High Reliability Pastoralism

Emery Roe, Lynn Huntsinger, and Keith Labnow

University of California, Berkeley

Abstract: The literature on pastoralism is sufficiently rich to accommodate two very different models of pastoralism. Currently, virtually all attention given to pastoralism focuses on herder risk aversion, ecological adaptation and the need for herd mobility in the face of an unpredictable environment. In contrast to the model of risk-averse pastoralism, the disequilibrium-based models of ecological dynamics on rangelands, often referred to as the "new range ecology," enable us to see pastoralism as a high reliability institution. From this perspective, high reliability pastoralism is the search and attainment of reliable peak performance through utilizing and managing highly complex technologies. Which model is closest to reality remains an empirical issue. Nonetheless, the policy implications for pastoral development and rangelands are very different, if pastoralists are found to be primarily reliability-seeking rather than risk-averse. Moreover, the implications for our understanding of pastoralism and its future are profound and differ appreciably from current conventional wisdom.

Introduction and Summary

A great deal of ink has been spilt over what is seen as the degradation of the largest land category in the world, this planet's arid and semi-arid drylands, by its largest group of users, pastoralists. Much of this written record needs to be rethought entirely, and what follows is, we believe, the first major rethinking of pastoralism in the last fifty years. Its major implication is that only after we have determined a more appropriate model of pastoralist behavior can we then address both what pastoralists are actually doing and the policy implications of their actions.

Why do we need a new model of pastoralist behavior? The study of pastoralism continues to center on the implications of unpredictability in rangeland ecosystems for pastoral societies and land use patterns (Sandford, 1983a, is the starting point of any such study). Conceiving unpredictability as risk, and pastoralists as risk averters, makes sense where rangeland unpredictability is in fact the core exogenous driver of pastoralist decisionmaking. We argue instead that the central concern of pastoralists is to manage a predictably unpredictable environment better, so as to establish a "reliable" flow of life-sustaining goods and services from rangeland ecosystems that are in fact an endogenous part of their production systems.

As a start, think of the risk-averting pastoralist as engaged in an attempt to avoid or escape the high hazards of ecological unpredictability, given that the pastoralist has no control over the probability

of those hazards occurring. The pastoralist in search of reliability is, in contrast, actively engaged in ongoing efforts to reduce the probability of those hazards he or she cannot avoid by managing temporal and spatial diversity in grazing opportunities and diversity in livestock capabilities and response. Rather than being risk averse in trying to avoid hazards altogether, pastoralists accept and even take risks in order to respond to high-consequence hazards they cannot altogether avoid. In that other high reliability organizationnuclear power plantsyou find computers, cooling systems, alarm systems and other intricate mechanisms operated so as to harness the unstable power of the atom. Pastoralism is harnessing the unstable power of the range. In both cases, the attempt is less to avoid hazards than to accept and manage them better.

While there is some overlap between the risk averse and high reliability models, the high reliability model differs diametrically from the risk aversion one on the most important dimensions, and these differences have contrasting implications for pastoral development, policy and our understanding of pastoralist societies. To appreciate those aspects better, we turn to what has been called the "new range ecology."

The New Range Ecology

The New Range Ecology seeks to accommodate range management to the complexity and inherent unpredictability of rangeland ecosystems. Clementian models of plant succession have long provided a simple, hitherto compelling framework that permits evaluation of the ecological status of rangeland ecosystems and prediction of response to management, including development of recommended carrying capacities and stocking rates (Behnke et al, 1993). Increasingly, the shift in rangeland ecology is away from simple assumptions and models to a greater appreciation of the complexity of vegetation dynamics as they interact with herbivory and random events or "disturbance" on rangelands. In between alarmingly simplified succession models that assume an orderly and predictable progression to a "climax" equilibrium state, and equally alarmingly complex diagrams that appear frequently today in symposia papers purporting to illustrate rangeland ecosystem dynamics, the need has grown for models or frameworks that maintain the virtues of simplicity but capture more of the complexity (and thus the unpredictability) observed on many rangelands around the world. The New Range Ecology, which fo

cuses on the disequilibrium conditions of rangelands, offers state and transition models as one such option (Westoby et al, 1989).

State and transition models do not assume a linear progression for vegetation development on a given site. Instead they offer a variety of end and transition states, with a number of possible pathways or transitions between them. In the disequilibrium environments characteristic of rangelands, state and transition models make it clear that operating scale and management objectives take on different roles than those conceptualized in more traditional models of rangeland vegetation dynamics. In many rangeland environments, transitions are more strongly driven by random, unpredictable events and contingencies than by stocking rates or patterns of herbivory. Managing for particular vegetation states becomes unrealistic in these cases, where instead the process of maintaining reliable production over different operating scales despite unexpected shifts takes priority. For ecologists, this means both rethinking approaches to the evaluation of ecological change and the development of goals and standards of management and focusing on the inter-relationships of ecosystems, management, and disturbance.

What does this mean for our understanding the ways pastoral people interact with rangeland ecosystems? The answer: It is now much easier to see pastoralism as more akin to what has been termed a high reliability organization. The details follow, but simply put, the New Range Ecology tells us that achieving a "steady state" or equilibrium between stocking rate and forage production is not possible on many, if not all, arid and semi-arid rangelands (Behnke et al, 1993). Instead, the interaction of pastoralists and their surrounding conditions has created an institution of reliably maintaining high (i.e., peak) levels of livestock in order to exploit the expected but unpredictable production of low-quality vegetation. In this view, livestock are best thought of as part of the complex technology adapted by pastoralists to utilize and "smooth" a highly volatile, unpredictable and inherently "dangerous" forage resource key to their production system. Livestock are mobile, have a sophisticated digestive capacity to process fibrous vegetation, and are ideally suitable to a diverse landscape where production varies over time and from place to place and where grasses and shrubs are the forage supply. In effect, the rangeland-livestock production system is technology adopted and adapted by pastoralists to provide a steady flow of human sustenance from the herbivory of animals in an environment whose only certainty is its high spatial and temporal diversity. While herd and herder mobility is an important management

tool, ability to cope with diversity and change in any given place is fundamental.

Why and how specifically pastoralists do this becomes clear in the following sections.

Risk Aversion and High Reliability Theory

Unfortunately, the New Range Ecology has not until this point precipitated a thorough reassessment of pastoral social organization and management strategies. In fact, the new understandings about the biology of grazing systems have been grafted onto the dominant model of risk-averse pastoralism without realizing they enable us to recast the conventional understanding of pastoralism in a fresh light.

What we call "risk-averse pastoralism" views pastoralism as a risk averse adaptation of herders and herds to environmental and ecological stress, arising in large part from external factors such as the variability in rainfall and grass cover. Accordingly, so the argument goes, anything that ensures or assists herd mobility is to be encouraged and anything that hinders it is to be discouraged. The frequently-drawn conceptual and policy implications of risk-averse pastoralism are that the range as a common property or otherwise communal resource is a good thing, while fixed boundaries, agricultural encroachment (particularly on dry season grazingland), privatization of the range, and fenced ranches are a bad thing, because they inhibit or otherwise retard herd mobility. In this view, pastoralism is fast becoming impossiblea conclusion at odds with the high reliability pastoralism described below. The assumption that pastoralism is a risk averse adaptation to a highly variable exogenous environment has been unchallenged over the last fifty years. It dominates analysis about, as well as development projects for, pastoralists (e.g., see Swallow, 1994; Dahl and Hjort, 1976; Galaty and Johnson, 1990).¹ It is now time to challenge that assumption, both because of the New Range Ecology and because the performance of many development projects has been so disappointing (e.g., de Haan, 1994; UNDP, 1994, p.1). A different model as to why pastoralists do what they do is found in high reliability theory.

The starting point in rethinking pastoralism must be questioning the centrality given to herder risk aversion. A moment's reflection makes clear some problems with the risk aversion modelif it were simply a matter of pastoralists being risk averse, then why ever do they stay and live in such a hazardous environment? Why haven't they all long ago flooded into the towns or agriculture or into something less risky?

The answer is that herders, instead of seeking to avoid risk altogether, have in all the important senses accepted it. To see why, we must be clear about what is meant by risk aversion and high reliability, at least as ideal types. Formally, risk is defined as the magnitude of a hazard multiplied by the probability of that hazard occurring. In risk-averse pastoralism, both the hazard and the probability are large, with the paradigmatic case being the inevitable droughts in an arid and semi-arid environment. From the risk averse perspective, the way herders respond is (i) by accepting that the probabilities of these hazards occurring are largely outside the pastoralist control and, then, (ii) by trying to avoid the hazards altogether (e.g., they move the herd to where they think the grass and water are better) or by trying to reduce the magnitude of the hazard directly (e.g., "spreading the risk" through separating the herd across a large geographical space). In both cases, the attempt is to ensure a minimum survival level of the herd, thereby ensuring the survival of the herders (e.g., Ellis, 1993, p. 89).

The high reliability ideal type interprets pastoralist behavior quite differently. Both the hazard and its probability are large in high reliability pastoralism, though, unlike risk-averse pastoralism, it is much more focused on probability than hazard. Accepting full well what the hazards are of living and working in an arid and semi-arid environment, pastoralists act in ways that enable them to maintain their large herds and ways of life in as highly reliable a fashion as they can. In formal terms, pastoralists seek to ensure their capacity to maintain very high, peak levels of livestock in the face of the hazards that cannot be otherwise avoided. They do this by avoiding management mistakes and other failures in a manner that allows them to reliably reduce the probabilities associated with otherwise highly probable hazards of living in their dryland environment. In high reliability pastoralism, the hazards cannot be avoided and the probabilities must be managed instead, in large part because what the risk averse model treats as exogenous to pastoralism, the high reliability model treats as endogenous to pastoralist production, namely, the highly variable environment of rainfall and grass cover.

