There are no larger human groups nor any greater challenges to collective action theory than polities, the autonomous political communities that characterize human macrosociality. Their most recently emergent form, the nation-state, represents an especially acute problem because, in these colossal "imagined communities" (Anderson 1991), no member knows more than a tiny fraction of the rest, and yet somehow they manage to cohere and continue through time. In one guise or another, the nature of polities and the processes that propel their development have occupied anthropology and archaeology from their earliest days. Recent thought, however, dates to the 1950s and the beginnings of political evolution1 as a research field. Since then, three basic types of theory have emerged, and though each has focused primarily on the evolution of polities, none could avoid either explicitly addressing or implicitly assuming something about their nature-about why they are formed in the first place and how they are socially reproduced. From the 1950s to around 1980, when the field was still dominated by cultural anthropology, these three approaches became known as the voluntaristic (also the integrative or functional ) approach, the conflict (or coercive) approach, and the systems (or multivariant) approach. In their essentials, none of these lines of thought was original; all can be traced back to the Enlightenment or even earlier, to thinkers such as Hobbes, Locke, and Rousseau. Their claim to distinction, though, was to modernize, expand, and clarify earlier thought and embed it more securely in the ethnographic and archaeological evidence. Voluntaristic theories (e.g., Service 1975) were based on the proposition that humans form polities to capitalize on some benefit that can best or only be secured through collective action. Members might perceive a common benefit to building or extending irrigation systems, subsistence redistribution mechanisms, surveillance and defensive systems, or some other public work, and they designate or strengthen a political center in order to organize the system for the benefit of the group. Conflict approaches (e.g., Carneiro 1970) took a far darker view of political evolution and the nature of polities, presenting both as the product of exploitation. In Carneiro's circumscription theory, for instance, political evolution is pictured as stimulated by population growth and enabled by circumscribed conditions (i.e., circumstances that tie people down by making it difficult or undesirable for them to relocate). Polities then expand in size and become politically more centralized as one polity succeeds in conquering, incorporating, and then exploiting the labor of others. By implication, since people do not voluntarily submit to exploitation, the approach assumes that polities are held together not by collective interest but by coercion: a minority military elite oppressing a majority under conditions that prevent the latter from escaping the oppression of the former. Systems theories sought to combine the voluntaristic and conflict approaches. One version proposed that, at some points in the trajectory of their political evolution, systems develop through voluntaristic processes, while at other points the processes are coercive. Another version proposed that both processes operated simultaneously (for a summary, see Cohen 1978). By the 1980s cultural anthropology had become more interpretative and had largely surrendered the study of political evolution to archaeology. The voluntaristic approach was transformed into managerial (or adaptational ) models (see review by Diehl 2000). To the extent that human behavioral ecology and "collective action" models address large-group cooperation and political centralization, they too fall into this camp (e.g., Blanton and Fargher 2008; Shennan 2008). The conflict approach became the political (or exploitative) model (e.g., Arnold 1993; Dye 2009; Hayden 1995). As for the systems approach, archaeologists had already sketched its elementary forms (e.g., Flannery 1972; Rathje and McGuire 1982; Wright 1978), and more recently it is apparent in appeals to combine heterarchy with hierarchy (Crumley 1995) or "bottom-up" with "top-down" approaches (e.g., Carballo, chapter 1). The names may have changed, and there have been many elaborations and modifications in the details, but the core approaches remain largely the same. Criticisms of these three approaches are numerous. The voluntaristic approach proposes that polity members voluntarily surrender their sovereignty to a political center in return for a set of expected benefits, but it is difficult to understand in what sense individuals can be said to surrender autonomy. The conflict approach views polities as held together by force, but force generates distress in those subjected to it and is therefore an exceedingly costly means of control. Systems theories claim to synthesize voluntaristic and conflict approaches, but they fail in practice to explain in any detail how such starkly opposed views of groups and processes might be reconciled. These theoretical differences persist, I think, because we have failed to recognize that terms like polity, chiefdom, state, and the like actually collapse together two political phenomena that are better kept apart. The issue is neatly summarized in Giddens's (1985: 17) observation that the term state carries two quite different meanings in ordinary language. It can refer to the "apparatus of government or power," as in "The State." Or, it can refer to "the overall system subject to that government or power"-the totality of individuals who make up a body politic. More generally, the term polity gets applied to both a political center (albeit sometimes little more than an elder or two in the case of a band) and a political community. The problem is that voluntaristic and conflict approaches fail to distinguish these two phenomena, and to confuse matters further they implicitly focus on different aspects. Voluntaristic arguments concern themselves primarily with the polity qua system or political community. While they do not ignore the polity qua governing apparatus or political center, they analytically reduce it to the collective goals of the group, representing it as the means by which the collective benefit is realized. Conflict approaches, by contrast, focus on the polity qua center, analytically reducing the polity qua community to the consequences of its military or otherwise coercive actions. As a result, the two approaches have always talked past one another. As for the systems approach, it too is unaware of the problem, and as a result necessarily fails to amalgamate them. In this chapter, I reexamine the whole issue of polities and their evolution in light of this paradigmatic confusion. My central point is that the polity qua governing apparatus and the polity qua group are different phenomena, and the processes that govern their emergence and development should therefore be analyzed as such. I shall argue that polities as political communities are and were almost everywhere defensive organizations, aimed at securing the collective benefits of mutual protection against enemy attack. In contrast, polities as political apparatuses are hierarchies of power relations created, reproduced, and extended to advance elite agendas. Now, to assert that the vertical and horizontal dimensions of polities are distinct in theory is not to say that they are unrelated in practice. As both a central elite and a defensive group, a polity comprises knowledgeable and capable agents with interests, and to the degree that these interests intersect, the two dimensions will affect one another. Mutual interests will generate cooperative linkages, opposed interests will produce conflict, exploitation, and/or oppression. And in this articulation of interests, I argue, we have a basis on which to reconcile voluntaristic and conflict approaches to polities and political evolution. I then briefly sketch how this perspective can shed light on the constitution of both empires and failed states. Within the confines of a single chapter it is difficult to develop an argument about the political history of the world without seeming crassly reductive, and it is even more of a challenge to provide adequate empirical support. To demonstrate that the analysis is not without empirical foundation, however, I draw on evidence from both ends of the political complexity spectrum. At one pole, I use data from contact-era New Guinea, an ethnographic theater that was home to "band," "tribal," and even "petty chiefdom" polities, the political forms conventionally viewed as some of the least complex on earth. At the other pole, I shall consider nation-states, in particular the United States of America, a nation-state whose formative rationale is perhaps better documented than any other. Against the charge of reductionism, I argue in conclusion that attempts to identify major imperatives in the formation and evolution of polities are not to be confused with prime-mover or deterministic models of human history. © 2013 by University Press of Colorado. All rights reserved.