[Help] [Aide] [Up]

Science Tribune - Article - December 1997


Looking back on past forecasts : How experts in chemical reactivity saw the future 10 years ago

J.C. Doré1 and M. Karsky2

1URA 401 CNRS, Muséum National d'Histoire Naturelle, 63 rue de Buffon, Paris 75005, France
E-mail : dore@mnhn.fr
2KBS, 287 rue St. Jacques, Paris 75005.
Website : http://myweb.worldnet.fr/~kbs-fr
E-mail : kbs-fr@worldnet.fr

In 1986, we initiated a study (a) on how chemists from 4 countries (France, Great Britain, Germany and Japan) perceived ongoing and future developments in the area of chemical reactivity. Information from face-to-face interviews was used to generate maps of keywords describing various facets of the subject. The location of the key-words in the maps reveals that, according to their country of work, scientists had different views at the time of the interview and made different forecasts. The methodology we used can easily be transposed to other fields and might prove an interesting way of performing objective analyses of data obtained in surveys. However, we have not been able to check which forecasts made by the chemists have proven most accurate and which were false and now, 10 years later, call upon Web browsers to provide us with their personal appreciation of chemical reactivity today.

Decisions, and also forecasts, are based on three types of information :

(i) Numerical data. These are scant, in principle objective but not always trustworthy, and often of poor information content. The success of numerical data is partly due to the ease with which they can be handled by computer.

(ii) Text. Text is encountered more frequently than numerical data; it is richer in information but more subjective and not easy to quantify.

(iii) Speech. By far the most common vector of information, speech is redundant, fleeting, disorderly etc but, if the noise were filtered, it would tell us most about the thoughts, intentions and wishes of people. Oral speculations and forecasts are rarely put down in writing.

The aim of our study

We obtained oral information during interviews with specialists in chemistry, all of them knowledgeable about what was going on in their own field through publications, congresses, meetings etc... However, as they themselves admitted, they were not always aware of the latest developments in 'neighbouring' related subjects, nor had they any clear idea of how advances in these subjects might influence their own work. This was the "fuzzy "world we wanted to analyse. If we take the well-worn image of the forest and the trees, we wanted an aerial global view of the forest, of the thickets (dense research areas), clearings (gaps between different areas of chemistry), intertwining branches (concepts common to scientists specialising in different areas), and new undergrowth (emerging topics and techniques). We were not particularly interested in a description of any of the individual trees.

Data collection and analysis

We interviewed some 50 chemists from each of 4 countries (France, Great Britain, Germany and Japan) on the subject of chemical reactivity. Because we were seeking new information (b), we used a broad open questionnaire of 13 questions which was completed on the spot by the interviewer using keywords for the concepts mentioned. We obtained information on :
(i) current research topics and ongoing work
(ii) their views on the evolution of concepts, themes, methods during the next 2 to 3 years
(iii) their long-term forecasts (5 years and beyond)
(iv) their expectations relating to advances in fields that might impinge on their own work.

Oral data being noisy, we filtered them as follows :
(i) Extraction of all keywords from the completed questionnaires with no regard to their frequency, finality or to the links among them. Up to 3000 keywords were recorded per country.
(ii) Attribution of each of these keywords to about 50 generic keywords that describe the main concepts, themes and techniques to which reference was made during the interviews. (The choice of generic keywords was validated by chemists not taking part in the study.)
(iii) Multidimensional analysis of the matrices describing the frequency with which key concepts were mentioned by the scientists [50 generic keywords x 50 interviewees]. A factor analysis based on chi2-metrics (correspondence analysis) (c) and thus suitable for the analysis of occurrences was used to highlight the relationships among the generic terms according to time-period (present, short-term future, long-term future) and according to country.

How chemists from each of four countries perceived chemical reactivity 10 years ago

The factorial map in Figure 1 displays 90% of the information on the views of scientists from different countries on what were, according to them, the main trends in chemical reactivity at the time of the interviews.

Fig.1. Trends in chemical reactivity according to country

The views of French chemists stand out in isolation on the left of the map; they were opposed to those of Japanese scientists on the one hand (top right-hand quadrant) and to German and British scientists on the other (bottom left-hand quadrant). The outlying shaded zones indicate what were the main generic keywords highly specific to each nation : The French put the emphasis on small molecules and collisions, the Japanese on large molecules and bioelectronics, the British and Germans on computer sciences, theory and spectroscopy. There was a central consensus area for all four nations and three areas of common interest between pairs of nations.

