12
Network Strategy and War
Peter N. Peregrine And Carol R. Ember
One of Richard Blanton’s most influential papers is the 1996 “A Dual-Processual Theory for the Evolution of Mesoamerican Civilization.” There, Blanton and colleagues argue that there are two basic strategies that leaders follow, to varying degrees, to maintain authority: corporate strategy and network (or exclusionary) strategy. Leaders following a corporate strategy attempt to build a power base by developing and promoting activities that reinforce the corporate bonds tying members of the polity together. A common corporate strategy is, for example, to mobilize goods from across a polity for large public rituals or construction projects that bring members of the polity together in corporate-affirming activities. Leaders following a network strategy attempt to build a power base by controlling access to networks of exchange and alliance both within and outside the polity. Thus a network strategy is one in which leaders attempt to monopolize sources of power, while a corporate strategy is one in which leaders attempt to share power across different groups and sectors of a polity.
The paper has been influential because it provides a way of understanding variation in the nature of Mesoamerican polities, some of which appear to have obvious, self-aggrandizing leaders, and others that appear “faceless” despite clear evidence of centralized leadership. The value of this perspective has appealed to many scholars, and it has been widely applied to other prehistoric polities (e.g., Mills 2000; Butler and Welch 2006). Despite its influence, the “dual-processual” paper had two flaws. First, it did not adequately clarify that the “dual processes” of political strategy formed a continuum rather than two types of political strategy, so that other scholars often took them as reflecting cultural “types” rather than as expressions of variation in political strategy (e.g., Yoffee 2005:177–179). Second, the paper did not explain how variation along this continuum changed over time.
In recent work the authors of the dual-processual paper have attempted to address these flaws by clarifying the idea that the “dual processes” of corporate and network strategies form a continuum from one more inclusive of citizen participation in the polity’s political action and decision-making (corporate strategy) to one more exclusive of such participation (network strategy) (Blanton and Fargher 2008; Peregrine 2001, 2008). The first author of this chapter has also attempted to demonstrate variation on the corporate-network continuum over time; to develop an extended multivariate approach to political strategy; and to argue that balancing stability and the “cost” of implementing political strategies are one reason that variation occurs (Peregrine 1998, 2003, 2009, 2012). In this chapter, we suggest that variation on the corporate-network continuum might be related to socialization for mistrust, unpredictable natural disasters, and external warfare (see also Earle, chapter 12, this volume). And we explore the mechanisms that might underlie these connections.
Ember and Ember (1992a) found that both socialization for mistrust of others and unpredictable natural disasters are independent predictors of warfare frequency. Peregrine wondered whether xenophobia might be more common in network-oriented polities, as leaders might encourage xenophobia to prevent individuals from seeking contact with external polities. While socialization for mistrust (a variable coded by Barry and colleagues [1976]) is not the same as xenophobia, one might presume that if children are told to mistrust others in the community, they might develop more generalized mistrust. Using the Embers’ data set, and a measure of the corporate-network continuum of political strategy developed by Bramm (2001), Peregrine (2009) found evidence to support the idea that network-oriented polities socialized for mistrust more frequently than did corporate-oriented societies, and had more frequent external war. By “external warfare” the Embers meant warfare taking place outside of the culture or linguistic group (Ember and Ember 1992a, 1992b). This means that one chiefdom, for example, fighting another chiefdom within the same linguistic or cultural group would not be considered an example of external warfare, but one chiefdom fighting another speaking a different language or being part of a different cultural group would be considered external warfare.
Following upon that work, Peregrine also wondered whether, given Ember and Ember’s (1992a: 256) argument that “fear of future economic problems (rather than current problems) is the major motive for going to war,” the relationship between warfare and the “fear of future economic problems,” as measured by the presence of unpredictable natural disasters, might be associated with network-oriented polities because of their leaders’ emphases on the control of external resources. This chapter provides evidence that network political strategies are related to unpredictable natural disasters and warfare, but in a more complicated way than initially expected.
