The advocates of each of the three views of demarcation have provided some evidence that their respective view is a necessary condition for the development and maintenance of a progressive, or at least. active science. On the other hand. each of these advocates claims that the others contribute both to stagnation and instability in science What I would like to show now is the role each of these views should play in the structure of science. and that to the extent they have played these roles in the past, science has reaped substantial rewards. The first step is to acknowledge that science is complex. and the key to understanding the interrelations between scientific activities lies in a proper appreciation of the .structure of complexity.

Most people recognize that some things are "complex," and more often than not the term is used either as a synonym for "complicated" or as an excuse for accepting vague or ambiguous explanations based on the idea that complexity can not be expressed in simple language. However, it is a major point in this thesis that complexity functions to simplify complicated systems. For example, an automobile is a complex system. It consists of a number of sub-systems such as the engine. drive train. and body. Each of these in turn exists as a set of sub-sub systems such as the pistons, the valves the differential, the transmission. If in order to drive an automobile it was necessary that each of us interact with all of these separate levels individually, few of us would even attempt it. When we drive an automobile we abstract out one level, that of the operating controls and assume that all of the lower levels will do what is expected of them. In fact if they don't most of us would call a mechanic before getting under the hood for more than the most rudimentary problems at to deal with a problem where we lack the prerequisite techniques.

Unlike a complex system, a complicated system is one which has a great number of degrees of freedom which would make it difficult if not impossible for us to control if we were forced to deal with all of them at once. A complex system exists on several levels such that we are able to abstract out one level and ignore the others or at least take them for granted. The technique of the auto mechanic or engineer are just not important to the driver. What is important to him is that their designs or repairs have been done reasonably well. In fact engineers and mechanics have very anarchistic ideas on their own role and very often different mechanics and different engineers disagree violently on specific techniques All this matters little to the driver as long as the techniques they use prove successful.

From a slightly different perspective. the engineer is constrained to design the automobile to serve the purpose of the driver. or else it just will not sell. The mechanic too is constrained by the driver and owner of the automobile. If his techniques are not effective or too expensive he will soon lack customers. The constraints on the driver are also determined by the ability of the engineer to design a car that will do what he wants it to do and by the ability of the auto mechanics to keep it doing those things. Each level in our automobile system has a great deal of autonomy. but at the same time each is constrained by the levels above and below and in turn constrains them .

The same can be said of science. A scientist studying a new concept in bridge design does not concern himself with the molecular pattern of his materials. He may be interested in the tensile strength. of course, but have only an academic interest in the molecular pattern that led to that strength. The bridge designer is Constrained by the ability of the metallurgist to control the strength of his bridge materials. The metallurgist is constrained by the variability possible in the molecular structure of his metals. Despite the constraints. each also interacts with the entire system only on those levels where they maintain control and abstracts out those that are on levels above and below that.

Now these concepts are intuitively obvious. but a more formal description of complexity may not be so easy. The statement, for example. that complex systems exist on more than one level seems straight forward enough, but in many such systems there is no clear dividing line between levels. A good example is the driver who uses a vacuum gauge and a tachometer to increase the efficiency of his driving. The level with which he interacts when driving is considerably broader than that of the man who uses only the standard idiot lights The first person, in a sense, encroaches on what is normally the level of the auto mechanic. We can usually tailor the breadth of our level of interaction according to our needs within limits.

Returning to our discussion of complex systems, they are structured as a special kind of hierarchy. Unlike the more common dominance hierarchy, in complex systems levels can be made up of lower levels and in turn can make up the higher levels. For example, the lower level of our automobile system such as the drive train and the engine make up the automobile itself. As a result If one level is removed then the higher level disappears. and the lower levels lose their organization. You can not build a car without the engine or the drive train and it is difficult to visualize an engine peering at a pile of pistons and connecting rods.

The most important quality of complex systems is that they make possible an increased amount of variety. The twenty-six letters of the English alphabet can be combined to create a large number of words. With syntax and punctuation these words can be combined to form an even greater variety of paragraphs. and these paragraphs an be combined to form an even greater variety of manuscripts ranging from Popeye to Shakespeare.

If we return again to the structure of complex systems, we find that there is a special kind of constraints that biologists have discovered. called "Hierarchical Control Programs" which operate at the interface between levels. They limit the They Limit the behavior of the lower level in such a way that they promote the freedom of the Upper. This may sound contradictory, and in a sense it is. To obtain a clear picture of this mechanism we need to look into the interactions within complex systems a little closer. We can begin with the idea that each level is made up of the level below it and helps to make up the level above. Words are made up of letters, molecules; of atoms, social systems of people's ideas. The emergence of an upper level can be represented by a pattern of limitations on the level below. A molecule is a pattern of constrained atoms, a word is a pattern of constrained letters.

When we talk about hierarchical constraints, however, we do not mean simply limitations. on the behavior of lower level elements. This is a constraint which chooses from among the variety of interactions that exist in the lower level. The Criteria for such choices are determined by the upper level. In this way it can be seen that. the forces that determine the lower level interactions may or may not have any relationship to the emergence at the higher. The role of constraint is limited to choosing from among the lower level variety those behaviors which successfully meet the Upper level criteria. Atoms may interact in a great many ways but there are only a limited number of patterns that will emerge into viable molecules. Letters too can be assembled in a great many ways but only some of these combinations result in meaningful words. Even more important is the fact that the letters, the words, and even the choice of grammatical construction may be irrelevant to the merit of a Piece of Literature.

