The constructivist movement in the social sciences (Berger & Luckmann, Gergen) argues that social institutions are the constructions of a community of human agents. This implies that people possess the power to radically change those institutions. This raises what may be called the values problem in constructivism, which, in its simplest form, is put in the title of the present chapter: What should guide reality construction? What principles or values should guide our construction of economic institutions, foreign policies, roles for the sexes, schooling, care for the sick? This values problem is accentuated in the sort of social research that is informed by constructivist principles. While the objectivist researcher of yesteryear could ignore the value implications of her research activities, the constructivist is forced to acknowledge her influence on the research situation and her role in co-constructing the reality of her hosts. This acknowledgement implies responsibility and, hence, a personal set of values. If the values problem is central in constructivism as a general outlook on the world, it is even more so in the practice of constructivist research, since the values aspect has been actively suppressed in the scientific community, whereas in the general community it is at least acknowledged as an issue. The values problem in constructivism has been addressed by a number of writers in second-order cybernetics, who offer reflections on ethics and values that are quite tentative and call for further elaboration. The remainder of the chapter is an attempt to provide a broader context for a constructivist discussion of values and ethics. To do this I discuss the concept of the "implicate order," an ontological framework proposed by the theoretical physicist David Bohm (1980b). As will be shown, this ontology implies no simplistic realism; it is an ontology highly compatible with constructivist epistemology. Here, it serves to provide inspiration for the formulation of two normative principles, to be called the unity principle and the diversity principle. These will be combined into a simple model that describes some common ethical positions (relativism and absolutism) as distorted or degenerate cases of an ideal case, characterized by unity-in-diversity. This model is offered as a framework for a discussion of the values problem in constructivism, particularly in social research.