Data and Methods
The research presented here begins with data coded by Carol R. and Melvin Ember for their study of the conditions favoring warfare. The Embers employed the Standard Cross-Cultural Sample of 186 societies, which provided them not only with a large and relatively well-documented set of cases, but also allowed them to use variables coded by other scholars for the same sample (Ember and Ember 1992a, 1992b). The Embers coded 43 variables concerning the type and intensity of warfare, pacification (pacified societies were not used in their study), outcomes of warfare, individual and social aggression, resource problems, and sex ratio. The coding method employed two naïve coders. If the coders disagreed in their coding of a particular case, they developed a “resolved” score for that case. The Embers found that simply using resolved scores yielded weaker results than using resolved scores where coders more closely agreed in the first place, and so they dropped all cases where reliability was not that strong (reliability scores greater than 6). We follow a similar procedure here except that we only use the societies in eHRAF World Cultures (http://ehrafworldcultures.yale.edu).
To ensure our data selection and statistical procedures matched the Embers’, we first replicated their findings to make sure that the results were still significant in our subsample. The Embers employed a carefully selected subset of their coded cases in their analyses (dropping island societies, pacified societies, and cases with poorer reliability), using only 30 cases in their final analyses. Our procedures were slightly different here in that we did not dichotomize the natural disasters variable because we wanted to maximize variation. They also used a different statistical package (Systat) than we did (SPSS). We found that our data and statistical package replicated the Ember’s results satisfactorily (see table 12.2a).1
Once we replicated the Embers’ results, the first author coded five ordinal variables, each focused on a separate facet of corporate/network strategy. These five variables were then summed to create an interval scale of corporate/network strategy, with more corporate-oriented polities having lower scale scores and more network-oriented polities having higher scale scores. The codebook is presented as Appendix 12.A, and the raw data as Appendix 12.B. While the first author did the coding by himself, he followed the basic strategy identified by the Embers. In their original study they found that cases where the information did not allow for a clear score (i.e., where the coders disagreed) added “noise” to the data, and they dropped them from their analysis. Following that idea, the first author coded only cases where the information implied an obvious score. Because of this selectivity, and because the first author only coded those cases with primary source material available in eHRAF World Cultures, only 11 cases are included in the analyses that follow.2
Results
First we examine whether unpredictable natural disasters predicts more network-oriented polities. The bivariate Pearson r is .544, p < .04, one-tailed, which is consistent with the hypothesis that network-oriented polities are more common where there are conditions of unpredictable natural disasters.3 Figure 12.1 shows a box plot of the relationship. The second hypothesis is that there would be more xenophobia in network-oriented polities. As measured indirectly by socialization for trust, the hypothesis is not supported, and for reasons we cannot explain (r = .365, p < .27, two-tailed). We know from the Embers’ study that unpredictable natural disasters are related to more warfare, as is socialization for mistrust (both of them presumed causes of warfare). Recall that Peregrine (2009) found that network-oriented polities had more external warfare.
Figure 12.1. Relationship between network strategy and natural disasters.
How might these variables relate to each other? It seems plausible to us that network-oriented leadership may be adopted in situations of crisis. Researchers examining the psychology of survival have found that network-style, authoritarian leadership is common in the early stages of a crisis, and their presence often leads to a group successfully overcoming that crisis. As explained by Leach (1994:140–141), “The initial leader will usually be authoritarian, he will be decisive and will lead by example . . . an authoritarian, military style of organization is not only acceptable but may even be welcomed in the initial stages of a disaster.” A network-oriented leader, who controls the polity unilaterally or with a small cadre of peers, is well-positioned to respond quickly and decisively to a crisis. In polities with an ongoing threat of unpredictable natural disasters, such a leader might be desired by the members of the polity, and might be able to maintain power much more readily than a leader employing a corporate strategy. Being attacked by others might also increase the “crisis mode” of a society. So warfare, particularly external warfare, might increase the likelihood of network-oriented societies.
Table 12.1 shows a multiple regression analysis with network-oriented societies as the dependent variable and unpredictable natural disasters and external warfare as independent variables.