The global perception of trends in chemical reactivity by chemists from all four countries

Figures 2A to E illustrate how chemists from all four countries, taken together, saw the future. Each panel shows the location along a non-linear time-scale of keywords relating to five aspects of chemical reactivity (d) : trends in concepts, theoretical tools, chemical disciplines, application areas (molecular support), physico-chemical and analytical tools. In general, keywords close to the "present" referred to items not expected to undergo any spectacular development, those near the "short-term" point were approaches about to become standard or 'fashionable', and those close to "long-term" described the scientists' forecasts.

Figs 2A & B Forecasting trends in concepts and theoretical tools (combined responses of all 4 countries)

When considering concepts (Figure 2A), issues that the scientists thought might be important in the future were active sites, molecular recognition, interactions and chaos. On the other hand, they did not project chemical stability, functionality, conformational analysis, etc into the future. The concepts of structure, selectivity and reactional dynamics were in a half-way-house.

All the keywords describing theoretical tools (Figure 2B) were part of the future. The most 'futuristic' were the most recently developed approaches such as fractals, expert systems, and computer-assisted design, but also databases. Keywords of a more general nature (theory, modelisation, computerisation) were not projected as far ahead. Quantum chemistry was closest to the present.

Fig 2C Forecasting trends in disciplines (combined responses of all 4 countries)

The distribution of the chemical disciplines (Figure 2C) along the time-axis revealed that aqueous chemistry and also solid state chemistry were long-term predictions whereas the recent disciplines of space chemistry and intrastellar chemistry were part of the short-term future together with several more traditional disciplines (radio/electro/surface chemistry and biochemistry). The other disciplines - whether specific such as photochemistry or general such as ion chemistry - were quoted as issues of primarily contemporary interest. Overall, there seemed to be a shift in focus from small to large molecules with time.

Fig 2D Forecasting trends in work on experimental entities (combined responses of all 4 countries)

The time distribution of the experimental entities (reactive species, atoms, molecules, materials ....) that characterized the above disciplines (Figure 2D) matched that of the disciplines themselves. The future was to focus on macromolecules (enzymes) related to biological systems and low energy reactions rather than on small entities (radicals, ions), high-energy reactions, and the classical physical views of oxydation and hydrogenation. Amorphous materials, semi-conductors, catalysts, plasma were immediate concerns projected into the short-term future.

Fig 2E Forecasting trends in techniques (combined responses of all 4 countries)

Of the techniques, three belonged decidedly to the realm of the future (laser, mass and fast spectrometry), three were in widespread use but likely to be superseded (fluorescence, polarimetry, chromatography), and the remainder were expected to be still in current use in the short-term future.

We also analysed trends for each country. These are represented in factorial maps with three attractor poles corresponding to the present (To), short-term (T1) and long-term (T2) (d) (Fig. 3).

Trends in chemical reactivity according to French chemists

A 2D-map carries more information than a single axis. The keywords in Fig. 3 are dispersed within the map according to their differential relationship to each time-period.

Fig 3 Forecasts made by French chemists

Location A includes : Adsorption, hydrogenation, irreversible phenomena, dissociation, conformational analysis, cross section, catalysis (homogeneous), energy transfer, structure reactivity, coordination complex, energy distribution, relaxation, chemical waves, chirality, excited state, geometry of reaction, topological methods, intermolecular interactions, pot. energy surface, ionization potential, charge transfer, kinetics, oxydation, fluorescence, structural chemistry, time evolution, photodissociation, photo ionization, support, radical react., organic conductors, organic synthesis, synthesis, clusters, zeolites.

Location B includes : transition metal chemistry, concert. processes, selectivity, molecular graphics.

Location C includes : X-ray analysis, biology, mass spectrometry, expert systems, biological systems, gas-surface reaction, ultra-fast laser, modelisation, electrochemistry, fractals, interstellar chemistry.

Keywords nearest to T0 : These were words associated with the present (e.g. absorption, hydrogenation, zeolites, clusters, etc....). In general, they were concrete, precise terms referring to techniques and fields of application.

Keywords closest to T1 : In 1986, molecular graphics techniques had only just emerged in France and no specific development of the technique was foreseen.

Keywords around T2 : These refer to new paradigms or to new ways of using old paradigms. They include : interstellar chemistry, expert systems, fractals, modelisation, X-ray crystallography and mass spectrometry (for the determination of the structure of biological macromolecules), gas-surface interface.

Keywords within the triangle formed by the three time-points are timeless, the blue-chip values of research into the understanding of chemical reactivity : collision, theory, catalysis and catalyst, quantum chemistry, spectrometry.

Keywords between T0 and T1 : These are approaches whose widespread use appears imminent : heterogeneous catalysis, the study of reaction intermediates.

