The social world consists of facts and the perspectives in which we see them. The number of possible perspectives is always very large; the number actually in use in the models presented by social sciences at any time is bound to be relatively small. All the social sciences have passed through periods of consensus during which one 'paradigm' seemed paramount, and subsequent periods of controversy characterized by a multitude of paradigms espoused by rival schools. Nevertheless the number of paradigms in actual use is always an exiguous proportion of those possible.
All analytical thought requires abstraction, but the more inclined we are to concede this need, the more apt we become to forget that that which has been abstracted from may become important, if not at present, perhaps at a later stage of our enquiry. In the economic thinking of our age this tendency has been greatly strengthened by the well known proclivity of model builders to 'close' their models and thus to impart to the relationships described in them, which in reality, often are relationships between contingent facts, an appearance of 'necessity' that seems to lend them a higher methodological status. This is usually done by introducing an array of restrictive assumptions.
Some methodologists believe this to be a necessary, indeed a welcome, step in expanding and unifying knowledge. 'A lot of effort is expended to assure that the theorems proposed will be “necessarily true” once the language is only used “correctly”.' (Leijonhufvud 1976:8). It is not surprising that the apparent necessity of the theorems proposed in our models soon begins to