This points out the difference between simple and complex interactions. ln a simple interaction the initial conditions, determined primarily by energetic relationships, existing at the time of the interaction determine fully the final conditions. Once energetic equilibrium has been achieved, then all further interaction between the elements is "frozen out" no further interactions are possible. In a complex interaction the final outcome is not necessarily determined. Even though multiple coincident interactions may reduce the variability of a system, it is the emergence of hierarchical constraints and the requirements of the upper level system which determine the final condition. At high levels of complexity there may be arbitrary stable outcomes. The final choice may be determined as. much by chance as by any specific forces or conditions. For example, the emergence of a language among communicating people can proceed ln innumerable different directions. Witness the number or viable languages Systems in the world today. It is only in retrospect that we can find causative patterns, This is the same mechanism that governs. the emergence of criteria of demarcation. It is no wonder that much modern scientific philosophy is developed by men like Agassi, Lakatos, and Kuhn who are oriented toward scientific history.

Now we will pursue an examination of the structure of science from the perspective of complexity theory. There is. no reason why we cannot select, as our lower level starting point, the interactions between working scientists in their normal activity That activity is the search tor scientific answers to real-life problems, whether theory building or practical application. The first question we might ask is just what is the role of demarcation in the activity or the working scientist

To begin with he must know which of the theories, methods, and attitudes he is using are those that his contemporaries consider to be scientific. It is; the nature of science that ones contemporaries will judge the success or failure of a scientific endeavor. This is a matter of Popper's pragmatic third world and not his autonomous. These methods, theories, and attitudes which scientists in general accept (paradigms) have resulted in the past in "scientifically acceptable" results. However, as any sceptic will point out, if, as; we have shown, the results of complex interactions are not determined, then the fact that a theory, method or attitude was successful in the past does not imply that it will be successful in the future. To overcome this problem, a scientist, even one solving real life problems, must be aware of the criteria of demarcation that his colleagues accept as proper. He must, however, also he prepared to abandon them and to pursue any method that helps him to solve the problem. If he is successful then he must e prepared to justify the use of unacceptable procedures to his contemporaries. It is they, and the future, who will determine whether the work is or is not scientific.

The true sceptic will object to this. He Will insist that such justification is not only unnecessary, but regressive also, insisting that the success of solving the problem is sufficient justification. The reason he is mistaken is that it Is not necessarily the successful solution that is being justified, but the unorthodox procedure. This procedure has resulted in the successful solution to a problem that acceptable procedures Could not solve. Obviously. either it or the limits of scientific procedure are problematic. It is as much in the interest of science in general as it is to his own work that this be determined to everyone's satisfaction.

There is a difference between the kinds of activity, of theories. and attitudes. that permeate science and those for which non--scientific fields. are noted. That difference has gone through continuous changes throughout the centuries. In the heyday of lnductivism and Newtonian optimism science dealt with empirical facts, with a priori truths. The fall of Newtonianism started a scramble for other equally clear and unambiguous; criteria. In the future I suggest that scientists will develop new and even more intriguing criteria for demarcation. The details of demarcation criteria prevalent at one period or another are not critical to our study of the structure of science. Science is essentially what scientists say it is, and there is an emergent level of the complex scientific system. Scientists generally accept the successful procedures, theories and attitudes of the past, or at least an eclectic assembly of them, as the important criteria. Some may have good reason to believe that this has been an obstacle to scientific progress, and that it may even continue to be. Dogmatic adherence to arbitrary criteria to the detriment of new and novel developments is. neither good nor necessary. It has undoubtedly happened in the past and probably will in the future. It represents a clear danger inherent in the structure of science and one which all scientists must be aware of and on guard against.

Poppers discussion of demarcation is open to the charge of ambiguity, but one of his points that is continually reiterated by many of his followers is that a clear set of demarcation criteria will help to separate the rational from the irrational. That may be so, but it seems to me that the most important element of Popperian philosophy is the critical attitude it teaches. As a hierarchical control system ln the structure science, demarcation should act to increase the growth of science by choosing, from among the variety of activities scientists are engaged in, those activities that best augment the set of attitudes, theories, and methods that constitute the prevailing paradigm. They can accomplish this only if they provide clear guide-lines to the working scientist. When he exceeds those guide linea he should know he is working in a questionable area and he should take special care to document what he is doing. Later, he and others should examine it critically to see if he has in fact expanded the role of science, or if he has fallen beyond the acceptable so that his work: could be brought into question.

The action of determining that a given piece of scientific work does or does fall into accepted criteria for demarcation should not be taken lightly. Future historians will find and point out those times that mistakes are made that result in a stagnation of science. However, too often it is only in the future that the correct choice will really be known. As a rule for action the best we can hope for are scientists with open minds and critical eyes.

For the reasons we have been discussing, the existence of a criteria for demarcation developed from Popper's autonomous third world is highly questionable. I have also shown that the sceptic's call for abdication is equally problematic. We come now to the elitist point of View. The danger of the elitist attitude is that it becomes the platform of the academic power-broker. Criteria for demarcation are not chosen by scientists. They are chosen by the results of all those who recognize new successful theories and have the courage to test them, often even outside of the field in which they were developed. When a scientist working in his own area of specialty develops a method or theory that solves a scientific problem while violating some criteria of demarcation. Then, if it is found that modifying that criteria to include these new ideas results in other scientific problems being solved, or as Lakatos would call it, a progressive problemshift, the inevitable result is a victory for science and an expansion of the paradigm. Therefore, the elitist position cannot be a role for the censor, It can only be determined by a critical analysis of all relevant details totally free from dogmatism.