functional or non-functional etc.) are themselves assumed and cannot be proven to be objective and determinate in nature. Thus, in some periods of a society's history, for example, violence towards a certain class of people may have proven to be functional as regards maintaining a 'social balance' or a status quo. Nonetheless, the particular social circumstances judged by the sociologist as being 'social balance' cannot be proven objectively or by reasonably empirical means whatever to be either positive or necessary for that society (or for societies in general).

The use of functional explanations in human sciences is mostly still highly controversial in that there exist no theories that have achieved widespread acceptance among practitioners of the science involved. Functional explanations are not demonstrable with the same exactitude or empirical control as are causal explanations (as for example in molecular biology). So they are frequently contested by scientists working in the same or adjacent fields.

This type of explanation is a special case of functional explanation and is found in general biology, including ecology. Morphological explanation is an empirically-based account of an organism's particular 'form' (Gr. morphe) as being functional towards some naturally pre-determined goal. (it is sometimes also called teleological explanation} telos Gr. goal, purpose).

Teleological explanation has its main historical roots in the biology of Aristotle (i.e. the doctrine of the 'final' cause) and in the botanical and anthropological discoveries of J.W. Goethe. It requires the assumption of some principle of the nature of a purpose at work in nature. Thus, in its occurrence even in the strictest Darwinist theory, the principle or purpose of 'survival' is assumed as the goal. In general biology, for example, the form of an organism (such as the specific colour of a flower) at a specific time plays an observable role in the interplay of ecological events (such as attracting particular animals or insects for pollenation). The goal (fertilisation) is primary to the means (flower colour).
The chief difficulty of morphological explanation is how to decide which principles or purposes are inherent in nature, since this cannot be empirically-demonstrated as such. There is no strictly scientific means by which one can decide between principles like 'mere survival of the species' (Darwin) and 'the maximalisation of ecological diversity' or 'the fulfilment of a supra-intelligence's plan for further evolution' (P. Teilhard de Chardin - biologist, paleontologist)

It is widely accepted, that the natural or physical sciences aim at a system of generalisations about observed facts referred, to as 'causal explanation' of events.

The layman view of a cause is that it is something which has some peculiar quality or power which it 'possesses' to produce the 'effect'. Here lies a main divergence from the scientific point of view which sees the cause as en event. The term 'cause' does not express any 'peculiarity' that an event possesses - but is rather a satisfactory name for the event which precedes another to which it is repeatedly related.

Furthermore the popular view differs in that, where a complexity of events is observed, it regards that event which is most striking or predominant as the cause rather than what is essential and 'immediate', for practical purposes it is common to name the cause vaguely - as when we want to avoid the spread of an infectious disease like T.B. we practice hygiene because unsanitary conditions cause tuberculosis. A medical scientist will of course isolate the cause of the incidence of T.B. to the introduction of particular germs into the system, examining the chain of events that is essential to each case of that illness. For the scientist, then, each typical event has a typical immediately preceding cause.

A logical definition may be: An immediate cause is one event which is directly antecedent to another, called the effect - the two events being so related that the one can always be observed as following the other in the same sequence.

Both are part of a continuous process such that no distinct line can be drawn between the two events - only a vague separation conceived in thought: 'A line of separation, but a line without definite edges' has been used to explain this. Since both events are part of what is best described a 'process' - and since a process requires duration of time - however short or long - the 'cause' always precedes the 'effect' in time. neither are the two events separated in sequence by any other event or events which are unrelated.

This definition is good in so far as we are concerned with only one isolated 'cause' and its 'effect'. In most material situations there is a complicity of events which seem to be inter-dependent in the most complex ways. Hence it is called an 'immediate' cause. This means one cause related to one, single 'effect' and isolated in thought from all the other inter-related events - the immediate cause being that which is related closest to the effect and that which is essential to its production.

It is firstly necessary to distinguish between laws arrived at by genuine induction and summary laws. A summary law is one which is merely the sum total of previous observations - or - in most cases the term refers to a summary of previously established laws. Boyle's law is a summary law in the sense that it is merely an addition of the established laws of causal connection established in the case of each known gas. Hence, we cannot obtain explanation of any of these causal laws by mere reference to the summary of them (i.e. Boyles Law).