This is not to say that all hazards are avoidable (from a risk averse perspective) or that all probabilities of hazards are reducible (from a high reliability perspective). Consider the hazard of low and erratic rainfall. No pastoralist behavior, whatever its nature, can change the hazard arising because a semi-arid and arid rangeland has, by definition, 20 inches (500mm) of rainfall or less per annum nor can pastoralist behavior change the hazard arising because even this little rainfall is erratically distributed

over any given area there. On the other hand, how the hazard of a severe drought having a 0.5 or greater probability of killing half or more of the herds one out of every ten years is understood and treated by pastoralists varies by whether one is talking about risk-averse or high reliability pastoralism. From a risk averse perspective, pastoralists will be searching for better grazing and water as a way of escaping the worse effects of the drought and as a way of increasing the chances of being left with a minimum survival herd after that drought. From a high reliability perspective, pastoralists will be managing the spatial and temporal diversity of their rangeland as a whole by using different areas in different ways so as to increase the chances of producing and maintaining peak herd sizes, even through the drought. The implications of these differences are considerable, as we shall see.

At the most general level, then, risk-averse and high reliability pastoralisms overlap in the sense that high reliability pastoralists are very "risk averse" when it comes to one special class of risksthose where the consequences of management error are catastrophic. That said, at all the other levels that matter, the difference between, on the one hand, averting failure by avoiding hazards whose probabilities cannot be controlled so as to maintain a minimum survival level of the herd and, on the other hand, seeking reliability by managing high-hazard, high-probability events into lower probability ones so as to avoid failures and maintain peak herd levels cannot be overstressed. The difference entails other formal distinctions between risk-averse and high reliability pastoralism important for our understanding of pastoralism and its wider implications. The contrasts between the two ideal types of pastoralism are summarized in Table 1. We take up these distinctions in the following sections.

High reliability theory is a relatively recent development of organization theorists, many located at or around the University of California at Berkeley, who have been interested in how complex organizations and institutions maintain their activities in situations where failure, error and accidents are highly probable. The question is (e.g., Demchak, 1996, p. 97), How do some institutions, with complex technologies and in predictably unstable environments, still manage to perform in a reliable, safe fashion, even when their first mistakes might well be their last? High reliability organizations studied in the West have included air traffic control systems, nuclear power plants, electricity companies, hospital intensive care units, and naval air carriers, among others. We nominate pastoralism to the list of high reliability organizations, by arguing that pastoralism shares mostmaybe even allof the same attributes

as these other organizations.

According to two high reliability theorists, Todd La Porte and Paula Consolini (1991, p. 21), the operating challenges of high reliability organizations "are twofold: (1) to manage complex, demanding technologies, making sure to avoid major failures that could cripple, perhaps destroy, the organization; at the same time (2) to maintain the capacity for meeting periods of very high peak demand and production whenever these occur." Notice the goal is twofold: maintain safety and at the same time meet peak production requirements (see also La Porte, 1996). Maintaining peak livestock loads is precisely the core problem pastoralists face when confronting erratic rains and other potential disruptions.²

The pastoralists' biggest challenge, like the challenge confronting all high reliability institutions, is to reduce the number and magnitude of their errors, given that the hazards associated with making highly consequential mistakes are so high. In the words of another high reliability theorist, Gene Rochlin (1993, p. 16): "A highly reliable organization is often defined as one that has already been judged on empirical or observational grounds to provide a desirable activity, product, or service at a desired or demanded level of performance while maintaining a low rate of error or accident." High reliability institutions are all the more notable, according to Rochlin (1993, p. 15), because "these organizations have not just failed to fail; they have actively managed to avoid failures in an environment rich with the potential for error." That ability to actively and reliably manage to reduce the chances of hazardous mistakes occurring, rather than to avoid the hazards, has been the distinguishing hallmark of much of pastoralism.³ For us, it is not risk aversion only or even primarily, but rather high reliability, that explains the pastoralist's "uncanny ability to survive and sometimes prosper under considerable adversity" (World Bank, 1987, p. 29).

Moreover, complex technologies and the wider, uncertain task environment ensure there can be no "equilibria" to which high reliability institutions are driving or from which they are departing. Here the link to the New Range Ecology could not be more explicit: "In the research conducted by the Berkeley group, virtually every manager interviewed appeared to believe that there is not a steady-state or a stable 'resting point' for the high performance system under their management," according to high reliability theorist, Paul Schulman (1993, p. 35). In order to identify the links better, the following section sketches nine of the twelve features that have been identified for high reliability institutions, and illus

trates how these attributes are to be found in the pastoralist literature, a fraction of which has been reviewed here.

High Reliability Pastoralism

Clearly reliability matters to pastoralists, and examples abound. According to Reckers, "The diversity of herd stocks [among Kenya pastoralists] allows a more efficient use of the rangelands and facilitates a more reliable supply of food" (1994, p. 49). Reliability is at a premium for pastoralists, because they are found in areas "where the natural resources that population is technologically capable of exploiting are unreliable" (Spooner, 1973, p. 4). What does "reliability" mean precisely?

Rochlin (1993) has summarized the principal features of high reliability institutions in his "Defining 'High Reliability' Organizations in Practice: A Taxonomic Prologue". The work of other high reliability theorists, particularly Todd La Porte, is used to supplement and extend this list of primary features. We have collapsed the features identified by Rochlin, La Porte and their colleagues into a dozen interrelated characteristics of high reliability institutions, of which nine principal ones are summarized here. No pretense is made that all theorists describe the features in the same way or that the features they describe apply to all pastoralists. What is important for our purposes is that each feature comes from the work of high reliability theorists and is found in the very same literature that has been used to justify the model of risk-averse pastoralism.

High Technical Competence

High reliability institutions "manage technologies that are increasingly complex, requiring specialized knowledge, specialized management, and a variety of esoteric skills at the operational level" (Rochlin, 1993, p. 14). "High technical competence" is how La Porte (1993, p. 1) summarizes this characteristic.

It has long been noted that pastoralists have had and continue to have very detailed knowledgesometimes called indigenous technical knowledge (e.g., UNDP, 1994)of the range, key resources, cattle and other livestock they manage as part of their production systems (e.g., Dyson-Hudson, 1988, p. 702 on the Turkana, Hobbs, 1989, p. 111 on Egyptian Bedouins, and Oba, 1985, pp. 37, 39 on other

Kenya pastoralists). Several important corollaries follow from the high technical competence requirement of high reliability institutions.

The first is what the requirement entails, namely, "continuous training" (Roberts, 1988, Figure 3). As La Porte (1996, p. 63) puts it: "Continuously attaining this [technical competence] entails attention to recruiting, training and staff incentives. It puts a premium on recruiting members with extraordinary skills and/or an organizational capacity to develop them in situ via continuous training". "The East Pokot possess an immense plant knowledge," according to Reckers (1994, p. 51), where "Every child is able to learn and identify plants and knows their value in terms of human and animal consumption". As Spooner (1973, p. 17) notes that the pastoralist's "intimate and detailed knowledge of his territory...must be continually rehearsed and revised. This knowledge is the foundation of the technology of successful nomadism, and represents a huge and continuous investment in time and energy".

A second corollary of maintaining high technical competence is the need for, as La Porte (1993, p. 3) argues, an "Extensive data base characterizing technical processes [and] the state of the system in operations, including performance data" (see also La Porte, 1996, p. 63). Charles Perrow (1994, p. 218), another organization theorist who has engaged high reliability theory, calls this the need for "Experience with operating scaledid [the institution in question] grow slowly, accumulating experience, or rapidly with no experience with the new configurations or volumes?" The more extensive the data base and experience with differing operating scales, the greater the chances the organization can act in a reliable fashion, other things being equal.

The detailed experience with and knowledge of different operating scales in livestock herding and holdings constitutes one of the pre-eminent features of pastoralism. Some of pastoralism's scale dependence is captured in statements such as "the pastoralists' flexible responses to the spatially and temporally distributed pasture and water have enabled them to exploit an environment which would have been impossible to exploit under other forms of land use systems" (Oba, 1985, p. 52). Or, "Such periods of stress [e.g., droughts] are met by migrating to other areas, ie. expanding the spatial scale of exploitation" (Manger, 1994, p. 14). Or, "In the arid context, it is precisely the two factors of space and time, rather than number of animals, that determine sustainable carrying capacity and have been used efficiently by traditional pastoral managers of extensive rangelands" (UNDP, 1994, p. 21). Or, as Behnke,

Scoones and Kerven (1993, p. 73), summarize:

"Cattle forage at different spatial scales. At the widest scale, animals move between different savanna types....[S]patial heterogeneity thus has an impact on the interactions between population and resources, at different levels in a hierarchy of scales. The persistence of cattle populations can be interpreted in terms of the exploitation of environmental heterogeneity at different spatial scales".

As will be seen, the experience of herders with different operating scales and the fact that different resources are scale-dependent have profound implications for reconsidering the importance of land tenureespecially common property regimesin pastoralism.

Just as experience with operating scales is important to high reliability institutions, so too is their experience with critical phases of operation. "Experience with critical phases," Perrow (1994, p. 218) tells us, is imperative in this regard: "if starting and stopping are the risky phases, does this happen frequently (take-offs or landings) or infrequently (nuclear plant outages)?" Here too the cognate is well-known in pastoralism in the form of pastoralist cycles of moving between wet and dry season grazing and other resources over time (e.g., Reckers, 1994, p. 51).⁴

High Performance And Oversight

High technical competence in a high reliability institution must be matched by continual high performance. "The public consequences of technical error in operations have the potential for sufficient harm such that continued success (and possibly even continued organizational survival) depends on maintaining a high level of performance reliability and safety through intervention and management (i.e., it cannot be made to inhere in the technology)" (Rochlin, 1993, p. 14). Accordingly, "Public perception of these consequences imposes on the organizations a degree of formal or informal oversight that might well be characterized as intrusive, if not actually comprehensive" (Rochlin, 1993, p. 14). "The organization will be judged to have 'failed'either operationally or sociallyif it does not perform at high levels. Whether service or safety is degraded, the degradation will be noted and criticized almost immediately" (Rochlin, 1993, p. 16). La Porte (1993, p. 7) adds that: "Aggressive and knowledgeable formal and informal watchers [are] IMPORTANT. Without which the rest [i.e., high reliability] is difficult to achieve."

