A theory that magnifies the role of decisions, or free selection from amongst equally possible alternatives, in order to show that what appears to be objective or fixed by nature is in fact an artefact of human convention, similar to conventions of etiquette, or grammar, or law. Thus one might suppose that moral rules owe more to social convention than to anything imposed from outside, or that supposedly inexorable necessities are in fact the shadow of our linguistic conventions. In the philosophy of science, conventionalism is the doctrine often traced to Poincaré that apparently real scientific differences, such as that between describing space in terms of a Euclidean and a non-Euclidean geometry, in fact register the acceptance of a different system of conventions for describing space. Thus one can no more ask whether Euclidean geometry is true than whether the metric system is true. Poincaré did not hold that all scientific theory is conventional, but left space for genuinely experimental laws, and his conventionalism is in practice modified by recognition that one choice of description may be more convenient than another. More recent holistic approaches to theories and to meaning find it impossible to separate out the objective or empirical from the conventional or linguistic (see also Quine). The disadvantage of conventionalism is that it must show that alternative, equally workable conventions could have been adopted, and it is often not easy to believe that. For example, if we hold that some ethical norm such as respect for promises or property is conventional, we ought to be able to show that human needs would have been equally well satisfied by a system involving a different norm, and this may be hard to establish.