In recent decades physical science has declined from the position it once held. Its connections with pollution and weaponry have lowered its prestige, its representatives are less welcome to pontificate, and even the literary genre still known as science fiction has turned towards occult fantasy. Neither the profession nor its supporters have retreated willingly, and a substantial rearguard maintains a defence. W. V. Quine, a prominent and respected philosopher, maintains that ‘our elaborate scientific theory of the world… leads persistently to successful prediction of observations.’  Not long ago, to question that statement would have brought automatic relegation to cranks’ corner; now we have greater freedom to examine it.
Holding strictly to Quine’s present tense would rule out all consideration of evidence; in order to support his statement he would have to show what scientific theory has achieved in the past, and we may do the same. Also, his phrase, ‘elaborate scientific theory of the world,’ cannot be taken in the singular as given; Quine may know of some one scientific theory covering the whole world, but if so it remains irrelevant for discussion of the social position of science; the status this activity has enjoyed, and still does enjoy though less abundantly, comes as a response to contributions made by a large number of theories; Simon Schaffer describes the sciences as ‘made up of varying and often contradictory communities of practitioners.’  The scientific ‘theory of everything’ (not a phrase Quine uses) which shall integrate them all remains an aspiration.
As science has pursued its investigations more widely many theories once accepted, for example phlogiston, the geocentric universe and the indivisible atom, have turned out to give predictions not agreeing with observation. They have been rejected in favour of others, but this does not provide ground for dismissing them as unscientific. To reserve the right to exclude from the canon any theories, once accepted as scientific, that might later be found to give false predictions, would constitute reduction of ‘scientific’ to a synonym for ‘successful.’ If this be done then the assertion, that scientific theories are successful, loses meaning; it amounts to saying only that successful theories are successful. Once the scientific community have accepted a theory as scientific they are stuck with it; if it fails under further test they have to modify their triumphalism. It may even be the case that the proposition, scientific theories do not lead to successful prediction of observations, carries greater truth than its contrary.
This raises difficulties, for if scientific theorising does not lead to reliably successful prediction it becomes hard to see why society has subsidised the activity in the past and harder to see why it should continue to do so in future. The cost of the equipment required, for the development of their theories, by the nuclear physicists alone has begun to strain the resources of even wealthy nations. What do we get for our money?
An approach to an answer requires a field of view wide enough, both in temporal and social terms, to take in more than science. Isolated instances of scientific activity or approaches to it can be traced back at least to Babylonian astronomy, but it started to achieve an enduring presence in Europe around the 17th Century and the beginnings of maturity in the 19th. People had been around for many thousands of years before that. They lived almost entirely without science, and on the whole they prospered. Beginning as a few scattered bands they spread to occupy most of the earth, and they did not do this without being able to make successful predictions. Every hutwife who fired a pot, every husbandman who planted a seed, showed themselves to be making a prediction (usually unvoiced but none the less real for that) and it was successful often enough to more than justify the expenditure of energy. Much the same holds good today. The number of ‘qualified scientists’ remains minute, and although embodiments of their theories, in the form of machinery and sophisticated equipment, are all around us, yet in using them we act as unscientifically as people have ever done. The prediction, that if I press this knob a light will come on, rests on the same ground of experience or instruction as the prediction that if I drop this stone on my foot it will hurt. Scientific theories do lead to successful predictions, but to say this and no more is to misrepresent by selection; they also lead to predictions that do not agree with observation, and predictions that do come also from non-scientific sources. Science does not stand alone in making successful predictions, and since it focuses its attention on areas as yet unexplored (or at least not yet fully mapped) while unscientific life tends towards repetition of success once achieved, it may well be that science achieves a smaller proportion of successful predictions than does non-science.
In social affairs as elsewhere a principle of parsimony operates (on the whole and in the long term) and it ensures that any institution consuming as large a proportion of resources as science does must be performing some useful and distinctive function. If this is not the provision of theories giving successful predictions, what can it be? A pointer to the answer lies in science’s tendency to quantify.