Table 12.1. Predictors of network strategy
Beta | Significance | |
---|---|---|
Constant | .181 | |
Natural and Pest Disasters | .427 | .412 |
External Warfare Frequency | .182 | .722 |
n = 9 | ||
R = .579 | ||
R2 = .336 | .239 |
The results are not statistically significant, although the beta values suggest that natural and pest disasters is the stronger predictor of network strategy in the model. But we have implied above that a network-oriented polity might also deliberately undertake more warfare, particularly external warfare, to defend external resources. If this is so, a network-oriented polity not only might become more likely with warfare (as we discussed above), but it also might increase warfare, particularly external warfare. (This would be a feedback loop.) To test this idea, we add the network variable to the Embers’ original model that has warfare as the dependent variable (Ember and Ember 1992b). (The Embers’ model with the societies in eHRAF World Cultures is shown in table 12.2a for comparison purposes.)
Table 12.2a. Predictors of overall warfare (Embers’ model)
Beta | Significance | |
---|---|---|
Constant | .000 | |
Socialization for Trust | –.407 | .022 |
Natural and Pest Disasters | .312 | .074 |
n = 29 | ||
R = .522 | ||
R2 = .272 | .016 |
The result shown in table 12.2b suggests that this expectation may be correct.
Table 12.2b. Predictors of overall warfare using Ember and Ember predictors
Beta | Significance | |
---|---|---|
Constant | .043 | |
Socialization for Trust | -.563 | .049 |
Natural and Pest Disasters | .435 | .123 |
Corporate-Network Scale | .411 | .167 |
n = 11 | ||
R = .823 | ||
R2 = .677 | .038 |
Not only do all the independent variables have high beta values, but the overall model predicting warfare is better when we add network-oriented strategy (the multiple R is .82 in table 12.2b compared to the multiple R of .52 in table 12.2a). If we make the dependent variable external warfare, the effect of network polities is even stronger (compare table 12.3a and table 12.3b), the increase in R is even greater, and all the independent variables are statistically significant.4
Table 12.3a. Predictors of external warfare
Beta | Significance | |
---|---|---|
Constant | .041 | |
Socialization for Trust | -.285 | .077 |
Natural and Pest Disasters | .350 | .146 |
n = 25 | ||
R = .472 | ||
R2 = .223 | .063 |
Table 12.3b. Predictors of external warfare
Beta | Significance | |
---|---|---|
Constant | .010 | |
Socialization for Trust | -.707 | .003 |
Natural and Pest Disasters | .694 | .045 |
Corporate-Network Scale | .637 | .009 |
n = 10 | ||
R = .946 | ||
R2 = .895 | .002 |
How can we make sense of this? We suggest the explanation lies in the unique conditions of polities in which leaders employ a network strategy. Recall that leaders employing a network strategy not only attempt to limit the political participation of individuals, but also to restrict access to resources and knowledge from outside the polity. Thus natural and pest disasters, which result in disruptions to external communication or resources, would have a much stronger impact on network-oriented polities than on corporate-oriented ones. To better control those external resources, network-strategizing leaders might go to war with the external group. Of course, such war could take place within a single group, as in the case of peer-chiefdoms given above, but it might more often occur with external groups, particularly in states.
At the beginning of this section we suggested that a crisis is likely to push toward more authoritarian leadership. But there is a catch. Network strategy is not simply authoritarian, but is based on controlling access to power and resources. The authoritarian nature of network strategy might be beneficial in the initial stages of a crisis, but what about later? Here, according to Leach (1994:140), more corporate-style leadership will be preferred: “The later type of leader will be one . . . who will work with the rest and will organize and minimize differences amongst the group”; in other words, a corporate-style leader. Thus, while a network strategy may work in a crisis mode, it may fail to be accepted when situations improve. We argue that the reason that network strategy improves the predictive power of the Embers’ model for external war is because it is in situations where political leaders both restrict access to political authority and use connections to other polities to maintain and legitimate their own authority that they might go to war in the face of unpredictable natural disasters in order to assure themselves access to those resources upon which their authority is based. Furthermore, leaders in network polities are in a strong position to initiate a war—they tightly control the population and are leaders who function well in a time of crisis. External war may appear an easy solution to an immediate or feared resource problem.
