Michael Polanyi

The Logic of Liberty

The usual thing is, of course, that when a number of persons apply themselves independently to parts of the same task, their efforts remain essentially unco-ordinated. A party of women shelling peas represents no co-ordinated effort, for their total achievement is simply the sum of their individual outputs. Similarly, a team of chess players is essentially unco-ordinated; for each plays his opponent according to his own lights and the performance of the team is simply the sum of the games independently won by each other.

By contrast we can see the distinctive character of science coming into view; it is not conducted by isolated efforts like those of the chess players or shellers of peas and could make no progress that way. If one day all communications were cut between scientists, that day science would practically come to a standstill. Discoveries might continue to be made during the first few years of such a regime at about the normal rate, but their flow would soon dry up and henceforth progress would become fitful and sporadic, and the continued systematic growth of science would cease entirely. The co-ordinative principle of science thus stands out in all its simple and obvious nature. It consists in the adjustment of each scientist's activities to the results hitherto achieved by others. In adjusting himself to the others each scientist acts independently, yet by virtue of these several adjustments scientists keep extending together with a maximum efficiency the achievements of science as a whole. At each step a scientist will select from the results obtained by others those elements which he can use best for his own task and will thus make the best possible contribution to science; opening thereby the field for other scientists to make their optimum contribution in their turn—and so on indefinitely.

We are faced here—it would seem—with a basic principle, leading quite generally to co-ordination of individual activities without the intervention of any co-ordinating authority. It is a simple principle of logic which can be demonstrated by quite trivial examples. Suppose, for example, we had to piece together a very large jigsaw puzzle which it would take one person several days or even weeks to complete. And imagine that the matter were really urgent, the discovery of some important secret being dependent on the solution. We would no doubt engage a team of helpers; but how would we organize them? There would be no purpose in farming out a number of sets of the puzzle (which could be duplicated photographically) to several isolated collaborators and then adding up their results after a specified period. Though this method would allow the enlistment of an indefinite number of helpers, it would bear no appreciable results. The only way to get the job finished quickly would be to get as many helpers as could conveniently work at one and the same set and let them loose on it, each to follow his own initiative. Each helper would then watch the situation as it was affected by the progress made by all the others and would set himself new problems in accordance with the latest outline of the completed part of the puzzle. The tasks undertaken by each would closely dovetail into those performed by the others. And consequently the joint efforts of all would form a closely organized whole, even though each helper would follow entirely his own independent judgment.

Moreover, it is obvious what would happen if someone believing in the paramount effectiveness of central direction, were to intervene and try to improve matters by applying the methods of central administration. It is impossible to plan in advance the steps by which a jigsaw puzzle is to be put together. All that a centralized administration could achieve, therefore, would be to form all helpers into a hierarchical body and direct their activities henceforth from one centre. Each would then have to wait until a decision is taken at the supreme level. In effect, all participants except the one acting as the head of the organization would cease to make any appreciable contribution to the piecing together of the puzzle. The effect of co-operation would fall to zero.

We can thus see confirmed here the twofold claim that on the one hand the actions of individuals acting according to their own judgment may become spontaneously—and yet efficiently—co-ordinated to a joint task, while on the other hand subordination of the individual efforts to a central authority would destroy their co-ordination. Moreover, we can see clearly adumbrated the applicability of this logic to the self-co-ordination of scientists in the pursuit of discovery. For this logic seems to consist simply in the extension of an unknown pattern by individual steps, under the twofold condition that each suggested new step can be readily judged as to its correctness or otherwise, and that each new step is rapidly brought to the notice of all participants and taken into account by them when they make their own next move.