Precisely this need for constant oversight (surveillance and monitoring), both formal and infor

mal, in order to maintain pastoralist survival, safety and reliability is nicely captured in several references. Spooner (1973, p. 17) quotes a researcher who found that the Tibetan nomad "remains alert to the faintest whisper that riders have left their own encampments or have been seen at large". Lancaster and Lancaster (1986, p. 42) write of the herder:

"He needs an enormous amount of information. The most likely place to acquire such information is from other people, so he needs to be able to assess the reliability of informants. To do this he must know who they are. Usually the most reliable informants are those to whom he is related, for their interests are in common, the reliability diminishing the further away genealogically that the informant is."

As the quote implies, the success of surveillance and monitoring depends crucially on the number of reliable informants and participants. Perrow (1994, p. 218) describes this feature as follows: "Organizational density of the system's environment (vendors, subcontractors, owners/operators...etc.)if rich, there will be persistent investigations and less likelihood of blaming God or the operators and more attempts to increase safety features". Some of this drive to a denser form of management in pastoralism is glimpsed in statements such as Spooner's (1973, p. 9): "Where a nomadic group relies on a number of different species, herding requirements...require a higher degree of cooperation between families than would be necessary if they could specialize in one species". We return to the importance of families from a high reliability perspective in a moment.

Constant Search For Improvement

A feature related to high technical competence and continual monitoring in these high reliability institutions is the constant drive to improve operations. "While [high reliability organizations] perform at very high levels, their personnel are never content, but search continually to improve their operations" (Rochlin, 1993, p. 14). They seek "continually to search for improvement via systematic gleaning of feedback" (La Porte, 1996, p. 64).

For Ekvall (quoted in Spooner, 1973, p. 16), the pastoralist is not "satisfied with generalities: he has a well developed question and elicitation technique that seeks specific and essential items of information," adding that "At the same time he is continually on the alert for other resources to exploit" (Spooner, 1973, p. 23). "Every morning the herd owner decides upon a new route for the daily livestock migration," Reckers (1994, p. 50) found among Kenya pastoralists.

The constant search for improvements is aided by what La Porte (1993, p. 7) calls "Systematic, consistent individual and group rewards for discovering error, performing acute analysis, and proposing potential solutions." Indeed, such search is often in the form of "Formalized processes of discovery and review" (La Porte, 1993, p. 7). Roberts (1988, p. 37) adds, "personnel are trained that when they see a problem they own it either until they solve it or until someone who can solve it takes responsibility for it." Some of this formalized search and discovery in assuring high reliability is captured in the following quotes (the first from Spooner, 1973, p.16; the second more recent in Agrawal, 1992, p. 23):

"Herders leave their campfire to waylay passers-by for news, and men ride out from the encampments to intercept approaching or passing caravans for news of places and events both near and far....The price for completely reliable information is high, but information for which a price has been paid rates well as evidence and has a standard market value..."

"The nambardar [chosen leader] in the dang [social unit of migratory pastoralists] often undertakes reconnaissance missions to gather information regarding rainfall and grazing availability. They travel 5 to 20 miles ahead of the dang and gather information on the state of vegetation over the proposed route. They look at the state of water-points and find out if the farmers with whom they are acquainted are present in their villages. Since dangs move almost every day, such reconnaissance missions are invaluable for getting advance information which will help the movement of the dang."

The constant search for and discovery of new information and improvements leads to permanent innovations in some pastoralist behavior and technology. Hobbs (1989, p. 113) concludes of the Egyptian nomads he studied: "The Khushmaan apparently have had to innovate in order to protect their resource base because their ideal of common access to resources could not guarantee sustainable resource use during times of prolonged drought. . .In view of the nomads' abilities to adapt themselves to some of the world's most marginal conditions, this innovation is not surprising."

Highly Complex Activities

"The activity or service [of a high reliability organization] is inherently complex, in that tasks are numerous, differentiated, and interdependent" (Rochlin, 1993, p. 15). This too is found throughout pastoralism (in addition to preceding quotes, see the literature review by Dyson-Hudson and Dyson-Hudson, 1980). "All nomadic movements involve complex decisions," in Spooner's (1973, p. 21) summation.

One of the conceptual advances of the New Range Ecology has been to clarify better the inseparability of herder environment and technology in the arid and semi-arid areas of the world. The great failing of risk-averse pastoralist model has been to draw too rigid a line between the two, arguing that the herder's environment is unpredictable and complex, while herder technology is simpler and more flexible (see Spooner, 1973, p. 4; Abercrombie, 1974, p. 13). In fact, the herder's production technology is very complex. It includes the spatial and temporal variation and interrelated dynamics of water, vegetation and other key resourcesall too often lumped misleadingly together as "the external environment"as well as the links between this variation and dynamics and the livestock behavioral and physical characteristics. Coping with this complexity necessitates specialization across communities and within the herder household and the herds themselves (Bonte, 1981). Different household members or communities specialize in different aspects of livestock and environmental management, for example. At the same time, the composition of the herds are often specialized at the household level.

For high reliability organizations, "It is impossible to separate physical-technical, social-organizational, and social-external aspects; the technology, the organization, and the social setting are woven together inseparably" (Rochlin, 1993, p.16). But not totally inseparable: "...it must also be shown that the reasons for failure did not arise from exogenous factors outside the organization's possible range of control (e.g., through an 'act of God' or a technical failure in an area of design or implementation not within its power to identify or correct)" (Rochlin, 1993, p. 22).

The notion that high reliability depends on what we can manage or fail to manage, rather than on what we cannot manage or is outside our control, has implications for the reconsideration of the role of drought and drought response in pastoralist behavior. A considerable amount of the literature on pastoralism, particularly the more recent variety, has been devoted to gauging the effect of drought on pastoralist peoples. Much of this literature has been preoccupied with pastoralist drought responses (e.g., Dahl and Hjort, 1979 and de Haan, 1994).

High reliability theory suggests, however, that it is crucially important to distinguishif not practically then at least conceptuallythose droughts that are "an act of God" in the Rochlin sense from those that are the product of humanincluding pastoralisterror and mistake (which some droughts are). From a high reliability perspective, pastoralism does not "fail" when a drought unexpectedly oc

curs nor should pastoralism be evaluated in terms of how effectively pastoralists respond to such an "act of God." To reiterate, the successes and failures of pastoralism, like other highly reliable institutions, must be evaluated in terms of and equated with that which people can and actually do manage, such as peak herd size, movements and distribution.⁵ To insist that a widened range of drought response and management (e.g., food distribution programs, food-for-work projects) must become part and parcel of pastoralist behavior risks repeating the very same mistake made with fenced ranching schemes of the 1960s and laternamely, the misguided insistence both that pastoralists should be reliable in a ways they are currently not and that they should be evaluated on how well they meet these new standards of reliable performance.

High Pressures, Incentives And Shared Expectations For Reliability

"The activity or service [of a high reliability organization] meets certain social demands that require performance at the highest level of service obtainable within present safety requirements, with both a desire for an even higher level of activity and a penalty (explicit or implicit) if service slackens" (Rochlin, 1993, p. 15). The social demands on pastoralists to cooperate in order to maintain peak reliability (in this case the safety and high production of the herds) has been much commented upon. As was seen a moment ago, where nomadic groups rely on different livestock species, herding requirements often require a higher degree of cooperation between families than would be necessary if the groups were specializing in one species (Spooner, 1973, p. 9). For Lancaster and Lancaster (1986, p. 45), "the Rwala see themselves as managing an environment in co-operation with the other users of it for the benefit of all. This moral duty is not only to their symbiotic partners, but to generations on both sides of the nomad/settled equation as yet unborn".

Neville Dyson-Hudson notes that a Karamojong pastoralist manages his herd and family in "mutual association" (in Spooner, 1973, p. 13). "Families recognize a household head (usually male) who has most of the apparent responsibility for herd management decisions. All family members engage in herding activities. Children are especially important for the day-to-day maintenance of animals" (Kuznar, 1991, p. 99). "There are also traditions for mobilizing work parties or age-groups according to the age grades of young women with a female leader or young men with a male leader"

(Shanmugaratnam et al, 1992, p. 56). For Lancaster and Lancaster (1986, p. 44), "Owing to the general unpredictability of the environment no one could ever be sure where he would be the next year, the following year or the subsequent one. So a man needed a variety of khuwa [brotherhood] relationships with a variety of settlements." Similar passages can be found elsewhere in the pastoralist literature (e.g., Dahl and Hjort, 1976, pp. 133-134). One virtue of organizing high reliability around kith, kin and neighbors is they provide greater chances of, in La Porte's words (1993, p. 7), "Relatively assured resources for/to carry our failure-preventing/quality enhancing activities" as well as, in Perrow's words (1994, p. 218), "Close proximity of elites to the operating systemthey fly on airplanes but don't ship on rusty vessels or live only a few blocks from chemical plants."

