The importance of complexity in Ostrom’s work

Complexity is one of the main notions structuring Ostrom’s work. However, it is distinctive in that it characterises a methodological approach more than a research object as such. Indeed, if we assume institutional situations are ‘complex’ systems, then it follows that we need methodological tools to deal with them. This is how Ostrom conceives the role of ‘complexity sciences’ in social sciences:

Lewis (Reference Lewis2017) indicates that Ostrom’s notion of complexity includes a series of characteristics that she endeavoured to define and understand throughout her career. According to him, a complex system in Ostrom’s work should be comprehended as: 1) a set of elements (human beings, natural resources, cultural communities, etc.) interacting according to a structure of rules; 2) situated within a series of nested structures; 3) adaptive to its environment; 4) and producing emergent properties that are difficult to predict and irreducible to the properties of its constituent parts.

Ostrom’s recognition of complexity came gradually. The theme began to be explicitly recognised by her and her husband in the 1980s, when the development of the Institutional Analysis & Development (IAD) framework was her principal focus (Ostrom, Reference Ostrom1986a, Reference Ostrom, Kaufmann, Majone and Ostrom1986b; see Blomquist, Reference Blomquist, Foster and Swiney2022).Footnote ³ In January 1986, in the Ten-Year Prospectus of the Workshop in Political Theory and Policy Analysis they had created in Bloomington, they stressed the problem posed by the complexity confronting social scientists: ‘From our cumulative experience of working with theory, collective choice, and empirical consequences, we have become increasingly aware that the phenomena being investigated involve extraordinary complexity requiring multiple levels of foci of inquiry’ (E. Ostrom and V. Ostrom [1986] in Clark, Reference Clark2019: 102). Then, from the 1990s onwards, Ostrom focused her writings ever more on how to deal with complex systems. For example, while the theme is only touched upon in her seminal 1990 book Governing the Commons, her 1995 essay ‘Designing Complexity to Govern Complexity’ explicitly sets out how her approach and results from 1990 address it. She states:

From then on, the notion of complexity was to appear explicitly across all of Ostrom’s work in subsequent years. Since her primary concern had always been to build a ‘theoretical framework for understanding institutions and collective action’ across a variety of situations (Blomquist, Reference Blomquist, Foster and Swiney2022: 22), the development of tools and methods to address complexity was central to this aim, preceding in fact her research on CPR situations.Footnote ⁵ It is therefore no surprise that, towards the end of her career, Ostrom continued to assert that, ‘[g]iven the complex, nested systems of the biophysical world’, we still need to further develop ‘a social science of complex, nested systems’. (Ostrom and Basurto, Reference Ostrom and Basurto2011: 337).

For Ostrom, the notion of complexity thus refers more to a global, structuring methodological approach than to a research object as such. It remains to characterise in more detail the origins and attributes of the tools that she associates with it. This is precisely where the contribution of Simon is decisive.

Ostrom’s Simonian source of inspiration for the study of complex systems

Most of the commentators who link the two social scientists do so via the notion of bounded rationality (Chanteau and Labrousse, Reference Chanteau and Labrousse2013; Clark, Reference Clark2019; Tarko, Reference Tarko, Lemke and Tarko2021; McGinnis, Reference McGinnis, Lemke and Tarko2021). Of course, on several occasions, Ostrom explicitly asserts the decisive role Simon played regarding her commitment to bounded rationality (Ostrom, Reference Ostrom2005, Reference Ostrom2010a). In one of her final writings, she states:

As we can see, Ostrom also takes a step further by explicitly acknowledging that it is thanks to Simon that field research came to occupy such a central place in her work.Footnote ⁶

Nevertheless, it is in a short essay reflecting on the ten books that most influenced her that Simon’s role becomes most apparent (Ostrom, Reference Ostrom2004). Indeed, while it is not surprising that The Sciences of the Artificial (Simon, [1969] Reference Simon1996) features in the ranking, it is more striking to see that no mention of bounded rationality is made when justifying this choice:

As Ostrom explicitly states, it is Simon’s reflections on systems science that seem to matter most for her work. It is also interesting to note that in the same list, Ostrom cites Mayr’s book The Growth of Biological Thought (1982) to emphasise the systems approach to complexity developed there. Indeed, the approach used by Mayr (Reference Mayr1982: 53) is partly influenced by Simon himself. Considering Simon’s thinking on systems science and complexity, it thus appears that his influence on Ostrom’s work is more substantial than the notion of bounded rationality alone would suggest.