It is only when a law is arrived at by genuine induction that we can refer to it for explanation. In the case of Boyle's Law the genuine induction takes place when the generalisation "all cases of this gas show an inverse variation between pressure and volume at a constant temperature" is drawn from the experimental data.

There is a tendency to call these more comprehensive genuine inductions 'theories'.
The term "theory" has been used often in a depreciatory sense ... but, except in truly hypothetical cases, what has been established as theory is meant to be true in practice. The reason for the logical application of the term 'theory' to such laws as explained is justified by:-
(a) the laws to which they refer are in fact more theoretical... they are inductions based on laws, and thus further removed from the facts of observation. Hence, if a theory is true, then the laws from which it is induced must be true from the formal logical point of view)... whereas the laws may be true even if the theories are false.
(b) looking at laws objectively, and not merely as verbal formulae for them, it is seen that to "call the theories also law amounts to counting the same laws twice over.
Hence we normally talk of discovering a law, but of inventing a theory. Laws of a less comprehensive character which cannot be deduced from theories are often felt to be unexplained... and are thence called empirical (laws or) generalisations.

On the other hand, a 'theory' is sometimes used to refer to a suggested hypothesis which is, as yet, insufficiently verified. Its antithesis in this sense is the term 'fact' when it refers to "an event (in the widest sense) when it is the case in Nature".
To sum up... a theory can refer to hypotheses which are as yet insufficiently
verified -- in opposition to the term 'fact' ... or for the more comprehensive, all-including abstract laws, which may have been sufficiently verified. Such use of the term theory does not include 'summary laws'and is opposed to the more usual use of the term law for the sake of distinguishing these two terms.
The following diagram may be helpful in representing the relationships between the terms just described.
(most explanatory, theories least descriptive)

most explanatory least descriptive - greater degree of abstraction and comprehensivity

lesser abstraction, comprhensivity - most descriptive, least explanatory

theories

summary 'laws'

empirical generalisations

classifications of facts and hypothesis formulation:

observable facts

The term 'empirical generalisation' can be defined as 'generalisations of a less comprehensive character that cannot be deduced from theories'. They are therefore considered to be unexplained themselves and are more descriptive than explanatory. Being themselves in need of explanation, they are not sufficiently connected with other established generalisations or 'laws' to make them extendable to all cases. Empiricial generalisations arise in the more preliminary stages of any exact science before they can be brought under principles of greater generality. When this is made possible, usually by the further discover of principles at a high level of generality (i.e. such as the discovery of E= mc2 or of quanta theory, or of the D.N.A. double-helix structure), they are no longer isolated empirical generalisations but assume the status of established scientific principles.

Causal explanations are all extensional. So, even if the above 'model' of the structure of a scientific system, is universally valid in the natural sciences, it cannot be said to be so for the human-historical sciences. For these latter, the conscious designs of human will (i.e. intentional behaviour) are always somehow a factor in their subject matter, however indirect its influence. Causal explanation is still held up as an ideal by researchers in some branches of the social sciences (notable experimental behavioural studies, demographic or other 'mass' sociological studies). This raises the central issue as to whether or not human behaviour - including intentional behaviour - is always and only causally determined. If so, the physicalistic causal model (one effect -has only one immediate cause) becomes problematical. Further, the issue of whether meaningful acts can be studies only as physical events arises.

There are usually many different circumstances necessary for the production of an 'effect' - and the popular view is that each of these circumstances is a 'cause' of the 'effect'. This is called the doctrine of 'plurality of causes'. Now, in many cases, after closer analysis of the causes - and especially of the effect, this 'plurality of causes' disappears and we are left with one essential or 'immediate' cause. For example, when we say 'there are many causes of misery' we often neglect to realise that there are many kinds of misery - each produced by its own kind of cause - such as misery from lack of money, from the death of a near relation, from loneliness, etc. All logicians and scientists agree that plurality disappears in most cases after closer examination. Some logicians, however, hold that plurality cannot exist in a causal relation. They assert that the statement 'If A is the cause of B' is reciprocal (that 3 can only be produced by A). Other logicians deny the certainty that plurality cannot exist - and prefer to suspend judgement indefinitely. This seems more sensible since those cases where we have not yet succeeded in discovering the one immediate cause (where a plurality still seems to exist) may be cases in which plurality actually does exist. To this extent experience shows a plurality of causes. Most schools of thought recognise the cautionary value of the doctrine of 'plurality' - for when we observe seeming plurality we are reminded that we may also be observing a plurality of effect - and so we are prompted to analyse further.