Hazard-Driven Flexibility To Ensure Safety

"The activity or service contains inherent technological hazards in case of error or failure that are manifold, varied, highly consequential, and relatively time-urgent, requiring constant, flexible, technology-intrusive management to provide an acceptable level of safety [i.e., reliability] to operators, other personnel, and/or the public" (Rochlin, 1993, p. 15). This accent on flexibility in securing safety is a pre-eminent feature of pastoralism. For example,

"Nomadic pastoralism, as practiced in East Africa, presupposes much organizational and spatial flexibility. Not only do households constantly redistribute themselves over the terrain, in response to climatic fluctuations and the needs of herd management, but membership of pastoral households, too, is continually changing as labour is allocated and reallocated between management units." (Dahl and Hjort, 1979, p. 29)

For Spooner (1973, p. 22) "the seasonal variability...requires a certain amount of fluidity in social organization". "The [pastoralist] strategies are in harmony [sic] with the yearly cycle of rainy and dry seasons and are modified when needed: e.g. by expansion or shifting. In this way menacing situations are controlled" (Reckers, 1994, p. 51). According to El Wakeel and Abu Sabah (1993, p. 37), "transhumants roam large areas to look for better and safe grazing areas". Also, "the ungrazed belts between areas effectively used by wildlife and those used by livestock noticed during aerial surveys in Maasailand are not evidence of the abundance of pasture, but are 'safety belts' deliberately created by the pastoralists to protect their herds from infection by wildlife" (UNDP, 1994, p. 71).⁶

One primary way in which organizations aspiring to high reliability develop the required flexibil

ity is to ensure redundancy is built into their operating systems. By redundancy, organization theorists mean high levels of duplication (where two units perform the same activity separately), overlap (where two units have the same activity together), or both. The argument is that the more redundant (back-up) modes of problem-solving there are, the more reliable performance can be, if any one mode of problem-solving fails (Roberts, 1988; Rochlin, 1993; Sagan, 1993). Fallback or back-up water points and grazinglands, as well as overlapping grazing areas, have long been a property of some pastoralist systems (see, e.g., Fortmann and Roe, 1981; Sandford 1983b; UNDP, 1994). Redundancy extends to other pastoralist resources: "In sum, the Khushmaan have three redundant and reinforcing rules that protect their trees: no tree anywhere should be cut; groups of trees within particular areas should not be cut; and specified individual trees should not be cut," writes Hobbs of Egyptian Bedouins (1989, p. 106). Clearly, though, the most important redundancy built into the pastoralist system to ensure its high reliability is its many livestock. What outsiders perceive to be too many cattle are in reality cattle kept because they are duplicates of each other. When one dies, others are there to take its place. Such, indeed, is the meaning for pastoralists of the much-vaunted "safety in numbers."

But it is "safety in numbers" with a twist. According to risk-averse pastoralism, pastoralists keep as many animals as possible in order to increase the chances of being left with a minimum or better herd size after the next drought or other catastrophe (e.g., Hobbs, 1989, p. 103). In this view, safety in numbers means the greater the number of livestock, the better the chances some will survive, worse comes to worst. High reliability pastoralism offers a very different interpretation of large herds. From its perspective, pastoralists keep as many animals as possible in order to increase the chances of producing a peak herd size as safely and for as long as they can before and during the next drought or other catastrophe. What looks to be a minimum herd size left behind after the drought ends is, from a high reliability perspective, the peak herd size pastoralists have managed to maintain through the end of the drought. By actively assembling and maintaining large herds, pastoralists create a production system that pressures them to manage these herds in a safe and reliable fashion, such that they, their herds, and the societies they represent have even more to lose if they fail. La Porte (1996, p. 61) puts the point this way: "As societies come to depend on systems designed and deployed in ways that risk putting their operators, consumers and citizens in harms way, demands for [high reliability] performance are insistent". In

this view, safety in numbers means the greater the number of livestock, the better the chances that they will be managed safely and reliably, before and while the worst happens.

Culture Of Reliability

High reliability institutions "face the challenge of being highly reliable, both as producers (many under all manner of demanding conditions) and as safety providers (under conditions of high production demands). This suggests an organizational culture integrating the familiar norms of mission accomplishment and production with those of the so-called 'safety culture'" (La Porte, 1996, p. 64). For Roberts (1988, Figure 3), this organizational culture is a "culture of reliability".

Some of this culture of reliability and safety can be glimpsed in descriptions of pastoralists generated from the dominant risk aversion framework. According to Dahl and Hjort (1979, p. 18),"safety first" is the motto of subsistence operations like those of pastoralists, adding: "Insurance and security are central themes in the East African pastoral societies and expressed both in the social structure and on the level of individual action." "To distribute one's cattle resources is," in Spooner's view (1973, p.13), "a form of insurance against natural hazard and enemy depredation". For example, Western (1979, p. 94) also found that, among the important selection criteria for Maasai settlements is a "location...that will minimize the hazards, production losses, and general discomfort of the occupants, and provide the essential settlement materials." The problem with such descriptions is that they get only half the picture right: The safety is there, but so are the peak production requirements, where what is minimized is less the hazard of operations than the probability of their occurrence.

Reliability Is Not Fungible

"Because of the consequentiality of error or failure, the organization cannot easily make marginal trade-offs between capacity and safety. In a deep sense, safety is not fungible" (Rochlin, 1993, p. 16). What this means is that there is a point at which high reliability institutions are simply not able to trade-off reliability for other desired attributes, such as money.⁷ Money and the like are not interchangeable with reliability; they cannot substitute for it: in short, high reliability is simply not fungible.

When this principle is applied to pastoralism, it leads to an unexpected finding. Namely, the conventional argument that pastoralists keep cattle (or other livestock) because cattle have a variety of

different functionse.g., provision of draft, milk, dung, brideswealth, etc.gets it backwards. From a high reliability perspective, it is because cattle are not fungible that cattle have multiple functions. Indeed, a signal feature of pastoralism is the pastoralist's inability to trade off livestock for other ways to achieve "reliability" in herd numbers, like bank accounts or fenced ranches. Accordingly, it is not surprising that cattle or other such livestock have taken on multiple functions as little, if anything, else can substitute for them in terms of securing the pastoralist priority of high reliability. Just as there is no substitute for the high number of safely traveled airplane passenger miles in ensuring that passengers can use airplanes for a variety of different reasons, so too there is no substitute for the high numbers of livestock in ensuring that pastoralists can rely on those livestock for a variety of different purposes. From a high reliability perspective, because cattle are maintained in a highly reliable fashion, they can have multiple functions. We return to the issue of non-substitutability below.

Limitations On Trial And Error Learning

It is commonly said that pastoralism is a product of long trial and learning, e.g., "grazing systems developed through trial and error by pastoralists have been handed down from generation to generation" (De Boer, Yazman, and Raun, 1994, p. 24). Yet the high reliability organization, according to Rochlin (1993, p. 16), "is reluctant to allow primary-task related learning to proceed by the usual modalities of trial-and-error for fear that the first error will be the last trial". While high reliability institutions do have search and discovery processes, and often elaborate ones as we saw, they will not undertake learning and experimentation that expose them to even greater hazards than they already face. They undertake learning only within the bounds that they control. As Rochlin (1993, p. 16) puts it, "Because of the complexity of both technology and task environment, the organization must actively manage its activities and technologies in real time, while maintaining capacity and flexibility to respond to events or circumstances that can at most be generally bounded". High reliability organizations "set goals beyond the boundaries of present performance, while seeking actively to avoid testing the boundaries of error" (Rochlin, 1993, p.14). Trial and error learning occurs, but in ways that avoid testing the boundary between system continuance and collapse or how to act when that boundary is breached.

Such is why the experiments with fenced ranches for pastoralists have so frequently failed. They

failed not because pastoralists were "traditional"herders who constantly search for improvements and adopt them scarcely qualify as traditional in any conventional sense. Nor did the projects fail because they were too "modern"pastoralists know a great deal about complex technologies and decisionmaking, as we have seen. Rather, fenced ranches failed because they posed a kind of trial and error learning whose "errors" threatened the very reliability of pastoralismnamely, ranches were unacceptable precisely because they insisted fewer than peakload livestock be maintained. Over and over again and quite literally, when pastoralists tried to become ranchers, their first error as pastoralists proved to be their last trial as ranchers.

A Proposed Test of the Theory

While our review has been based on secondary data sources, clearly a remarkable congruency exists between what has been described in the pastoralist literature and the major features of high reliability institutions. Nonetheless, the review has not been comprehensive. Obviously, some secondary sources are out-of-date, pastoralist societies are constantly changing (indeed that is the point of the high reliability theory), while our source material originated in very different conceptual frameworks and models (particularly that of risk aversion). A much more difficult problem is that there is no clean way to distinguish risk-averse and high reliability pastoralism in terms of overt pastoralist behavior only, e.g., in both cases, herds and herders move and their responses vary over time and space. Only detailed research of pastoralists will uncover the kinds of distinctions identified above.

That said, these limitations do suggest a way to test the applicability of high reliability theory to pastoralist behavior. We did not find quotes or references for three additional characteristics of high reliability institutions identified by organization theorists:

"Their search for performance and suspicion of quiet periods continually regenerates operational challenges even during times when they [i.e., high reliability organizations] seem to be working well" (Rochlin, 1993, p. 14). In particular, "Groups compet[e] to discover error, 'located,' or attached to different levels of hierarchy [or] different reporting issues" (La Porte, 1993, p. 7). In terms of the pastoralist literature, we would be looking for references to the effect that pastoralists rarely if ever rest in their often competitive search for relevant information or are always on the lookout for such information.

"Common access to data bases [is found in high reliability organizations]. All operating...groups [have] access to same data, so can view, review situations before and after fact. Provides roots of realistic alternative views" (La Porte, 1993, p. 3). In particular, "Information on errorsis this shared within and between the organizations?...Can it be obtained, as in air transport, or does it go to the bottom with the ship?" (Perrow, 1994, p. 218). In terms of the pastoralist literature, we would be looking for references to the effect that pastoralists regularly come together to share the same information, e.g., during the evenings, around their fires, or in assemblies.