Some commentators have begun to highlight the role of Simonian ideas on systems and complexity in Ostrom’s work. Wall (Reference Wall2014) points out, for example, that Ostrom drew on Simon’s concept of ‘nearly decomposable system’ to extend the IAD to SES. Peneranda (Reference Peneranda2012) writes that the engineering approach developed in Simon’s work may have influenced Ostrom in elaborating the ‘design principles’. However, one feels that these insights focus primarily on explaining particular operational tools, giving only a fragmented and embryonic vision of what Ostrom borrows from Simon on systems and complex thinking. It perhaps explains why, while Ostrom (Reference Ostrom2005: 270) argues that her research on complex, adaptive systems is ‘strongly influenced by the work of Simon’, Clark (Reference Clark2019: 103–104) states that, before Ostrom, ‘the groundswell of systems theory and complexity science had not yet occurred’, and that her work in this field ‘proceeded without support or guidance from others who had come before’ her. In contrast to this, we uphold that Simon’s reflection on systems science and complexity profoundly structures her global methodology and the tools she uses to conduct her work.

Indeed, The Sciences of the Artificial, as conceptualised by Simon, ground a large part of the method employed by Ostrom on social systems, understood as complex systems, as she explicitly acknowledges in a 2004 article co-authored with Vincent:

The sciences of the artificial were born from the realisation that humans are now evolving in a world that is more ‘artificial’ than natural, made up of a growing number of increasingly complex ‘artifacts’. These artefacts range from the material objects we create and use to immaterial goods (knowledge, computer data, etc.), to the various organisations we evolve in, with its associated flow of information. What fundamentally characterises these artefacts for Simon is that they are ‘being moulded, by goals or purposes, to the environment in which [they live]’ (Simon, [1969] Reference Simon1996: xi). This malleability of artefacts in relation to the goals pursued by humans vis-à-vis the environment implies a certain ‘contingency’ in artificial phenomena, unlike in natural phenomena: ‘The contingency of artificial phenomena has always created doubts as to whether they fall properly within the compass of science… . The genuine problem is to show how empirical propositions can be made at all about systems that, given different circumstances, might be quite other than they are’ (ibid.). This, then, is what needs to be surpassed by the elaboration of a new science, as Avenier (Reference Avenier2019: 47, our translation) describes: ‘The central project of the sciences of the artificial is, then, to provide an epistemic framework better suited than that of the natural sciences to represent and understand phenomena in which both human intentions and perceived natural regulations are embodied’. The sciences of the artificial are thus composed of several disciplines: systems science, which enables the rigorous study of complex formal systems; cybernetics, which is more concerned with communication between systems and/or subsystems; design science, which provides systematic methods for system innovation; cognitive science, which studies the processing of information by any system; etc. (Durand, Reference Durand2017). This new research programme aims both to understand the functioning and evolution of artefacts and to formulate well-founded knowledge and methods for developing new artefacts with desired properties (Avenier, Reference Avenier2019: 47).

It is precisely this objective that Ostrom seeks to achieve in the study of the management of collective goods, particularly from the 1980s onwards. This is why she asserts that the issues and foundations laid down by Simon in The Sciences of the Artificial are the same as her own (E. Ostrom and V. Ostrom, Reference Ostrom2004). In one of her earliest studies on the institutional framework that would become the IAD, she states that ‘[n]o guidance or control is possible without an understanding of how complex systems work’ (Ostrom, Reference Ostrom, Kaufmann, Majone and Ostrom1986b: 473). Thus, what she finds in The Sciences of the Artificial is a series of concepts and tools already at hand for the study of social systems understood as complex systems:

It is striking to see that the figure of the systems engineer, linked to these sciences of the artificial, progressively takes on a central role in her work from the 2000s:

So, just as the contingency of artificial phenomena was a preoccupation for Simon, as we have seen, so it was for Ostrom, who ruled out the possibility, or even the desirability, of seeking to ‘produce universal laws of political process and outcomes that hold anywhere and at any time (similar to the laws of physics)’ (Ostrom, Reference Ostrom2006: 6). The work of the social scientist is more akin to that of a systems engineer: ‘when the outcome of a particular process is unexpected, important, or has serious consequences …, our task is similar to that of engineers who are asked to determine the cause of a failure of some complex equipment such as an airplane, ferry boat, or rocket’ (ibid.).

