Turchin & Gavrilets (2009, 2010) Hierarchy model

Turchin P & Gavrilets S (2009) Evolution of complex hierarchical societies. Social Evolution & History 8, 167-198.

Gavrilets S, Anderson DG, & Turchin P (2010) Cycling in the complexity of early societies. Cliodynamics 1, 58-80.

• The transition from small-scale to large-scale societies is not well understood.
• Apart from the scale of population size, other dramatic changes occurred in human sociality.
• From relying on face-to-face interactions to sustain social life and cooperation to other mechanisms such as symbolic markers to distinguish ‘us’ from ‘them.’
• From fiercely egalitarian to vastly inegalitarian.
• From a simple structure to a complex, centralized structure with many levels of hierarchical organization.
• From a stateless society to states – coercion-wielding hierarchical organizations managed by administrative specialists (bureaucracies.
• This paper reviews some theoretical approaches to the evolution of large-scale, hierarchically complex societies, and includes an agent-based model that looks at the rise of complex hierarchies in response to selection imposed by intergroup conflict (warfare).

From small-scale societies to states and empires

Two views on the evolution of the state have been (1) functionalist explanations focusing on the benefits brought by state organization and (2) conflict theories focusing on war-making capabilities of the state

A multilevel selection view allows us to integrate these.

Warfare is a potent mechanism of group selection. In small-scale socie- ties between 10 and 60 percent of male deaths are attributable to warfare (Keely 1997) and warfare is a major cause of cultural group extinction (Soltis et al. 1995).

3 ways by which social evolution can respond to selection pressure from warfare.

1. Groups can become more socially cohesive.

2. Warfare drives innovation and technological progress.

3. Intergroup competition is a major selection force in the evolution of larger group size. There are biological limits to the size of an egalitarian group.

How to get a big group (larger than 150) given the constraints of the human brain?

1. Symbolic markers to demarcate group membership (Richerson & Boyd, 1998). These include dialect and language, clothing, ornamentation, and religion.

2. Hierarchical organization, built out of a bond between a superior and an inferior ‘agents’. The growth of hierarchies occurs primarily by adding extra levels of organization and, therefore, is not limited by social channel capacity. Any member of a hierarchy needs to have a face-to-face relationship only with, at most, n +1 persons: the maximum number of subordinates (the ‘span of control’), n, plus an additional link to its own superior.

Hierarchical organizations can be human individuals as well as small-scale communities such as villages. The growth of hierarchically organized societies occurs by chiefly villages adding subordinate villages and by adding new layers of hierarchy on top of the pre-existing ones. Thus, hierarchical societies are also not limited by social channel capacity, and can potentially reach any size, as long as it is possible to add new hierarchical levels.

Hierarchical organization of human societies

• Among the nomadic and semi-nomadic Berbers of North Africa, from the Roman times on, an individual was simultaneously a member of a nested set of groups:
• a family, an extended family, a clan, and a tribe. Settled agriculturalists have similar organization. 
• Clans may live together in a local community (village), while a tribe unites a collection of such villages. Non-centralized, or acephalous tribes lack permanent leaders, while centralized tribes are led by a chief and, therefore, are called (simple) chiefdoms.
• Simple chiefdoms typically encompass some thousands of individuals. The next level of social organization is a complex chiefdom uniting several simple chiefdoms and having populations numbering in tens of thousands.
• It appears that an acephalous tribe is the largest social scale a human group can achieve without the benefit of centralized organization. Greater social complexity requires leaders – chiefs, kings, caliphs, presidents, prime ministers, or politburo chairmen.
• Adding extra levels of social organization beyond a complex chiefdom usually requires transition to a more formal political organization – the state.

If the elaboration of hierarchical complexity was driven primarily by the need to control and administer large populations, then we expect a fairly strong relationship between sizes of various political units (both polities and subdivisions within polities) and the number of control levels within them. This proposition can be tested empirically. Plotting the number of levels within a political unit against its size for six historical empires yields a series of parallel lines (Fig. 2).

Cycling and the evolution of social complexity

The model is described more fully in Gavrilets et al. (2010), so that’s where this part and subsequent notes come from.

Cycling: Once established, complex societies rose and fell over time, with centers of power and authority shifting from one location to another over the landscape.