The big, dramatic events, the revolutions, discoveries, breakthroughs and paradigm-shifts, do much to keep science newsworthy but they account for only a small part of its activity. Far more goes into the routine of counting and measuring, intended to render the match between observation and prediction more accurate. It is, indeed, largely accumulation of these quantitative advances that forces the dramatic qualitative changes; in Thomas Kuhn’s words, ‘novelty ordinarily emerges only for the man who, knowing with precision what he should expect, is able to recognise that something has gone wrong.’ 
Science has been said to use mathematics as its language, and its practice consists largely in the use of instruments which take accuracy beyond the capacities of our unaided senses. When making measurements in the farther reaches of the galaxy the astronomers may have to accept a margin of error measured in lightyears, but they still achieve a precision beyond the reach of the layperson, who can say only that the stars are far away, and it is mainly this that elevates their results to the rank of science. Both science and rule of thumb give predictions for the most part successful; the distinguishing feature of the scientific ones lies in their higher precision, with quantitative precision holding a foremost place.
The proposition with which we began implies the prediction that predictions will be found to fall into two categories: scientific ones which do agree with observation and unscientific ones which do not. This prediction itself fails to agree precisely with observation, for such a division hardly appears. Every scientific prediction is capable of further refinement and to that extent fails to correspond with observation, while it is difficult to think of any prediction, even the most unscientific, which does not give some agreement. Even ‘the sun goes round the earth’ succeeds with a main feature of the event, namely the presence of these two bodies; it just gets their relative movements wrong. Observation of predictions shows that rather than falling into two sharply distinct categories they lie along a range from more to less successful. By taking the distinctive feature of science to be the pursuit of ever-greater precision we bring our prediction into more precise agreement with observation.
We run some risk of swinging from overvaluation of science to denigration of it. In seeking precision science does more than add decimal points to measurements already accurate enough for practical purposes. Although it occupies itself mainly with refinement of predictions already made (even the remotest reaches of quantum theory are still trying to predict more accurately the behaviour of matter, something we all handle for the most part successfully), in doing so it introduces substantial novelty. In unassisted non-science the ‘noise level’ is often so high, the gap between theory and prediction so wide, as to rule out successful action; without science we should not have electric power on tap, motorised transport, medicines which often prove effective, or the prospect (more remote now than it once seemed) of unlimited clean power from fusion.
Unscientific activity aims at successful action, valuing precision only as it contributes to this object. Science follows a different course, setting up precision itself as the target and leaving technology, commerce and politics to decide whether and how to use the results obtained. It differs from non-science not only in degree but in kind; the transition, from the use of precision when convenient to the purposeful pursuit of it, involves a qualitative change in mental activity, a shift from one mode of thought to another.
In its concrete actuality science comes together with society and people, and both of these preceded it, maintaining themselves without its aid; more, they developed their capacities to a point where they were able to give science the resources it cannot provide for itself. They did this largely by making predictions which agreed with observation, and they have not lost this ability. Those predictions, and the actions based upon them, become more effective as they grow more precise, but although scientific precision comes as a new factor it still remains one factor in a complex. Science does not carry the overwhelming importance implied by the suggestion that it, and not these other modes of thought, permits the making of predictions which turn out to agree with observation.
We began with the observation that science no longer enjoys the respect and prominence it once did; both this decline and the previous advance have to be reconciled with the ideological pyramid, and they do not fall readily into place until we accept that this formation obtains with much exactitude only overall and in the long term. In daily practice it appears less as a fact than as a complex of tendencies restraining but not prohibiting deviations. In any particular place at any particular time one (or more) of the major ideologies is likely to exercise a greater (or lesser) influence, and successfully claim a larger (or smaller) share of resources than its position in the standard pyramid would suggest.
 TLS 3 July 92
 TLS 22 Nov 91
 Kuhn T.S. 1970 The Structure of Scientific Revolutions Chicago: University of Chicago Press, 65, emphasis in original.
from Ideological Commentary 57, August 1992.