Simon’s research programme on the sciences of the artificial is thus presented by Ostrom as a decisive source of inspiration, both conceptually and methodologically. To understand what Ostrom concretely drew from Simon’s work, we will now examine in closer detail three analytical perspectives he developed to respond to the challenge of complexity and how she appropriated them for her study of social systems.

Analysing complex social systems as nearly decomposable ones

Simon insists that most complex systems, whether natural or artificial, are to be understood and analysed as ‘hierarchic systems’ or, more specifically, as ‘nearly decomposable systems’ (Simon, [1969] Reference Simon1996). Hierarchic systems refer to ‘all complex systems analysable into successive sets of subsystems’.Footnote ⁸ Nearly decomposable systems form a particular class of hierarchic systems (ibid.: 185). For Simon, hierarchic systems can be decomposed into ‘the interactions among subsystems, on the one hand, and the interactions within subsystems that is, among the parts of those subsystems on the other’, with nearly decomposable systems being those hierarchic systems in which ‘the interactions among the subsystems are weak but not negligible’ (ibid.: 197, author’s emphasis).Footnote ⁹ While nearly decomposable systems would constitute most complex systems, the usefulness of the proposed categorisation actually goes beyond mere description. In the chapter ‘The Architecture of Complexity: Hierarchic Systems’, Simon states:

Thus, this categorisation has heuristic value for researchers seeking to study complex systems, as it aligns well with human cognitive capacities and available resources. Indeed, to analyse and design complex structures, ‘one powerful technique is to discover viable ways of decomposing [them] into semi-independent components corresponding to its many functional parts’, enabling each component to be thought of ‘with some degree of independence’ from the others (Simon, [1969] Reference Simon1996: 128). While this may seem abstract in relation to human situations, Simon points out that ‘[m]uch of classical organization theory in fact was concerned precisely with this issue’ (ibid.).

Ostrom drew on this very conceptualisation to found the IAD (Ostrom, Reference Ostrom2005) and to take better account of the ecological aspect of social systems (Ostrom, Reference Ostrom2007; see Wall, Reference Wall2014). In her 2005 book, in which the IAD plays a central role, Ostrom seeks to provide solutions for social scientists who want to study the social world by acknowledging its complexity. They face ‘a major challenge in identifying the appropriate level of analysis relevant to addressing a particular puzzle and learning an appropriate language for understanding at least that focal level and one or two levels above and below that focal level’ (Ostrom, Reference Ostrom2005: 12). To respond to this challenge, as she points out in her Nobel Prize lecture, she drew her ‘inspiration from the work of Arthur Koestler (Reference Koestler, Gray and Rizzo1973) and Simon ([1969] Reference Simon1981, Reference Simon, Morowitz and Singer1995) who both challenged the assumption that human behavior and outcomes are entirely based on a small set of irreducible building blocks’ (Ostrom, Reference Ostrom2010a: 646). For her, social systems are best viewed as hierarchic and nearly decomposable systems, in which ‘it is somewhat arbitrary as to where we leave off the partitioning and what subsystems we take as elementary’ (Simon, [1969] Reference Simon1981: 196 in Poteete, Janssen and Ostrom, Reference Poteete, Janssen and Ostrom2010: 359). This is why she intends to provide a tool capable ‘of “dissecting” complex systems into composite holons that are then dissected further’ (Ostrom, Reference Ostrom2005: 11–12).Footnote ¹⁰ This is precisely what the IAD enables, as Ostrom explicitly describes: ‘It is a framework for analyzing nearly decomposable systems ranging from families, to neighborhoods, to large communities, to governmental units at a local, regional, national, or international level’ (Ostrom, Reference Ostrom2006: 6–7, our emphasis).