High reliability organizations "structur[e] themselves to quickly move from completely centralized decision making and hierarchy during periods of relative calm to completely decentralized and flat decision structures during 'hot times'" (Mannarelli, Roberts and Bea, 1996, p. 84). In particular, these organizations have a "flexible delegation of authority and structure under stress (particularly in crises and emergency situations)" (Rochlin, 1996, p. 56), where "other, more collegial, patterns of authority relationships emerge as the tempo of operations increases" (La Porte, 1996, p. 64). In terms of the pastoralist literature, we would be looking for references to the effect that, in times of emergency, pastoralist elders and leaders would be delegating decision making powers to those who were dealing with the crisis first-hand.

As a test of the theory's applicability, we predict that primary research as well as a more thorough review of the secondary literature would find these features (as well those already described) to be central to much, if not most, of pastoralist livestock production.

Policy Implications of High Reliability Pastoralism

We have already discussed implications of high reliability theory for reconceiving features attributed to the risk averse model of pastoralism. There are, however, many wider policy implications of high reliability pastoralism, and we conclude by drawing them out for pastoralist development; for the key issues of pastoralist mobility, land tenure, herder risk aversion, and "overgrazing"; and for the future of pastoralism generally.

Development Implications

High reliability theorists warn against trying to use the theory to design reliable institutions (e.g., Rochlin, 1993, p. 13). There is no cookbook here. Equally important, there are very real costseconomic, organizational, personalto maintaining a high reliability organization; one simply cannot assume that all such organizations have net positive value (Rochlin, personal communication). For example, whatever one's position on overgrazing, keeping livestock as a form of "redundancy" entails

costs (if simply herd management costs) in addition to the benefits already mentioned. Boundary avoidance in trial and error learning is also not without its own costs. Other costs will become clear momentarily.

That said, development implications do follow from a high reliability perspective on pastoralism, and they are different from those currently dominating pastoralist development. For instance, commonplaces such as the following fourchosen at randommust be rethought and challenged, when the high reliability elements of pastoralism are foregrounded:

(1) "The future of pastoralism depends on the ecological restoration and sustainable utilization of the available rangelands, on the improvement of livestock productivity, and the resolution of resource conflicts in ways that facilitate the integration of pastoralism, agriculture, and silviculture in areas suitable for their coexistence" (Shanmugaratnam et al, 1992, p. 2).

Wrong. The future of pastoralists may have nothing whatsoever to do with the ecological restoration of rangelands, or livestock productivity, or the resolution of conflicts, or even the integration of pastoralist communities into the wider society (on the latter, see Roe, 1995).

(2) "Policies to sedenterize herders have hindered them from practising the most effective strategy for managing risks in areas of great environmental uncertainty: the option to move to areas of higher natural productivity in any given season or year." (UNDP, 1994, p. 39)

Wrong. If by managing risk what is meant is ensuring high reliability, then there is nothing that says sedentary pastoralists are not or cannot be highly reliable in their sedentary behavior (see Gefu and Gilles on sedentary pastoralists). While in no way justifying policies of forced sedentarization (see Johnson, 1993), the crux here is not mobility itself but developing strategies that provide forage as reliably as possible over time.

(3) "If accepted that rangelands are generally robust and resilient, future rangeland management policies should be more concerned about resolution of resource use conflicts/equity/civil security/drought preparedness and economic efficiency..." (Vedeld, 1994, p. 20)

Wrong. If the New Range Ecology and high reliability pastoralism are the case, then pastoralist-to-pastoralist links are the priority, and these may have little, if anything, to do with current resource conflicts, equity and security issues, let alone drought response (more in a moment).

(4) "Rangeland resources must be co-managed by local communities and government authorities" (Behnke et al, 1993).

Wrong. Co-management makes sense only when both the communities and government agencies concerned approximate high reliability institutions. Anything otherwise is a recipe for mismanagement (more also below).

A high reliability critique of the commonplaces that dominate thinking about pastoral development and management could easily be extended. What pastoral development should not be, however, is rarely as useful as knowing what it should be, and high reliability theory is significant precisely because it has positive implications for the directions pastoral development and management should take. Because of space limitations, three development implications are identified. None of the following recommendations is new. What is innovative, however, is that these recommendations must be the core of pastoral development strategies, where the pastoralism of interest is high reliability rather than risk-averse pastoralism:

(1) First and foremost, the central driver of any major pastoral development (e.g., encouraged by the donors and major lenders) should be those activities that promote pastoralist-to-pastoralist links. Such links, of course, are already underway, though at the periphery rather than in the center. Currently, contacts between pastoralists as individuals and in groups are promoted primarily by NGOs and through small government projects rather than by the major donors and lenders; moreover, they are never treated as the central engine of pastoralist development. That has to change when dealing with high reliability pastoralism.

The core priority of promoting pastoralist-to-pastoralist links follows directly from the major role that search, improvement and learning play in a pastoralism bounded by the avoidance of error.

Pastoralist-to-pastoralist links that facilitate such bounded search, improvement and learning can be encouraged in three ways:

· first, through intra-pastoralist innovators and innovations (e.g., diffusion of innovations developed within a given pastoralist group or community);

· second, through inter-pastoralist links focused on the so-called "traditional" pastoralist learning and innovation areas of livestock disease control, breeding, and feed supplementation, where innovators in one pastoralist community work with pastoralists from other communities, be they within or outside the region or country of concern; and

· third, through inter-pastoralist links that focus specifically on major challenges analogous to that Boeing faced when it moved from propeller engines to producing jet engines. Namely, how can pastoralists maintain a culture of high reliability when they move from "producing" livestock to undertaking other "livestock" activities, such as ostrich (or wildlife) farming, ecotourism, or other income-generating innovation in livestock production and utilization.⁸

Note that the kind of links we have in mind are much more varied than the popular wisdom that pastoralists should get to know how other pastoralists run their common property institutions. How one actually designs specific pastoralist-to-pastoralist links will vary from case to case (again, there are no recipes here; in one case that focus may be on veterinary care, in another, wildlife management). Clearly, pastoralist associations, which are the focus of more and more pastoralist development efforts (e.g., de Haan, 1994), could facilitate some of the links we recommend. Nevertheless, a high reliability perspective suggests having a specific pastoralist agency (ministry, government initiative, national commission) whose goal is to ensure that all key pastoralist development projects focus primarily if not exclusively on promoting pastoralist-to-pastoralist contacts.⁹

(2) Another important development implication follows from the fact that pastoralists are not the only high reliability institutions in arid and semi-arid environments. Each of the nine features for high reliability could well have been found in the Government of Botswana's Veterinary Department and the Botswana Meat Commission in the 1970s and early 1980s. Agropastoralism in Botswana and these Government of Botswana operations were very similar at that time, notwithstanding the conventional view that the former was "traditional," the latter "modern."

Why is this important? Because there is nothing "exotic" about pastoralism (see also Gefu and

Gilles, 1990). What is of note is the high reliability, given their complex technologies and uncertain task environments. That is, what really links activities like pastoralism, veterinary departments and abattoirs is their commitment to highly reliable peakload production, not that they all have something to do with livestock or are lumped together in something called "the livestock sector."

Why is that important? Because without that commitment to high reliability there can be no real "livestock sector"which explains why Botswana had one in a way that was manifestly not the case in Kenya during the same period, when the Kenya Meat Commission was in a shambles and the Government of Kenya had veterinary control problems in pastoralist areas of the country. There are times when the best thing one can do to improve the reliability of pastoralism is to stop trying to get pastoralists to be "more modern" and instead improve the reliability of linked institutions such as vet departments and meat commissions. If these latter institutions cannot be made reliable in their own peakload throughputs and if the objective is to keep pastoralism as a highly reliable institution in those areas in which it operates (an objective we commend), then it is best to get rid of the latter institutions altogether in the areas concerned.

(3) The third development implication of a high reliability perspective runs counter to the popular recommendation that pastoralists require common property tenure regimes, because such tenure arrangements facilitate herd mobility and movement. In high reliability terms, the relevant issue is not that, but rather the fact that the more experience high reliability institutions have with different operating scales, the more reliably these institutions can perform. There is an access issue here, but it is one of ensuring access to scale-dependent resources. From a high reliability perspective, what is threatening about loss of grazing area due, e.g., to agricultural encroachment, is not that it restricts herd mobility as such but that it reduces the experience of herders with managing herds at different scales of operation.¹⁰ Reduce our experience with operating scales and you threaten our ability to perform in a highly reliable fashion.

One herder response to a reduction in operating scale has been to compensate for that loss by accessing other scale-dependent resources hitherto un(der)-accessed. For example, the herders' response may be a shift to confined feeding of stock at selective times of the year, using improved grasses grown

in plots near the herder's compound or gardens. Whatever the case, the land tenure issue cannot be one simply of ensuring land remains common property, as prevailing wisdom would have it. Grazing land may better remain common property if it is not encroached; but one cannot argue in the same breadth that the plot of improved grasses should as well remain common property. The better argument is that such plots should be private property. The issue is really one of access to scale-dependent resources, not that all scale-dependent resources should be common property, or for that matter, private property. Different scale-dependent resources will have different access requirements, and therefore need different tenure regimes.

Little in the three recommendations should be surprising, but taken together they constitute a new flashpoint not only in the way development practitioners think about pastoralism, but also in the way they undertake that development. In this time when many larger donors have turned away from anything to do with livestock and pastoralists, especially in Africa, it is important they know they have turned away far too early and with much yet that can be done, albeit what needs to be done is not what they have been trained to do in the past.

Implications for the Key Pastoralist Issues of Mobility, Land Tenure, Herder Risk Aversion and "Overgrazing"

Four themes dominate the literature on risk-averse pastoralism: the need for herder mobility, the need for stable and secure common property tenure regimes supporting that mobility, the herder risk aversion said to be driving those needs, and the "overgrazing" that arises when trying to meet those needs.