Conclusion
One of the limitations of the dual-processual paper (Blanton et al. 1996) was that variation on the corporate-network continuum of political strategies was not well explained. We suggested one possible explanation here—that network strategies may be adaptations to situations of resource unpredictability. Because of small sample sizes, we only have suggestive evidence to support this relationship. But we have better evidence for a more complex relationship with a feedback loop such that network-oriented polities themselves may create more warfare, particularly external warfare. We suggest that, because adding variation in the corporate-network continuum improves the predictive power for the Ember’s model of war, network orientation may itself be a factor behind increased frequencies of war. We have argued that, in conditions of resource uncertainty, leaders employing a network strategy might find going to war a good, and easily implemented, solution to a potential resource shortage.
We realize that our explanation for the increased predictive power that accompanies network strategy in our regression model for the frequency of war is not entirely satisfying,5 but we hope the reader might overlook this to accept a broader point: the corporate-network continuum of political strategy that Blanton developed to explain variation in the archaeological signature of Mesoamerican polities has explanatory power far beyond the narrow contexts within which it was developed. The idea that political strategy is actively promoted by leaders, and that those strategies impact polities in predictable ways, has had profound impact. It represents the best of anthropology’s insights, in that it both allows us to understand others and to better understand ourselves and the world in which we live (a world in which there are still corporate-oriented and network-oriented polities, acting in predictable ways). It also represents the best of Blanton’s work—strongly theoretical, but also practical, allowing for explanation of variation in the past and the present. Whether focused on cities, political strategy, or collective action, Blanton has demonstrated that anthropology has the ability to develop concepts of great explanatory power, and we hope our brief contribution to this volume has expanded the explanatory power of Blanton’s corporate-network continuum, if only in a small way.
Notes
1. If natural disasters were dichotomized as no or rare versus more, the natural disasters variable would have a higher beta than socialization for mistrust.
2. Relying only on cases in eHRAF World Cultures meant that we could not code all the cases in the Embers’ sample. We hope to do so in the future. Thus this should be taken as a preliminary study.
3. Natural and pest disasters were recoded into a three-category variable from a four-category one for this correlation and the regression shown in table 12.1 . The full variable was employed in all other analyses.
4. We must point out that the sample size upon which these analyses were performed is very small, and we must be cautious in interpreting them. However, we feel that the results are so strong, and the improvement in the regression model so great, that they cannot be ignored, despite the potential problems of such a small sample.
5. Blanton would refer to our explanation as an example of “rinky-dink theory,” replete with “ho-hum hypotheses.”
Appendix 12.A
Codebook
Column 1: Society Name
Column 2: Standard Cross-Cultural Sample Number
Column 3: Differentiation among leaders and followers
1. none
2. leaders have some privileges and/or access to resources others do not
3. leaders have extensive privileges and access to resources others do not, including special housing and sumptuary goods
4. leaders have exclusive privileges and exclusive access to special housing, resources, and sumptuary goods
Column 4: Leader identification
1. none
2. leaders are identified by treatment or appearance
3. leaders are identified by recognized symbols of power or special behaviors
4. individual aggrandizement and/or cult of leaders
Column 5: Sharing of authority
1. leaders share power extensively with others
2. leaders share power with a large cadre of other leaders
3. leaders share power with a few other leaders
4. leaders exercise exclusive power
Column 6: Emphasis of authority
1. emphasis placed on group solidarity and group survival
2. emphasis shared between group and leader, with greatest importance given to group survival
3. emphasis shared between group and leader, with greatest importance given to leader survival
4. emphasis placed on leaders as the embodiment of the group
Column 7: External contacts
1. few or unimportant
2. external contacts are part of leaders’ authority, but not exclusive
3. external contacts are key to leaders’ authority, but not exclusive
4. external contacts are exclusively controlled by leaders
Column 8: Sum of scale items (columns 3–7)
Column 9: Overall warfare frequency (from Ember and Ember 1992b)
Column 10: External warfare frequency (from Ember and Ember 1992b)
Column 11: Socialization for trust (from Ember and Ember 1992b)
Column 12: Natural and pest disasters (from Ember and Ember 1992b)