The IAD is conceived as ‘a multitier conceptual map’, in which the analysis of each component can lead to the analysis of its sub-components, which in turn can be broken down in greater detail according to the needs expressed by the researcher (Ostrom, Reference Ostrom2005: 14). Within this framework, an institutional situation is indeed defined by an ‘action arena’, itself composed of two holons: the action situation and the participants. The action situation is then characterised by seven variablesFootnote ¹¹ , and the participants’ behaviours are modelled through three main variables that must be adapted to each specific situation. However, it is also possible to consider three exogenous variables to the action arena affecting its structure: 1) the characteristics of the biophysical world attached to the situation; 2) the structure of the community (with its norms, habits, cultures, etc.); 3) the rules used by participants in their relationships. Each of these exogenous variables is made up of a large number of variables. Rules being the central object of institutions (Ostrom, Reference Ostrom1986a), Ostrom proposes, for example, seven fundamental classes of rules, echoing the seven variables structuring action situations. Nevertheless, it is still possible to go further by detailing the syntactic composition of rules, in particular to dissociate them from norms or collective strategies. Thus, Ostrom also proposes an institutional grammar, which she calls ADICO syntaxFootnote ¹² , to recognise these elements and their evolution in each institutional situation (Crawford and Ostrom, Reference Crawford, Ostrom and Ostrom2005: 139–140). Lastly, it is possible to step back and study the interactions between several action arenas ‘linked together either sequentially or simultaneously’ (Ostrom, Reference Ostrom2005: 15). As we can see, the IAD corresponds well to nested conceptual maps presented ‘at scales ranging from exceedingly fine-grained to extremely broad-grained’ (Ostrom, Reference Ostrom2005: 11). Similarly, when she seeks to take greater account of the ecological environment by developing the SES framework from the IAD, she points out that ‘SESs are partially decomposable systems’ (Ostrom, Reference Ostrom2007: 15182). Once more, the Simonian perspective is advocated: ‘[s]cientific progress has been achieved in the past when scholars have recognized that complex systems are partially decomposable in their structure’ (ibid.).

Recognising the adaptive evolution of complex social systems

In addition to this hierarchical conception of complex systems, Simon advocates an approach to the study of complex systems based more on the recognition of the variety of possible evolutions than on a principle of optimisation leading to a single solution, particularly in economics. This thought is similar to his criticism of standard rational choice theory (Simon, Reference Simon1955, Reference Simon1956).Footnote ¹³ Here, however, the argument has more to do with the methodology to be employed in studying economic complex systems. He states:

His speech to the Richard T. Ely lecture at the American Economic Association in 1977 helps us understand the argument advanced here. Simon criticised the systematic, even excessive, use of marginalist calculus inside and outside the ‘traditional domains of production and distribution’ (Simon, Reference Simon1978: 4). Instead, Simon promoted a functionalist analysis in the social sciences, which was concerned with the realisation and maintenance of individual or social ‘goals’ (Simon, Reference Simon1978: 3).Footnote ¹⁴ And more specifically, by stepping outside the traditional domains of economics, Simon pointed out that ‘it is not clear that anything new is added’ with marginalist calculation than with ‘cruder and simpler arguments’ of functionalist analysis, ‘since the parameters imputed to the system are largely unmeasured and unmeasurable’ (Simon, Reference Simon1978: 6-8). Functionalist analysis, on the other hand, enables us to understand how a system can evolve to maintain itself (whether it reaches an equilibrium state or not). Indeed, to consider a solution as such in the functionalist framework proposed by Simon, it is sufficient that it satisfies the objectives specific to the system’s survival: ‘all that can be concluded from a functional argument is that certain characteristics (the satisfaction of certain functional requirements in a particular way) are consistent with the survival and further development of the system, not that these same requirements could not be satisfied in some other way’ (Simon, Reference Simon1978: 4). This is all the truer when the system and its associated environment are complex, in which case ‘there is no assurance that the system’s momentary position will lie anywhere near a point of equilibrium, whether local or global’ (ibid.), with each element continually adapting to the other elements with which it shares a connection. This is why Simon states in The Sciences of the Artificial that ‘the evolution of firms and economies … is probably best understood through an examination of its history’ (Simon, [1969] Reference Simon1996: 48).