Mobility. From the perspective of a high reliability institution, herd and herder mobility is important, but only because it is a scale- and phase-specific routine comparable to the standard operating procedures for bringing in airplanes, providing hospital intensive care, or supplying electricity over a power grid. Mobility is not just movement; it is also defining and staking out the operating scales at which high reliability is to be achieved over the critical phases of pastoralist production. To reiterate, risk-averse pastoralism holds that, when pastoralists move from one part of the range to another over the course of the year, they are escaping dry conditions and need a large area to provide such a retreat in or

der to minimize the effects of or otherwise avoid altogether the hazards they cannot afford to risk. High reliability pastoralism holds that in making such movements, pastoralists are managing the spatial and temporal diversity of their operating scales and enhancing the reliability of high levels of production by using different resources over these scales at different times in response to hazards that must be risked. Mobility is not just about herds moving where; it is also about managing the where so that herds can move.

For example, a risk averse perspective would focus on how the Batswana shift their livestock watering over the course of a year from ephemeral, surface water sources during the rainy season to more permanent wells and boreholes late in the dry season. From this viewpoint, herders are retreating from surface water sources as they dry up and falling back to the fewer, year-round water points. In contrast, the high reliability perspective focuses on the multiple levels of water use and management governing any given specific water point in that fallback system: namely, the site immediately surrounding the water point, the locality in which the water point is found, and the compound locality in which the water point is located (that is, the set of different localities over which the users of that water point typically reside and work during the year).

When rural Batswana are physically at the water point, they are keenly aware of the physical condition of adjacent land as well as the physical type of water source in question (e.g., boreholes, because they are mechanized, are managed differently than hand-dug wells). At the locality level, how the water point is managed and used is affected by a host of factors, including the availability of alternative water points in the locality; the availability of labor for fetching water, which varies by locality (some household members move to the cropping fields for planting, while others, such as children who would otherwise fetch water, remain in village schools); and by the locality's prevailing land uses (e.g., villages and grazing areas are typically dominated by borehole development, while mixed cropping and grazing areas have had a greater variety of sources).

Moreover, because members of rural households in Botswana have often shifted their household compounds over the course of the yearin the cropping season, they have gone to the "lands" where their field homes are and, after harvest, they have returned to their village residences for the rest of the yearthe demand and supply of water have shifted as well over the course of the year across the locali

ties concerned. Indeed, it is at compound locality level that the specific water point operates as part of the fallback water point system, with herders falling back from the many surface water sources in the lands to the fewer permanent water sources in the village. In short, what is being reliably managed by the herders is not only a fallback water point system, but the production system(s) as well. The importance of such spatial and temporal diversity of multiple operating scales and phases key to ensuring the reliability of high levels of production thus becomes much more apparent and central in the high reliability model than in the risk averse one.

Land tenure. The problem with tenure is the reverse of the problem with mobility. While there has been too much fixation on movement only as movement in pastoralism, the preoccupation with tenure has been to focus too much on stability and security¹¹ to the exclusion of negotiation and change (on the importance of negotiation in high reliability institution, see Schulman, 1993, p. 44). Land tenure arrangements are important not because they provide stability, but because they are negotiated and changed in ways that realize highly reliable behavior over critical scales and phases needed to ensure that reliability.¹² The accent on negotiation follows from the fact that highly reliable behavior requires continual search, improvement, and informal mechanisms of oversightthree concepts at odds with a tenure that is prized simply because its rules are said to be secure, public and purportedly "tried and tested" (UNDP, 1994, p. 42) in promoting mobility .¹³

Herder Risk Aversion. As for the preoccupation with risk aversion, it is not difficult to see why pastoralist behavior has been so uniformly described as risk averse. For many the antithesis of reliability is risk (Schulman, 1993, p. 34), and, in practice, both risk aversion and reliability seeking may be going on together, e.g., as when Mace (1990, p. 2) writes of pastoralists "having to decide how to invest their livestock wealth between slow-breeding but relatively reliable and potentially fast-breeding but risky species" (in her case, camels and smallstock respectively). Similarly, much has been made in the literature about the risk-spreading practices of pastoralists who loan their livestock out to others (e.g., Dahl and Hjort, 1976, 1979; Sandford, 1983a); what also is going on is that in so doing pastoralists are learning about the different operating scales and critical phases over which herds can be managed more

reliably.

That said, it should be clear by this point that reliability-seeking behavior, as described through its nine principal features, is very different from the conventional view of risk-averse pastoralism. Behavior that is developed around high technical competence and highly complex activities, needs sustained high levels of performance, oversight and flexibility, is continuously searching for improvements, maintains great pressures, incentives and expectations for safety, and establishes a virtual culture predicated on maintaining peak (not minimum) livestock numbers in a highly reliable fashion is not what we customarily think of as "risk-averting" behavior. Again, apart being averse to catastrophic risks, high reliability pastoralism is characterized in its most important respects by risk-accepting, if not risk-taking, behavior.¹⁴ Similarly, as developed above and summarized in Table 1, the contrast between risk-averse and high reliability pastoralisms, at least as ideal types, could not be starker on virtually all the key dimensions.

"Overgrazing." Last but not least, what about overgrazing and the putative land degradation caused by pastoralism? It is the singular failing of the vast literature on pastoralism that so much of it starts out by assuming "overgrazing and degradation" are the issue to be explained.¹⁵ In writing this article, we put an embargo on even talking about overgrazing until after we had thought out and articulated the high reliability model of pastoralism. Our assumption was that, if we had anything new to say about overgrazing, it should follow from our approach, rather than be a prior assumption to that approach.

It turns out that we have something new to say. As noted earlier, the most commented-upon finding of the New Range Ecology is that it is difficult, if not impossible, to define the carrying capacity of a given rangeland. This does not mean, however, that there is no carrying capacity for pastoralists. There is. In a recent article one of us (Roe, 1997) argues that carrying capacity is really a theory (or theories) of knowledge generation and change for the given rangeland population and area. One such theoryand the one we commend to all readershas been presented in this paper, namely, that of high reliability theory. The upshot here is that while there is no carrying capacity of the range, there is the carrying capacity of high reliability pastoralism.¹⁶ What sets the carrying capacity of this pastoralism are the limits

on peakload production already described for high reliability pastoralism: Knowledge about the livestock production on a given rangeland is generated through pastoralist experience of working at different operating scales, through different critical phases of production, and under the pressures that limit trial and error learning and put a premium on ensuring reliable production through thick and thin. Carrying capacity is defined, in other words, as the limits on peak production derived through the long familiarity and experience pastoralists have with the range over which they operate. While the limits on what the rangeland itself can support are as important as the amount of electricity the power lines can take, the number of beds in the intensive care unit, and the level of contamination that water in the nuclear reactor can support, these factors only become limits by virtue of being part and parcel of the complex production technology that drives the institution in question to be a highly reliable one. Sustainability matters, but it is sustainability of the production system that ensures the sustainability of the land in high reliability pastoralism.

Thus, we are not certain if the "overgrazing issue" makes any sense whatsoever as currently formulated. We are certain, however, that any scenario about pastoralist overgrazing of rangelands has zerorepeat, zeropolicy relevance in the absence of its demonstrating what activities should be in place so that pastoralists can manage whatever they manage in as reliable a fashion as they currently are managing their livestock. Moreover, such demonstrations are most likely to evolve, if at all, through the pastoralist-to-pastoralist links recommended above. Or to put it another way: If overgrazing is important as a major threat to the rangeland, then as a threat it literally has no meaning until tied directly to the carrying capacity of high reliability pastoralism being studied. As we shall see in the next and final section, there is a threat to high reliability pastoralism that strikes at the heart of its capacity to produce peak herd sizes reliably over timeand that threat is not overgrazing.

Implications for the Future of Pastoralism

The past fifteen years have witnessed a major shift in prevailing views about pastoralists, initiated in great part by Stephen Sandford's path-breaking and path-setting Management of Pastoral Development in the Third World (1983). In the past, though less so today, the Mainstream View held that pastoralism had many problems and what was wrong was by and large internal to pastoralism itself. The

"endogenous" perspective argued, among other things, that pastoralist systems put a premium on the short-run objective of accumulating as many livestock as possible and on using traditional communal grazing systems, both of which, in this view, led to a disregard of longer-term considerations of rangeland conservation and environmental enhancement. What happened over the last fifteen years has been a shift to a more "exogenous" perspective on pastoralism, where rangeland deterioration has been explained by factors largely external to the pastoralist systems, including agricultural encroachment on wet-season grazing, ill-conceived government policies (e.g., nationalizing rangeland, enforcing pastoralist resettlement), and a highly variable climate that makes it very difficult to determine just what is rangeland "deterioration" and the role that livestock numbers may have in "it." In crude terms, the older Mainstream View held that pastoralists were acting irrationally because of their preoccupation with the short-run, while the newer view holds that pastoralists are actually acting rationally in the face of factors largely outside their own control. The risk averse model of pastoralism is never more inadequate than when used, as has too often been the case, to support the endogenous and exogenous perspectives both at the same time.

The model of high reliability pastoralism pulls us back to an endogenous perspective on pastoralism and thus runs counter to much of the exogenous perspective developed by the revisionist literature of the last two decades. In a nutshell, the benefits of high reliability pastoralism are precisely its dangers. La Porte (1996, p. 67) summarizes the problems generally:

"[High reliability organizations] are at once a source of benefit and worry. As the benefits become crucial and the potential damage from mis-steps becomes grave, the difficulties of maintaining public trust and confidence grow. . .The degree of difficulty will, in large part, be a function of the following conditions. . .

· Operations are beneficial but hazardous in their design, that is, the work is intrinsically dangerous;

· hazards are evident and likely to extend well after the benefits have been gained;

· the benefits of the production system have already accrued to past and present generations with high costs still to be borne by future ones;

· overall success or failure of the operations is hard to determine for several work generations;

· there is reasonably rapid change in the technical aspects of the work, the core technologies, or information about the environment where it is deployed; and

· there is hostility to current or future operations based on learning from past. . .practices.