Keywords between T1 and T2 refer to more concrete and realistic items than those associated with T2 alone. They include sophisticated techniques such as fast spectrometry, computer-aided design, NMR spectrography, but also database use, solvent effects, and the study of active sites.

Keywords between T0 and T2 belong to a paradoxical area covering items that are about to undergo deep changes or make new adepts. In France, these are chaos, photochemistry, reaction dynamics, physico-chemical techniques in biochemistry.

The keywords belonging to a family can be linked to form a network. Thus, for instance, one notes a progression in time from molecular graphics to computer-aided design to active site and modelisation.

Differences in the emphasis of the forecasts made by each country

By comparing the factorial maps for each country, we were able to pinpoint several differences in the ways they viewed trends in chemical reactivity.

Computer and computerisation : For French scientists, these terms lie along the T1/T2 gradient whereas for British scientists they are closer to T0/T2 and, for German and Japanese scientists, actually within the triangle. This is a sign of a greater familiarity with computer use in Germany and Japan.

Enzymes : For Germany and Japan, as for France (Fig. 3), the keyword enzyme lies within the triangle, whereas for Great Britain, it lies far out, beyond T2.

Nuclear magnetic resonance (NMR) : NMR - insofar as it is used to understand chemical reactivity and is not just a method for structure analysis - belonged decidedly to the future in the opinion of French scientists, was in the center of the triangle for British and Japanese scientists, and closer to the present for the Germans. The interpretation of these positions is ambiguous. Either the German scientists had cottoned on to the immediate benefits afforded by NMR in the understanding of chemical reactivity before the French did, or the French had a clearer idea than the Germans of the potential of the method in years to come (e.g. for supraconductors).

Radicals and radical chemistry : Only Great Britain placed these keywords closer to T1 than to the present.

Electrochemistry : This keyword, which was projected far into the future by the French (Fig. 3), was part of the present for the Japanese. On the other hand, collision, which was part of the present for the French, was projected into the short-term future by the Japanese.

Theory was within the triangle for the French, British and Germans but well beyond the T2 pole for the Japanese suggesting, perhaps, a more pragmatic frame of mind.

Both kinetics and quantum chemistry held timeless positions within the triangle for all four countries.

The specific mention of certain keywords by one country and their complete absence in the map of another also reflects different viewpoints. Thus supraconductivity, molecular electronics, and molecular recognition were not "in-words" in France; of these three keywords, only the term molecular electronics was mentioned by the British scientists and was located close to T0; the Germans, however, placed it along the T0/T2 gradient and supraconductors beyond T2 but made no specific mention of molecular recognition. Only the Japanese mentioned all three terms : molecular electronics was within the triangle but nearest to T2, molecular recognition along the T1/T2 gradient and supraconductors way out beyond T2

A post-scriptum in 1997

The above analysis was initiated over 10 years ago. The two to three years covering the short-term predictions and even the 5 years of the long-term predictions have gone by and we should now know how accurate were the forecasts made by the chemists we interviewed and how useful this approach might be in obtaining a global understanding of the evolution of a specific area in science. Ideally, the answers to these questions could be obtained by means of a second survey of the same scientists but, alas, this is not possible.

We are therefore asking Web browsers with expertise in chemistry to tell us what they think about these forecasts of 10 years ago. Do the factorial axes and maps yield information that is spot on or did the scientists make several misjudgements ? Has the notion of chemical reactivity evolved differently in the four countries ? Are there aspects of chemical reactivity - new keywords - that the chemists completely ignored ? Has the field progressed faster or slower than expected ?

We are looking forward to hearing from you.


(a) The study was performed under the auspices of the Department of International Relations of the CNRS headed by Jean-François Miquel. The authors are extremely grateful to Isabelle Suzuki for carrying out the interviews of Japanese chemists and to Eleanor Ardeaillan and Janos Angyan for technical support.

(b) Interviews were held in English with British and German scientists, in Japanese with Japanese scientists, and in French with French scientists.

(c) For details of correspondence analysis, readers may refer to textbooks (Correspondence analysis in the social sciences by J. Blasius & M. Greenacre, Academic Press, 1994, 352 p. or to the following website http://pascal.math.yorku.ca/Cologne/).

(d) The projections onto the main factorial axis of the two sets of variables - time-periods (present, short-term future, long-term future) and key-words - was obtained by a correspondence analysis of the matrix [time x key-words] for data obtained from all interviewees regardless of country of work.

(e) The projections onto the two first factorial axes of the two sets of variables - time-periods (present, short-term future, long-term future) and key-words - was obtained by a correspondence analysis of the matrix [time x key-words] for data obtained from all interviewees of each country.