The study of the commons proposed by Ostrom (Reference Ostrom1990), resulting in the eight design principles, is perfectly in line with the Simonian logic of considering the adaptive evolution of systems rather than their optimal solutions. While Ostrom (Reference Ostrom1990, Reference Ostrom2005, Reference Ostrom2007) calls for never relying on ‘universal solutions’ or ‘panaceas’, whatever the institutional situation under consideration, the elaboration of the design principles shows a way of getting around this problem. For her, a design principle refers to an ‘essential element or condition that helps to account for the success’ of institutions managing a CPR, promoting a sustainable management of the CPR and the conformance of users to the institutional rules, generation after generation (Ostrom, Reference Ostrom1990: 90).Footnote ¹⁵ These principles are special because they do not indicate ready-made solutions, but instead correspond to what the rules of an institutional situation must respect to enable agents to find solutions adapted to their situation. As Peneranda (Reference Peneranda2012: 28, our translation) points out, these design principles find their foundation in an ‘engineering approach that consists in designing procedural rules’ where ‘[i]t is not the decision that is optimized’ but ‘the process of coordinating actors in decision-making that is strengthened’. As a result, several sets of rules, depending on the specific and historical situation of the problem under consideration, can fulfil the same goals, such as the survival of the CPR and agents’ conformance to collective rules. In other words, we find here the notion of functional equivalence that Simon ([1969] Reference Simon1996, Reference Simon1978) defended concerning complex systems, as we saw above. For this reason, Ostrom stated the following about the design principles in 2005:

More generally, she explicitly states that no notion of optimality is automatically attached to the gradual adaptation of institutional rules since each institutional situation is too complex ever to be sure of achieving any optimality. Ostrom emphasises that it is the very notion of ‘optimality’ that poses a problem for institutional situations: ‘In fact, given the high levels of uncertainty involved and the difficulty of measuring the benefits and costs, it would be extremely difficult to obtain a meaningful measure of optimality’ (Ostrom, Reference Ostrom1990: 59–60). And she continues in the associated footnote: ‘I prefer to argue that optimality is not well defined in a changing environment, including the capacity to change the institutional rules themselves’ (Ostrom, Reference Ostrom1990: 224; see also Ostrom, Reference Ostrom2005: chap. 8; Ostrom and Basurto, Reference Ostrom and Basurto2011: 319–321). This clearly echoes Simon’s comments (1978: 4): ‘when the system is complex and its environment continually changing (that is, in the conditions under which biological and social evolution actually take place), there is no assurance that the momentary position will lie anywhere near a point of equilibrium, whether local or global’. Other influences also play an important role in Ostrom’s adaptive conception of institutional rules. These influences include evolutionary theories in economics (Nelson and Winter, Reference Nelson and Winter1982; Schmid, 2004), notably with the notion of ‘path-dependence’ (North, Reference North1990), in psychology (Campbell, Reference Campbell1975; Holland, Reference Holland1995), and in biology (Mayr, Reference Mayr1982; Boyd and Richerson, Reference Boyd and Richerson1985). Remarkably, many of them explicitly incorporate Simon’s work into their approach (Mayr, Reference Mayr1982; Boyd and Richerson, Reference Boyd and Richerson1985; North, Reference North1990), and some consider it fundamental (Koestler, Reference Koestler1967; Nelson and Winter, Reference Nelson and Winter1982; Schmid, 2004; Holland, Reference Holland1998). This reveals Simon’s central reference in this field of research, although he is obviously not its only representative. Thus, it is not surprising to see Ostrom combine Simon’s concepts – the satisfaction criterion and adaptive solutions – with North’s path dependency within the same line of reasoning:

Therefore, Simon’s emphasis on the adaptive evolution of complex social systems had a significant direct and indirect influence on Ostrom’s work.

Relying on realistic assumptions and empirical investigation as a warrant for scientific inquiry