Where one lacks exact methods to demonstrate single essential immediate causes, such as in the social and historical sciences generally, the reason may lie in the nature of the subject matter. Insofar as any study in psychology, sociology or history deals with human acts - whether at the individual or the collective level -rather than with natural phenomena, the thesis of voluntarism or free will have to be considered. The thesis of voluntarism holds that, since human acts are not compulsory reactions to stimuli, but are deeds consciously-evaluated and chosen among possible alternatives, they cannot in principle be explained causally with scientific rigour. In accordance with this thesis, human acts are distinguished as such by being essentially meaningful to the agent(s) and not being unintentional reactions to stimuli. Thus, acts cannot be observed or measured as if they were natural phenomena because this is to reduce them to mere physical behaviour and lose the essential phenomenon, the meaningful agency that makes an act what it is.

The above thesis of voluntarism, if accepted, implies a denial of the Principle of Universal Causation. This principle holds that 'every event has a cause'. As already discussed, there is ho way in which this principle can be proven empirically or even by conclusive logical argument. The view of human action as being a meaningful choosing between alternatives almost invariably implies a thesis of indeterminacy, holding that not all phenomena are-determined according to invariable laws by proceeding events. As empirical evidence to support this principle the common experience of freedom of volition is forwarded. If one understands what it means to make a choice without being under external or internal pressure, one has the experience. Further, Heisenberg's principle of indeterminacy in quantum physics is often referred to as supporting evidence for the thesis of voluntarism. Heisenberg's principle implies that the movements of the individual atomic particle are unpredictable in principle. Only at a mass level can the movements of elementary particles be predicted. This thesis is used to argue for indeterminacy of human acts, both because it is an instance of a type of 'uncaused event' itself, and by analogy. The analogy is that, just as the behaviour of an individual particle cannot ever be predicted with scientific certainty, nor can the individual person's acts be predicted with certainty. True, some may prove to be predicted correctly in advance of the act, but the prediction is no better than an informed guess in principle. Similarly, the movement of large numbers of particles can be predicted on average, leading to the analogy that the behaviour of large numbers of persons can be predicted on average.

For example, one may predict with a very high degree of likelihood that the citizens of India will not, on average, earn a wage higher than the citizens of the U.S.A. within a decade. The prediction, is not certain (since the U.S.A. might be destroyed by a nuclear catastrophe of 'local' proportions in the meantime), but it is highly likely if other conditions remain largely within the same limits that have applied since the beginning of the century. Meanwhile it is evident that one cannot predict whether any individual Indian citizen will not earn more than any individual U.S. citizen with any degree of certainty. All that is required to invalidate such a prediction is that an individual wins a large lottery and can then live from interest earnings. In this example it is seen how precise prediction of human activities or affairs is a chimera, even if it were' desirable, due to the practical unfeasibility of knowing the operative conditions sufficiently well to have a situation even remotely comparable to that achievable in the laboratory. The operative conditions of any human action cannot be delimited, ultimately they extend to the overall movement of world history.

Those who reject the thesis of voluntarism, however, hold that the experience of freedom of will is merely a subjective accompaniment of an objective series of physical events which are probably subject to causally-determining precedent events (causes). Though the thesis of indeterminacy is denied, the opponents of voluntarism invariably hold back from asserting that 'every event has a cause' is true. Hume's view is most common here, that we cannot prove universal causation, but we must work on the assumption that a cause will eventually be found for everything, including human acts. Such a view rejects the distinction between 'causes' and 'motives'. Human acts - whether individual or collective acts - can be explained causally as physical events, according to anti-voluntarists. The psychic experiences (say of freedom of will, effort of will etc.) which accompany these can also be explained, for they are reduced to physiological or electro-chemical events for causal analysis, holds the voluntarist. It is pointed out that this explanation is possible in principle, but so far impossible in practice due to our limitations in neurological knowledge (i.e. of the brain).