These conditions, along with ever present competition for resources generally, combine to re-inforce the sense of public dependence on the skills and integrity of managers and operational leaders. This, in effect, intensifies the public's perceived vulnerability and their hopeperhaps against hopethat organization leaders are worthy of the public's trust and confidence".

The difficulties for high reliability institutions overall have a remarkable fit for pastoralist institutions as well. Yes, outside factors such as encroachment and government policy have made pastoralist operations even more hazardous than before (the first bullet above), but it is because pastoralism is already a very dangerous occupation that these factors come to pose even more hazards. To continue with the bullets, the benefits of pastoralism ebb and wane, but hazards persist; past generations have clearly benefited from pastoralist practices in ways that future generations might well not; as it has been demonstrated again and again, only after a lengthy period can the success or failure of pastoralism practices be determined (e.g., with respect to assessing livestock-induced changes in range condition); and the technical core of pastoralism has been changing with the increasing adoption by pastoralists of modern veterinary care and the cash economy, among others changes. Perhaps the one threat that embodies most of these concerns is pastoralists expanding their peak herd sizes to a point where irreversible collapse of the forage, water or other input resource is threatened (see Ludwig, Hilborn and Waters, 1993)though this concern is precisely why high reliability pastoralism resists such boundary testing and insists on constant oversight of resource and production conditions. To repeat, there is a carrying capacity at work, albeit of high reliability pastoralism itself and not of the land per se.

Finally and most important for our purposes, high reliability pastoralism is profoundly preoccupied with the short run rather than the past immediately behind or the longer-term ahead. As Schulman (1996, p. 74) puts it for high reliability institutions generally: "The organization is only as reliable as the first incident in front of it, not the many successful operations behind it." High reliability pastoralism, in the newer endogenous view, cannot afford not to be preoccupied with the short-term, as so much is riding on reliably maintaining peak herd sizes. Accordingly, irrationality is an important part of this newer endogenous view as well: It would be incredibly irrational and irresponsible of pastoralistsjust

as it would be for hospital intensive care units, nuclear power plants, air traffic controllers, and large utility corporationsnot to be fixated on their short-run performance. Indeed, we would be irrational to accept anything less in a world that is populated, as this one is, by hospitals, power lines, airplanes and the millions upon millions of pastoralist livestock all over the globe.

The good news, in conclusion, is that high reliability pastoralism fills a unique niche in the world's arid and semi-arid lands. To adopt Demchak's description of high reliability organizations (1996, p. 97): "In general, their functions are non-substitutable and considered socially essential, forming a natural monopoly. . ." The bad news is that this natural monopoly for making what has been in many cases the best human use of an otherwise inhospitable landscape is itself a highly precarious niche. It is precarious because what the exogenous perspective takes to be external factors, such as highly erratic climate and rainfall, are in fact endogenous factors core to pastoralist technologies and production systems. Hazards do not "happen" to pastoralists; they are built into pastoralist production by virtue of their accumulating and keeping peak herds in such a way that high-probability hazards would pose even more catastrophic consequences for pastoralists, were they not able to reduce the probabilities by managing their livestock safely over time and space. Drought is all but inevitable, but pastoralists are the ones who make it a hazard whose consequences have to managed, for better or for worse.

In this way, the great betrayal of many, if not most, government and donor-initiated policies and projects for pastoralists has been that they have been killing pastoralism from the inside. Fenced ranches, grazing schemes, resettlement projects, and other rangeland reforms, combined with the unwillingness and inability of governments to stop the whittling away of pastoralist areas due to outside incursionsagricultural, security or otherwisehave served only to up the stakes for pastoralism and make it even more hazardous, posing new risks that pastoralists must try to manage but find all the more difficult to surmount. It is like taking workaholics who are already producing to the limit, and then knowing they will continue to try to produce that much after you have ensured that they have even less to work with than they had before. The real threat here, pace La Porte and other high reliability theorists, is that pastoralists themselves will lose trust in their own capacity to respond reliably to such misguided interventions.

So what is the future of pastoralism? The answer, this article insists, depends profoundly on just

what pastoralism we are talking about. If pastoralists are only risk averse in the conventional sense, then the future is bleak: It depends on giving back to pastoralists the mobility and tenure arrangements that, quite frankly, will not be given back to them. If pastoralists are highly reliable, then the future is less bleak, at least to the extent that they have significant things to learn from each other by way of improving their changing operations. Either way, the future of pastoralism is precarious. Understanding why this is so is today's real challenge for the environment and development of the world's arid and semi-arid lands.

Acknowledgments

The authors would like to thank.....

Bibliography

Abercrombie, Frank D., Range Development and Management in Africa (Washington, D.C.: Office of Development Services, Bureau for Africa, United States Agency for International Development, 1974).

Agrawal, Arun, "The Grass Is Greener on the Other Side: A Study of Raikas, Migrant Pastoralists of Rajasthan" (London, England: IIED Drylands Issue Paper, 1992).

Behnke, Roy H., Ian Scoones, and Carol Kerven, Range Ecology at Disequilibrium. (London, England: Overseas Development Institute, 1993).

Bonte, Pierre, "Ecological and Economic Factors in the Determination of Pastoral Specialization," in Galaty, John and Philip Carl Salzman, Change and Development in Nomadic and Pastoral Societies (Leiden, The Netherlands: E.J. Brill Publishers, 1981), pp. 33-49.

Dahl, Gudrun, and Anders Hjort, Pastoral Change and the Role of Drought, SAREC Report R2 (Sweden: Swedish Agency for Research Cooperation with Developing Countries, 1979).

Dahl, Gudrun, and Anders Hjort, Having Herds (Stockholm, Sweden: University of Stockholm, 1976).

De Boer, A. John, James A. Yazman, and Ned S. Raun, Animal Agriculture in Developing Countries: Technology Dimensions, Development Studies Paper Series (Morrilton, Arkansas: Winrock International Institute for Agricultural Development, Feb. 1994).

de Haan, Cornelis, "An Overview of the World Bank's Involvement in Pastoral Development," Pastoral Development Network Paper 36b (London, England: Overseas Development Institute, 1994) pp. 1-6.

Demchak, Chris, "Tailored Precision Armies in Fully Networked Battlespace: High Reliability Organizational Dilemmas in the 'Information Age'," Journal of Contingencies and Crisis Management, Vol. 4, No. 2 (June 1996), pp. 93-103.

Dyson-Hudson, Rada, "Ecology of Nomadic Turkana Pastoralists: A Discussion," in Whitehead, Emily et al, Arid Lands Today and Tomorrow, the proceedings of a conference (Oct. 20, 1985), held in Tucson, Arizona (Boulder, Colorado: Westview Press, 1988) pp. 701-702.

Dyson-Hudson, Rada, and Neville Dyson-Hudson, "Nomadic Pastoralism," Annual Review of Anthropology, No. 9, (1980) pp. 15-61.

Ellis, Frank, Peasant Economics (Second Edition, Cambridge, Great Britain: Cambridge University Press, 1993).

El Wakeel, Ahmed, and Mohamed Azim Abu Sabah, "Relevance of Mobility to Rangeland Utilization,"

Nomadic Peoples, No. 32 (1993) pp. 33-38.

Fortmann, Louise, "Peasant and Official Views of Rangeland Use in Botswana: Fifty Years of Devastation?" Land Use Policy (July 1989) pp. 197-202.

Fortmann, Louise, and Emery Roe, The Water Points Survey, Report (Gaborone, Botswana: Ministry of Agriculture and center for International Studies, Cornell University, 1981).

Galaty, John G., and Douglas L. Johnson, editors, The World of Pastoralism (New York: Guilford Press, 1990).

Gefu, Jerome O., and Jere L. Gilles, "Pastoralists, Ranchers and the State in Nigeria and North America," Nomadic Peoples, No. 25-27 (1990) pp. 34-50.

Gilles, Jere L., and Cornelis de Haan, "Recent Trends in World Bank Pastoral Development Projects: A Review of 13 Bank Projects in Light of the 'New Pastoral Ecology'," Pastoral Development Network Paper, No. 36b (London, England: Overseas Development Institute, 1994) pp. 7-14.

Gilles, Jere, "The World Bank and Pastoral Development", Pastoral Development Network Paper, No. 36b (London, England: Overseas Development Institute, 1994) pp. 15-16.

Hobbs, Joseph, Bedouin Life in the Egyptian Wilderness, (Austin, TX: University of Texas Press, 1989).

Johnson, Douglas L., "Nomadism and Desertification in Africa and the Middle East," Geojournal, No.31.1 (September, 1993) pp. 51-66.

Kuznar, Lawrence, "Transhumant Goat Pastoralism in the High Sierra of the South Central Andes: Human Responses to Environmental and Social Uncertainty," Nomadic Peoples, No. 28 (1991)pp. 93-104.

Lancaster, William, and Fidelity Lancaster, "The Concept of Territory Among the Rwala Bedouin," Nomadic Peoples, No. 20 (March, 1986) pp. 41-48.

La Porte, Todd, "High Reliability Organizations: Unlikely, Demanding, and At Risk," Journal of Contingencies and Crisis Management, Vol. 4, No. 2 (June 1996), pp. 60-71.

La Porte, Todd, "'Organization and Safety in Large Scale Technical Organizations': Lessons from High Reliability Organizations Research and Task Force on 'Institutional Trustworthiness," paper prepared for a seminar on Man-Technology-Organization in Nuclear Power Plants, Finnish Centre for Radiation and Nuclear Safety, Technical Research Centre of Finland, Olkiluoto, Finland (June 14-15, 1993).

La Porte, Todd and Paula Consolini, "Working in Practice But Not in Theory: Theoretical Challenges of 'High Reliability Organizations,'" Journal of Public Administration Research and Theory, Vol. 1, No. 1 (1991) pp. 19-47.