One final analytic perspective on complexity that Ostrom adopted from Simon is his continual commitment to relying on realistic assumptions and conducting empirical work through field surveys, experiments, and simulations. In addition to his work on The Sciences of the Artificial, Simon ([1947] Reference Simon1957) is best known for his book Administrative Behavior, in which he delivers a major critique of the science of administration, and, in particular, the ‘administrative principles’ traditionally used to define the type of organisation that works best.Footnote ¹⁶ Simon disapproves the aprioristic and hardly operational nature of these administrative principles, since they are not derived from empirical analyses based on the ‘efficiency’ criterion that organisations actually aim at: ‘studies of administrative agencies have been carried out without benefit of control or objective measurements of results, they have had to depend for their recommendations and conclusions upon a priori reasoning proceeding from “principles of administration”’ (Simon, [1947] Reference Simon1957: 43–44). He aims to give a stronger methodological foundation for administrative science by promoting a more empirical research programme: ‘What is needed is empirical research and experimentation to determine the relative desirability of alternative administrative arrangements’, knowing that the ‘methodological framework for this research is already at hand in the principle of efficiency’ (Simon, [1947] Reference Simon1957: 42). This emphasis on empirical investigation would remain a defining feature of Simon’s research throughout his career: in his view, only empirical testing could ensure the validity of a hypothesis (Simon, Reference Simon1997). This position appears even more compelling in the context of complex systems, which are characterised by the difficulty of inferring their structural properties solely from knowledge of their individual components (Simon, [1969] Reference Simon1996: 184). This is why ‘the genuine problem [of the sciences of the artificial] is to show how empirical propositions can be made at all’ (Simon, [1969] Reference Simon1996: xi). Hence, Simon (Reference Simon, Alt, Levi and Ostrom1999: 118) goes so far as to call himself an ‘empirical scientist’, insisting that abduction is a legitimate method for conducting social science research, and is always careful to return to the empirical evaluation of hypotheses, especially when they refer to complex situations (Simon, [1969] Reference Simon1996: 23, 52, 121).Footnote ¹⁷

The importance of empirical research in the study of any system is precisely what would influence Ostrom to conduct field investigations in metropolitan areas and later in CPR situations. If the importance of conducting empirical research in Ostrom’s work is well established (Aligica and Boettke, Reference Aligica and Boettke2009; Frischman, 2013; Herzberg, Reference Herzberg2015), few have noted the role again played by Simon. As early as 1971, in an article co-authored with V. Ostrom, they claimed to promote a new approach to the study of public administration, based in part on the public choice research programme, but also on Simon’s observations and guidelines for administrative science (V. Ostrom and E. Ostrom, 1971).Footnote ¹⁸ It is from this point onwards that Ostrom clearly expressed her mistrust of any solution presented as panacea, and asserted the need always to rely on empirical studies, whether in developing an operational vocabulary, establishing measurements or conducting experiments. This explains in particular the numerous field surveys conducted during the 1970s in the case of metropolitan areas, as she explicitly states (Ostrom, Reference Ostrom, Cole and McGinnis2017: 407). It also clarifies her remark in her 2004 article: ‘Simon also urged scholars to add simulation techniques to their tool kits. He did this very self-consciously. He felt that many of the overly simple mathematical models used in policy analyses were dead wrong and led to incorrect decisions if not tragedies’ (Ostrom, Reference Ostrom2004: 45). Her criticism of economic models on collective action problems, which she regards as mere ‘metaphors’ disconnected from actual CPR situations, echoes Simon’s criticism of ‘traditional administrative principles’, which he regards as mere ‘proverbs’ disconnected from actual administrative organisations (Simon, [1947] Reference Simon1957). This alignment also sheds light on Ostrom’s late-career enthusiasm for agent-based models (ABMs), especially in situations ‘too complex to be modeled as a simple game’ (Ostrom, Reference Ostrom2005: 7; see also Ostrom, Reference Ostrom2007: 15186; Ostrom and Basurto, Reference Ostrom and Basurto2011: 337). And, even then, Ostrom emphasises the importance of confronting these models with empirical data (Janssen and Ostrom, Reference Janssen and Ostrom2006; Poteete, Janssen, and Ostrom, Reference Poteete, Janssen and Ostrom2010, chaps. 8–9).