Those who reject the above thesis of 'causal reductionism' or 'physical reductionism' -i.e. those who accept voluntarism -hold that, insofar as human behaviour is conscious and intentional it cannot be scientifically explained, but must be interpreted as to its intentional meaning if it is to be understood.

This view regards the psychological and social sciences as primarily concerned to understand (by Verstehen) the individual development of personality and the collective movements of society and history rather than seek causal explanations (Erklärung). As a secondary part of the task which applies insofar as human behaviour is non-intentional and non-conscious, causal explanations are to be sought for these aspects of human behaviour. In physiological-psychology, for example, a causal explanation of irrational behaviour - anger without cause etc. may prove to be the presence of brain tumour. So, what in ordinary language is often called 'anger without cause' proves to have a precise natural cause. What it lacks, is motive ... that is, a reason, grounds or what one thinks of as an 'intelligible' purpose.

The high prestige historically accorded, to natural science is generally recognised as being rooted in its ability to predict accurately with consequent 'useful' technological innovation as a result. This is also largely due to the precision of its hypotheses and/or theory, which can be mathematically formulated and applied.

In the natural sciences the ability to make accurate predictions is extremely high in some fields. Where the advancement of theory and experimental practice allow very accurate measurement and therefore the quantification of data, the likelihood of precise prediction is high. Such is the case in physics, including astronomy, optics, molecular biology etc. In other natural sciences such as general biology, including zoology, botany etc., or geology, - where 'qualitative' factors still play an important role because all factors cannot meaningfully be quantified, - the likelihood of accurate predictions in field circumstances is generally low.

In general, there are two main limitations in natural scientific predictability, as follows:-
1) There are in principle no certain guarantees of full predictability in any science. This has become widely held to be valid since Hume first forwarded the view in connection with his refutation of necessary connections being provable in causal events. (Note: Hume did not deny that the causal connection is necessary, but only that this cannot be proven conclusively). Also, as noted, in microphysics it is held that individual sub-atomic particles are unpredictable in principle. Only in the mass can their behaviour be predicted.

2) Accurate predictions do not only depend upon full knowledge of the appropriate 'laws' that apply in any case but also upon their correct application in the particular (experimental) situation. Human error is a factor' which cannot, in principle, be eliminated, hence full guarantees of scientific predictability can never be given. Human error may occur in calculating the expected results of a new experiment, in failing to find faults in the apparatus being used or in observing and analysing the results of the experiment. For example, during the Los Alamos 'Manhattan' project in the early 1940's many predictions of the explosive force of the first atomic bomb were forwarded, ranging from a fraction of the actual force finally measured to much higher, but 'unlikely' estimates that still proved less than the actual result. In short, Robert Oppenheimer and his colleagues were amazed at the magnitude of the explosive event.

In the non-natural sciences many attempts have and are still being made to emulate the accuracy and ability to predict of the natural sciences. As part of this programme, the quantification of one's material into 'hard' data has become widespread and the techniques offered by computer science have greatly improved the processes of analysing, comparing and presenting data in all branches of research. Insofar as the particular phenomena of any study are such that they allow of prediction, improved data techniques give improved predictability. However, the rise in population of a city or, say, the gradual shift of opinion from traditional labour to modern 'green' political parties, can be predicted only as long as - and to the extent that - they remain stable trends. The reliability of opinion polls for a period of a decade, however, can suddenly be overturned dramatically by the introduction of a new political issue or some other decisive factor.

In the human sciences, there is a possibility of the 'self-fulfilling prophesy' arising. This is the case where a prediction tends to become true due to the fact of its having been made. For example, a prediction of the following sort might tend to be self-fulfilling: 'On the basis of the information that only around 1% of moderate to large-scale tax evaders are ever caught, it is predicted that tax evasion will increase yearly throughout the decade'. This prediction may act as a suggestion to potential tax evaders. A similar function may be involved where predictions are based on the continuance of existing trends. For example, to predict the rate of increase in use of private cars as a basis for road and vehicle planning policy, may tend indirectly to increase the pressure on families etc. to own a private car (due to the reduction in public transport etc. that is brought about by the same trend and prediction). Planning on the basis of predicted trends can be likened to driving a car by watching the road only through the rear mirror. One robs oneself of the power to create policy by believing too much in the trend that results from previous policies.