Anthropology, if it is to survive in its present form, cannot afford to carry out empirical analyses of social change, then go on to argue for improvements whose only basis is the uncritical, often unconscious, acceptance of one among a number of possible value/ideological positions. It is sad to watch anthropology, indeed social science in general, disabling itself from active participation in policy formation by choosing or supporting one set of goals over others with nothing but advocacy to back it up. The choice itself is justifiable. Ultimately we must accept or oppose many social goals in our own society and others if we are to participate in programs to achieve change and improvement in particular situations. The problem emerges when we do so with little or no analysis of the goals and purposes themselves. these purposes defensible ethically, and in real world terms what will they produce? Are they feasible? Will they inhibit or destroy other desirable purposes?'(Cohen 1985: 253). These are the issues raised, and avoided, by the Collins (1986) article on smallholder agriculture in South America. The article argues convincingly that South American conditions create incentives and constraints forcing smallholders to shorten fallowing and to adopt monocropping. This is correlated with soil erosion and deterioration, declining yields, loss of peasant-owned farms, and increased social differentiation. Farms are then taken over by larger farmers. A few smallholders in expanding operations, or they move into other economically viable activities. Most, however, fail and become impoverished. To improve the situation, Collins concludes that these forces must be reversed and smallholder farming be able to succeed if there is sufficient political commitment (1986:8). Exactly how this can be accomplished is left for further research, but the author hints at better producer prices, credits, and improved marketing among other stimulants and constraints so that the need for off-farm labor will not be so great for small farmers. Supposing, and I have no reason to doubt it, Collins' analysis is valid and replicable in its findings across wide areas of tropical South America. Possibly it is even applicable to tropical agriculture elsewhere in Asia and Africa. Then what? Should we accept her policy propositions as well? And if not, why not? The answer lies in our willingness to be as careful and probing of ends and purposes as we are of ongoing social processes. Without disciplined analysis of alternatives and outcomes it is naive, possibly worse, to believe we have the right to say what interventions should be imposed on the communities and societies we wish to help. And what is our culpability should the advice we give lead to a worsening of suffering and impoverishment-as I believe has occurred in parts of Africa. Are we innocent and blameless if we help people to achieve the wrong goals and outcomes for the right reasons? Only, in my view, if we have carefully and dispassionately examined the evidence for all possible options and then advised about which goals are best. I have very little if any expertise about South America. However, using Collins' own materials it is clear that a wide range of options are possible. (1) We can accept Collins' own position. This involves trying in ways yet unknown to save smallholder farming, stop rural differentiation, stop takeovers by larger farmers, and stop the environmental degradation resulting from smallholder poverty. (2) We can opt for a laissez-faire approach, and allow the processes described to continue unabated. This will produce a rural landless peasantry who are impoverished, probably used as cheap labor, but will slow down or stop environmental degradation once land becomes a scarce resource requiring constant investment in nutrients and other forms of more scientific land preparation and care. (3) We could opt for a combination of each of the above, use special programs for the most conscientious smallholders, and allow the others to become farm labor and urban migrants. This would help the environment and slow down impoverishment. Except for the second laissez-faire position, these policy options involve extraordinarily far-reaching changes in the scope of governmental and bureaucratic power and authority over the forces and relations of production. Can they succeed? Would any of them be acceptable without massive changes in society involving the possibility of internal violence or external interference or both? And if accepted and implemented, would smallholders actually give up off-farm labor to pay greater attention to the enhanced efficiency of farm work without greater coercive organization of society by the state? There is a large array of regular and chance determinants that could vary all of the possible outcomes so enormously that any of the positions leads to unknown-and unknowable-outcomes. The responsible recommendation is therefore quite differ-