Prerequisites for the evolution of social complexity (drawing heavily from Carneiro):

1. Warfare

2. Circumscription (environmental, due to the resource concentration, or social, due to the presence of other human groups nearby). Circumscription was the factor that precluded losing communities from moving away and thus separating themselves spatially and politically from victors.

3. Existence of agricultural potential capable of generating surpluses

4. Significant variation in productive and/or demographic potential among local communities

5. The ability to delegate power and the invention of hierarchically structured control mechanisms in which each superior directly controlled only a limited number of subordinates.

The model

Structure:

• Consider a hexagonal array of initially autonomous communities. Each community has up to six neighbors.
• Each unit of time is a year, and is the expected interval between two consecutive “decisions” made by a community.
• Each community i is characterized by a baseline resource level f0_i. These are chosen randomly from a truncated normal distribution with mean 1 and SD sigma, which represents variation in productive/demographic potential between local communities. The actual resource of a community is f_i.
• Each community is part of a polity (which can be just one community).
• Each community in a polity except for the one at the top of the hierarchy (the “chief community”) has one superior community and up to L subordinates, where L is the maximum “span of control”. 
• Each subordinate community pays tribute by transferring a fixed proportion theta of its total resources to its superior.
• The total resources of a community are the sum of the resources f_i it produces and the tribute from subordinates.
• The power (wealth) of a polity F_i is the total resources available to its chief community.
• The complexity of a polity c_i is given by the number of levels of control above the level of individual villages.

Dynamics:

• Each chief community and each of their direct subordinates make exactly
• one decision every year. For the chief community, the decision is whether or
• not to attack a neighboring polity. For a direct subordinate of a chief
• community, the decision is whether or not to attempt to secede.
• Warfare: A polity selects its weakest neighbor, and calculates the probability of success of an attack upon it (which increases the probability of attack, and scales with relative increase in power) and the attack costs (which decrease the probability of attack, and are generally higher for unsuccessful attacks).
• The aggressor attempts to conquer communities of the victim, starting with border ones, and proceeding in a series of “battles” until either it suffers a defeat, or until the chief community of the victim polity is conquered. Thus, the aggressor either fails completely, seizes a part of the victim polity, or the whole victim polity is annexed.
• Conquest and annexation: If the chief community i has an open control slot, it will control polity j directly. If there are no open control slots, then the chief community will control directly the L wealthiest communities chosen (i.e., promoted) from the set of its L subordinates and the newly conquered polity j. The remaining (i.e., the poorest) community will be demoted and reattached to its geographically closest neighbor of the higher rank (i.e., by-passed in a process akin to linearization). If this neighbor has already filled all its control slots, further rearrangements will follow according to the same strategy.
• Secession: A community subordinate to the chief polity will secede if it estimates that the attack of its old master will be successfully repelled and is willing to pay the price of rebellion.
• The chief polity attempts to suppress the rebellion immediately.
• If a successful rebellion results in spatial separation between different parts of the master state, all communities that become disjointed from their superiors secede as well.
• To account for a possibility of secession upon the death of the paramount chief as a result of a struggle among subchiefs (which is a major source of instability in chiefdoms), we introduce an additional parameter T, the average time in power of the paramount chief. Upon the death of the paramount chief, a random number of subordinate communities become independent without war.

Results: They got the emergence of complex organization and cycling among states.

Discussion

Our model provides theoretical support for a view that the formation of complex polities is “a predictable response to certain specific cultural, demographic and ecological conditions” (Carneiro 1970).

The chief’s expected time in power T is one of the two most important parameters. Highlights the crucial importance of having well-defined and legitimate mechanisms of succession for the stability of polities. Creating and maintaining complex polities thus requires effective mechanisms to deal with both internal and external threats.

The other critical parameter of the model is the probability of success in war, which sets the relative effectiveness of stronger (wealthier) polities in internal and inter-polity conflicts. In our model, the stronger of the two polities does not necessarily win a conflict between them. This is reasonable as there are many other factors besides wealth that can affect the outcome of conflict. However increasing this implies a stronger dependence of the outcome of the conflict on the polities’ power (wealth). The degree of determinism in the conflict resolution is expected to increase with economic and political development.