Furthermore, Simon’s emphasis on introducing greater complexity into economic modelling and rational choice theory helps clarify Ostrom’s own perspective on these approaches. Indeed, from 1990 onwards, she began explicitly to deplore the lack of theory capable of accounting for the empirical observation that individuals succeed in organising themselves to manage a CPR: ‘We have not yet developed a behavioral theory of collective action based on models of the individual consistent with empirical evidence about how individuals make decisions in social-dilemma situations’ (Ostrom, Reference Ostrom1998: 1, author’s emphasis; see also Ostrom, Reference Ostrom1990: 25–26). Her central criticism is that standard rational choice theory is too simplistic to understand the complex reality of collective action, being limited mainly to ‘open competitive situations’ (Janssen and Ostrom, Reference Janssen and Ostrom2006: 39; see also Ostrom, Reference Ostrom2005: 100–101). Individuals possess cognitive capacities and limitations that have evolved in accordance with their social and ecological environments (Ostrom, Reference Ostrom1998; Reference Ostrom2005: 104). This is why she states that the analyst ‘needs to substitute the assumption of bounded rationality—that persons are “intendedly” rational but only limitedly so—for the assumptions of perfect information and utility maximization used in axiomatic choice theory’ (Ostrom, Reference Ostrom2010b: 18). Some commentators duly underscore the role of complexity in this critique (Lara, Reference Lara2015; Lewis and Petersen, Reference Lewis and Petersen2023) and others duly note that she appropriates Simon’s notion of bounded rationality (Chanteau and Labrousse, Reference Chanteau and Labrousse2013; Tarko, Reference Tarko, Lemke and Tarko2021). However, none of them goes so far as to explain Simon’s influence on Ostrom’s conception of rationality through the emphasis on complexity and the related general methodological approach. Yet, once again, a striking parallel emerges with Simon, who argued nearly fifty years earlier that ‘the concept of “economic man” … is in need of fairly drastic revision’, and should be replaced by a conception of rational behaviour aligned with actual human capacities and situated within specific environments (Simon, Reference Simon1955: 99).Footnote ¹⁹ It also explains why, when criticising previous research on collective action problems in her Nobel Prize lecture, Ostrom states that ‘the basic assumptions that were made at that time … have been challenged by scholars around the world, including the work of Herbert A. Simon (Reference Simon1955) and V. Ostrom ([1973] Reference Ostrom2008)’ (Ostrom, Reference Ostrom2010a: 642, our emphasis).

Therefore, Ostrom’s commitment to conducting empirical investigations and relying on realistic assumptions – particularly through the development of models of bounded rationality – borrows Simon’s insistence on recognising the importance of addressing the complexity of the systems under study through empirical and realistic analysis.

The sciences of the artificialor the sciences of the artifactual?

Although Ostrom explicitly acknowledges Simon and claims a direct intellectual lineage with his work, she opts for the expression ‘the sciences of the artifactual’ rather than Simon’s original term, ‘the sciences of the artificial’ (E. Ostrom and V. Ostrom, Reference Ostrom2004: 139). This terminological choice is far from neutral. It reveals Ostrom’s distance from certain aspects of Simon’s work. Two points of tension stand out in particular.

First, Ostrom (Reference Ostrom1990, Reference Ostrom2007) is wary of any methodological posture marked by a desire to design the organisations in which social actors evolve. Simon, however, is more ambiguous on this subject. As we saw in the previous section, both Simon and Ostrom adopt a bottom-up approach to science. It is their reluctance to believe in ‘panaceas’ that leads them to opt for such an approach.Footnote ²⁰ However, after conducting an empirical and contextualised analysis of an institutional problem, the question remains of who will participate in designing and implementing the solution. For Ostrom, it is ultimately up to the social agents themselves to assume this responsibility within the institutional context they inhabit (Ostrom, Reference Ostrom1990, Reference Ostrom2005).

For Simon, however, the answer is more complicated. On the one hand, he often refers to ‘designers of administrative organizations’ ([1969] 1996: 128) or social ‘planners’ who act from above on the behaviour of the members of the social system under consideration (ibid.: chap. 6). On the other hand, he points out that these members can never be regarded as ‘passive instruments’, but rather as ‘themselves designers’ of the institutions in which they find themselves, ‘seeking to use the system to further their own goals’ (Simon, [1969] Reference Simon1996: 153). According to Simon, this is because any complex system is primarily composed of ‘stable intermediate forms’ that operate ‘nearly independently of the detailed processes going on within the other subsystems’ (ibid.: 192–193), as discussed in Section “Analysing complex social systems as nearly decomposable ones”. Thus, as Fiori (Reference Fiori2010: 155) analyses, social systems evolve for Simon through the ‘continuous interaction’ of their subsystems, ‘without fixed goals’ guiding them since ‘every step in the interaction constitutes a new base for the emergence of new plans’ for these subsystems. And, it is on the basis of this interpretation that Ostrom takes up Simon’s defence regarding the ambiguity of the term ‘hierarchy’ he uses: ‘Simon overtly distinguished his concept of hierarchy from one involving human bosses at each level’ (Poteete, Janssen and Ostrom, Reference Poteete, Janssen and Ostrom2010: 377). More generally, this Simonian conception of social system evolution can, to some extent, be linked to the concept of ‘polycentric governance’, which Ostrom (Reference Ostrom2005, Reference Ostrom2010a) uses to characterise a public economy.Footnote ²¹

