The role of knowledge codification in the emergence of consensus under uncertainty: empirical analysis and policy implications

Abstract

The aim of the article is to empirically analyse the main factors, which allow the emergence of consensus. Moreover, we raise the question of the nature of the codified process involved which seems to be too reducing to promote knowledge flows within the community and, therefore, to contribute to learning processes and better choices about the future. Our analysis relies on a prospective investigation based on a Delphi type study carried out by BETA in 1994.

Introduction

The increasing complexity of the relations between technologies and economic problems combined with social pressure implies that governments are compelled to make choices, especially under radical uncertainty (non-stochastic uncertainty). Most of the time, these choices concern science and technology policy, which originate in increasing environmental, economic and social problems (Grupp and Linstone, 1999). In this context, experts are increasingly requested to bring knowledge and/or scientific legitimacy (Weil, 1998). In order both to bring some responses to socio-economic problems and to reduce the potential responsibility of governmental decisions, the policy-maker encourages the emergence of a consensus among experts.

Foresight approaches are a method which makes it possible to choose and to understand the role, the nature and the impact of science and technology policy portfolio and also to obtain a consensus among experts.¹ Nevertheless, radical uncertainty linked with the hypothesis of bounded rationality (Simon, 1982) implies that the actions of social systems cannot be predicted in terms of natural laws, and that future events cannot be determined by extrapolation, but are shaped by communities. In this way, traditional methods of forecasting using probabilistic predictions based on today’s knowledge base seem not to be relevant and, therefore, qualitative analyses is preferred.

The Delphi method appears as the favoured tool of modern foresight in many countries. One important characteristic of this method (in comparison with the other methods using expert groups) is that it is necessary to codify knowledge about the future in order to transmit it among experts. One first basic principle used in Delphi is that experts judgmental predictions should be made independently because of evidence that group pressures can spoil accuracy (Armstrong, 1999). Therefore, by providing the individual group members with the opportunity to express their opinions and judgements privately through the use of anonymous questionnaires, social pressure from dominant or dogmatic individuals should be avoided. A second basic principle of the Delphi method is that it is a method for structuring a group communication problem, which makes it possible to exhibit consensus or dissent about technological, economic or social subjects. In order to reach this consensus, iteration and feedback procedures are used and experts can revise their judgements by means of successive iterations of a given questionnaire with additional information such as the median value of the opinions of their anonymous colleagues. Generally, this method seems to be a better procedure for obtaining accurate judgements than standard interacting groups (Ayton et al., 1999) and moreover, this method provides the possibility to obtain a consensus. The interest of such an investigation is to provide also a ground of observation of mimetic behaviours insofar as two rounds characterise the Delphi inquiry. At the second round, experts can either converge on the general opinion (expressed by the median of the answers concerning the date of realisation of innovation), that emerges from the first round, or stick to and consolidate their initial position that differs from the consensus.

The aim of our article is to empirically analyse the main factors which allow the emergence of consensus. Moreover, we raise the question of the nature of the codified process involved which seems to be too reducing to promote knowledge flows within the community and, therefore, to contribute to learning processes and better choices about the future. Our analysis relies on a prospective investigation based on a Delphi type study carried out by BETA during the year 1994 (BETA, 1995). The object was to collect the opinions of experts of the public and private sectors on the evolutions of technology and its possible rupture in the next 30 years.

In the questionnaire, the experts answer a series of questions for every item belonging to the general theme of their expertise. Each expert has various levels of knowledge following the items, and one of the questions consists precisely in this self-evaluation (some “experts” are more “competent” for a given item). Other questions may concern the importance, the difficulty, the timing, etc. of the subject, if items are about potential innovations. The investigation is divided into 15 fields: materials, electronics, life sciences, elementary particles, marine science, raw materials, energies, environment, agriculture, manufacturing activities, town planning, communication, space, transport and medicine.

The assumption that we formulate is that the (self-declared) knowledge of the experts is the most important explanatory factor in order to understand mimetic behaviour and also the emergence of consensus. More precisely, we make the assumption that experts with a limited knowledge of a given subject are more inclined to concur in the general opinion while adopting a rational mimesis behaviour. To validate this assumption, we propose first a descriptive statistical analysis of the sample and secondly, we test an econometric model in which several explanatory variables (including knowledge) are also analysed. They are variables expressing the nature of the expert such as his/her age (reflecting the experience of the individual), his/her institutional membership (company or public organisation) and his/her membership or not of an R&D department. In order to apprehend the nature of the consensus, we also propose a conceptual analysis clarifying the importance of knowledge in the experts judgement.

In a first part, we present the conceptual framework of our empirical analysis. We present in a second part an interpretation of the results, descriptive statistical results and those derived from the econometric analysis. In the last part, we present some policy implications according to our results.