PROBLEM SOLVING: SOME THOUGHTS ON METHODS AND PEOPLE* By DERE E. TILL /lrthur D. Little, Inc., Cambridge 42, Mass. IT is RE^SOS^BLE to assume that our supply of problems is not going to run out in the foreseeable future and perhaps the market for problem4 solvers will also remain firm. The rapid and efficient solving of problem• is, of course, the primary aim of the research worker in industry. Whethen these problems are deliberately posed, such as in the quest for new products,I or dumped in the Technical Director's lap by Sales or Production is of little[ importance. The point is they must be solved quickly and efficiently. To some extent, we take the process of problem solving pretty much forl granted. We become familiar with it during our college training and subsequently practice our own particular aspect of it so that it becomes al-[ most habitual. Recently, however, some of our staff have thought it worth[ while to review various approaches to the solving of industrial problems. Not surprisingly, a great deal of lively discussion resulted, and I have drawn freely from the ideas expressed at these informal gatherings. It is always difficult and dangerous to generalize about science, and par- ticularly scientists, but for purposes of discussion it is possible to divide ap- proaches to problem solving into three main classes: the theoretical ap- proach, the experimental approach (although this is not the only approach to use experiments) and a more recent approach called Operational Crea- . tivity. We shall refer to the idealized people who populate these classes as model-builderss experimentalists and "blue-skyers" (for reasons which will be apparent later). These classifications should not be confused with other arbitrary divisions often drawn between something called "pure" science and something else '. called "applied" science. The divisions I have just made are concerned only with method and not with motives. To compare the value of the approach of the model-builder, the experi- mentalist and the "blue skyer" would be quite pointless. It would be like comparing the relative utility of a hammer, a screwdriver and a drill. To * Presented at the May 10, 1957, Meeting, New York City. 328

PROBLEM SOLVING: METHODS AND PEOPLE 329 argue that the screwdriver is better than the hammer would clearly be ri- diculous each is uniquely suited to perform certain tasks. Such is also the case with the tools available for problem-solving. First, let us consider the theoretical approach. Here, facts are fashioned into a theory, which must be proved by a carefully designed experiment. The experiment in turn produces more facts which may make it necessary to modify the theory and thus necessitate further experimental proof, and so on. Workers in this field often use mathematical language to represent their models. The recent tenth anniversary of the tragic Texas City dis- aster is a reminder that the cause of the ammonium nitrate explosion was determined by the model-building technique. The theoretical approach is essential here--after all, one can hardly go around exploding enormous quantities of ammonium nitrate in order to get additional data. A more recent and highly successful application of this method by scientists skilled in Operations Research is the calculation of physical constants and other engineering data for complicated piping networks. Complex "plumbing" problems may now be solved on a computer, instead of by an army of en- gineers carrying out dreary mathematical calculations. In this case, a problem which has already been solved has been resolved by a far more ele- gant method. This is a familiar pattern--many current problems being handled by the experimentalists may ultimately be resolved more rapidly and accurately by the model-builder, using the evidence accumulated by the experimentalists. The model-builder is sometimes compelled to resort to the empirical ap- proach. A good example of this is in the identification of materials by IR absorption spectra. Although the theory is highly developed and the in- struments very refined, the model-builder cannot positively identify a ma- terial from theoretical considerations. He must be able to compare his spectra with that of a known and closely related, or better still, identical, compound. Scientists engaged in this sort of work are usually highly trained, articu- late people who like to talk about what they do. It is probably fair to say, however, that they are much more interested in discussing their methods than their results. The extreme case of the model-builder is the scientist who is so preoccupied with his method that he selects his problems in order to provide opportunities to use his technique. At the other end of the scale, there is a gradual, almost imperceptible, change into the predominantly ex- perimental area. For instance, who can judge where the field of the poly- mer scientist ends and that of the advanced resin chemist begins ? The experimentalists' method for solving problems is perhaps the most familiar to industry. It can accurately be described as an empirical method, although this is a dangerous word because it has become identified in many people's minds with an approach which is dependent only on experi-