Ludwig, Donald, Ray Hilborn, and Carl Waters, "Uncertainty, Resource Exploitation, and Conservation: Lessons from History," Science, Vol. 260 (2 April) pp. 17, 36.

Mace, Ruth, "Pastoralist Herd Compositions in Unpredictable Environments: A Comparison of Model Predictions and Data from Camel-Keeping Groups," Agricultural Systems, No. 33 (1990) pp. 1-11.

Manger, Leif, "Managing Pastoral Adaptation in the Red Sea Hills of the Sudan: Challenges and Dilemmas, " Dryland Networks Programme Issues Paper, No. 52 (London: International Institute for Environment and Development, September, 1994).

Mannarelli, Thomas, Karlene Roberts and Robert Bea, "Learning How Organizations Mitigate Risk," Journal of Contingencies and Crisis Management, Vol. 4, No. 2 (June 1996), pp. 83-92.

Oba, Gufu, "Perception of Environment Among Kenyan Pastoralists: Implications for Development," Nomadic Peoples, No. 19 (September, 1985) pp. 33-57.

Perrow, Charles, "The Limits of Safety: The Enhancement of a Theory of Accidents," Journal of Contingencies and Crisis Management, Vol. 2., No. 4 (December, 1994) pp. 212-220.

Reckers, Ute, "Learning from the Nomads: resource and Risk Management of Nomadic Pastoralists," U.N. Desertification Control Bulletin, No. 24 (1994) pp. 48-53.

Roberts, Karlene, "Some Characteristics of High Reliability Organizations," School of Business Administration, University of California at Berkeley (February, 1988).

Rochlin, Gene, "Reliable Organizations: Present Research and Future Directions," Journal of Contingencies and Crisis Management, Vol. 4, No. 2 (June 1996), pp. 55-59.

Rochlin, Gene, "Defining 'High Reliability' Organizations in Practice: A Taxonomic Prologue," in Roberts, Karlene, New Challenges to Understanding Organizations (New York, NY: Macmillan Publishing Company, 1993), pp. 11-32.

Roe, Emery, "On Rangeland Carrying Capacity," Journal of Range Management, Vol. 50, No. 5 (September, 1997) pp. 467-472.

Roe, Emery, "More Than the Politics of Decentralization: Local Government Reform, District Development and Public Administration in Zimbabwe," World Development, Vol. 23, No. 5 (May 1995).

Roe, Emery, "Public Service, Rural Development and Careers in Public Management: A Case Study of Expatriate Advising and African Land Reform," World Development, Vol. 21, No. 3 (1993).

Sagan, Scott, The Limits of Safety: Organizations, Accidents, and Nuclear Weapons (Princeton, NJ: Princeton University Press, 1993).

Sandford, Stephen, Management of Pastoral Development in the Third World (Chichester, England: Wiley, 1983a).

Sandford, Stephen, Organisation and Management of Water Supplies in Tropical Africa, (Addis Ababa, Ethiopia: International Livestock Centre for Africa, 1983b).

Schulman, Paul, "Heroes, Organizations and High Reliability," Journal of Contingencies and Crisis Management, Vol. 4, No. 2 (June 1996), pp. 72-82.

Schulman, Paul, "The Analysis of High Reliability Organizations: A Comparative Framework," in Roberts, Karlene, New Challenges to Understanding Organizations (New York, NY: Macmillan Publishing Company, 1993), pp. 33-53.

Scoones, Ian, Living With Uncertainty: New Directions for Pastoral Development in Africa, Drylands Paper No. 6 (London: International Institute for Environment and Development, 1994).

Shanmugaratnam, Nadarajah, Trond Vedeld, Anne Mossige, and Mette Bovin, "Resource Management and Pastoral Institution Building in the West African Sahel," Discussion Paper No. 175 (Washington, D.C.: The World Bank, 1992).

Spooner, Brian, "The Cultural Ecology of Pastoral Nomads," Addison-Wesley Modules in Anthropology, No. 45 (Reading, Massachusetts: Addison-Wesley, 1973).

Swallow, Brent, "The Role of Mobility Within Risk Management Strategies of Pastoralists and Agropastoralists," in Gatekeeper Series, No. 47 (London: International Institute for Environment and Development, 1994).

United Nations Development Programme, "Pastoral Natural Resource Management and Policy," (New York: UNDP, 1994).

Vedeld, Trond, "Procedural Law: Land Tenure Reforms as a Long-Term Political Process," Pastoral Development Network Paper, No. 36b (London: Overseas Development Institute, July, 1994) pp. 17-22.

Walker, Brian H., "Rangeland Ecology: Understanding and Managing Change'" Ambio, Vol. 22, No. 2-3 (May 1993), pp. 80-87.

Western, David, and Thomas Dunne, "Environmental Aspects of Settlement Site Decisions Among Pastoral Maasai." Human Ecology, Vol. 7, No. 1 (1979), pp. 75-98.

Westoby, Mark, Brian Walker, and Imanuel Noy-Meir, "Opportunistic Management for Rangelands Not at Equilibrium," Journal of Range Management, Vol. 42, No. 4 (July, 1989), pp. 266-274.

World Bank, Economics and Policy Division, Department of Agriculture and Rural Development, Livestock Strategy Paper, draft (Washington, D.C.: World Bank, 1987).

¹ There, of course, have been criticisms of the ecological approach by some, e.g., the critique of Spooner (1973) by Dyson-Hudson and Dyson-Hudson (1980). For another critique of the ecological approach, see Bonte (1981).

² For example: "Whether movement is regular and seasonal, contingent, or a combination of contingency and regularity, the producer's strategy within non-equilibrium systems is to move livestock sequentially across a series of environments each of which reaches peak carrying capacity in a different time period." (Behnke et al, 1993, p. 11)

³ Obviously, high reliability organizations may attempt to control or manage the hazards directly, but more often than not their behavior is organized around trying to reduce or otherwise favorably affect the probabilities (Rochlin, 1993, p. 19).

⁴ Other critical phases include breeding, weaning and culling cycles.

⁵ The current view among many experts runs directly counter to this conclusion. According to one long-time observer (Gilles, 1994, p. 16): "Droughts continue to be regarded as 'acts of God' which are not part of the normal operation of semi-arid ecosystems. [Such traditional] understandings of resource management prevent the development of a means for systematically addressing drought and drought management."

⁶ For more on the importance of flexibility (sometimes called, opportunism or opportunistic management) in pastoralism, see UNDP (1994), Scoones (1994), Gilles (1994), and Sandford (1983a).

⁷ As Schulman (1993, pp. 34-35) puts it: "Reliability demands are so intense, and failures so potentially unforgiving, that only a sharply reduced amount of trial and error learning about causal relationships is permitted. Managers are hardly free to reduce investments and arrive at conclusions about the marginal impacts on reliability". We discuss the limitations on trial-and-error learning in the next subsection.

⁸ The challenge of keeping pastoralist high reliability in tact while such basic production changes are underway is akin to the well-known paradox in philosophy, i.e., "We've had this hammer in our family for so long we've lost count of all the times we've had to change its head and its handle!"

⁹ There, of course, are ecologists and other technicians who would insist that pastoralist-to-pastoralist diffusion of livestock and herding innovations could worsen rangeland conditions considerably (see our next section). Nonetheless, the fact that pastoral networking is a growing focus of the major donorse.g., World Bank, GTZ, Oxfamis very encouraging from the high reliability perspective.

¹⁰ The argument that agricultural encroachment is limiting the effectiveness of pastoralists bears a striking, albeit uncomfortable, resemblance to the argument that pastoralists are overgrazing their rangelands. For over fifty years now, veterinary and livestock officers in Africa have been saying that herders are seriously overstocking and degrading their rangelands, all the while the herders have been managing to increase their numbers and the numbers of their livestock (see, e.g., Fortmann, 1989). In parallel fashion, for over fifty years now people in Africa have been saying that pastoralists are seriously endangered by agricultural encroachment of their rangelands, all the while pastoralists have been managing to increase their numbers and the numbers of their livestock (see, e.g., Dahl and Hjort, 1979; Scoones, 1994). Both cases, rather than portending the imminent collapse of pastoralism, have demonstrated the high reliability resilience of pastoralists in the face of real and persisting adversities.

¹¹ With respect to the stability argument: "Spatial demarcation with secure, defensible usufruct rights is a basic condition for the sustainable management of the rangeland..." (Shanmugaratnam et al, 1992, p. 8). The importance of common property regimes based upon "well-established" pastoralist institutions is stressed by UNDP (1994, p. 16).

¹² On the negotiation of tenure rules, see UNDP, 1994. On the existence of flexible tenure regimes, see Scoones, 1994 (also Vedeld, 1994).

¹³ Needless to say, privatizing rangeland could (though not necessary would) interfere with a habitually renegotiated commons. That said, national policies and initiatives, which were to have led to privatization of rangelands, have not always in practice done so, i.e., privatization itself can be a negotiated tenure regime (see Roe, 1993).

¹⁴ Though high reliability behavior has nothing to do with and is in fact orthogonal to the-press-on-regardless-technique of relying on luck and chance that Brian Walker (1993, p. 87) seems to recommend in some livestock herding situations under arid and semi-arid conditions.

¹⁵ For example: "A basic parameter in the working of a pastoral system is the relation between carrying capacity for the area and demographic practices, i.e. the growth rate of man and animals. The basic issue is whether the stocking rate exceeds, is in balance with or below the theoretical carrying capacity" (Manger, 1994, pp. 5-6).

¹⁶ A commendable counterweight to static carrying capacity estimates of one beast per x hectares has been the notion of opportunistic strategies of pastoralists for adjusting livestock numbers to variable forage conditions. The notion, however, has the problem of tying these strategies to limits on what the land can support rather than limits set by pastoralism itself.