Nonetheless, Ostrom firmly rejects the idea of a ‘social planner’ acting above social actors, as it inevitably risks relying on ‘self-evident truths’ or ‘one-size-fits-all’ solutions (Ostrom, Reference Ostrom2000; Reference Ostrom2005: 270; Reference Ostrom2007). This is what led Ostrom to declare at the end of her career: ‘The term “design principle” has confused many readers. Perhaps I should have used the term “best practices” to describe the rules and structure of robust institutions’ (Ostrom, Reference Ostrom2010a: 653, our emphasis). Hence, the term ‘artifactual’ more accurately reflects Ostrom’s method than the term ‘artificial’ because the former emphasises the cultural and social aspects of institutions, whereas the latter inherently conveys an engineer’s design perspective.Footnote ²²

Second, Simon was suspicious of some of the tools Ostrom used to deal with bounded rationality in institutional settings, such as game theory and public choice theory (Simon, Reference Simon1985, Reference Simon, Alt, Levi and Ostrom1999). His refusal to participate in the 1997 American Political Science Association annual meeting, when Ostrom was president and invited him, perfectly illustrates this point. Simon fully associates Ostrom with the public choice movement and its ‘ridiculous ideas of perfect rationality’ (Simon, Reference Simon1996). While this certainly reflects Simon’s lack of familiarity with Ostrom’s work, it also indicates two different understandings and applications of bounded rationality in institutional settings. Simon seeks to understand the bounded rationality of individuals making decisions within specific ecological and social environments. Bounded rationality is therefore central and is defined in terms of adaptation to ‘constraints imposed both by the external situation and [human] capacities’ (Simon, Reference Simon1985: 294, author’s emphasis; see also Simon, Reference Simon1955, Reference Simon1956). Yet, as he himself acknowledges, the internal – cognitive – dimension has at times overtaken the environmental dimension in his work (see Petracca, Reference Petracca2021). This is why such constraints typically relate to the individual’s computational limits, the incommensurability of her various objectives, and uncertainty regarding both the environment and the range of possible alternatives. Faced with this complexity, agents can attend to only a limited and biased amount of information. They, therefore, construct simplified mental models of the situation while generating new alternatives that may transform it (Simon, Reference Simon1985). And they respond primarily to the criterion of satisfaction rather than maximisation, as discussed in Section “Recognising the adaptive evolution of complex social systems”.

From another perspective, Ostrom (Reference Ostrom1990, Reference Ostrom1998, Reference Ostrom2005, Reference Ostrom2010a) primarily seeks to understand how individuals solve collective action problems through collective decision-making. Bounded rationality is secondary in her work and is always applied within a social context. To understand the interactions between individuals that lead to collective choice, it is then useful in her perspective to turn to public choice theory and game theory, notably by linking them to laboratory experiments (Ostrom et al., Reference Ostrom, Gardner and Walker1994). Thus, Ostrom places greater emphasis than Simon does on reciprocal relations, norms, communication, and individuals’ commitments within a community to develop ‘second-generation models of empirically grounded, boundedly rational, and moral decision making’ (Ostrom, Reference Ostrom1998: 15).Footnote ²³

Therefore, Ostrom and Simon agree on the need to use models of bounded rationality that are consistent with the complex approach to social systems described throughout this article. However, they adopt different perspectives in doing so, reflecting the distinct phenomena each initially seeks to explain.

Having examined Ostrom’s appropriation of Simonian analytical perspectives to study social systems, we now turn to a case study to illustrate how this lens deepens our understanding of her institutionalism.

Institutional evolution through incremental change: a textbook case

One good example that illustrates our general argument is Ostrom’s study of the management of groundwater basins in California, as discussed in Governing the Commons. Although it was the case study of her 1965 thesis, she revisited it in 1990 when developing her design principles, offering a new interpretation. The problem was that many different users, including private landowners and numerous private or public water companies, exploited these basins. Until the mid-20th century, water rights in California were not clearly defined. And every unit of water used by one user was effectively withdrawn from the others. Yet, overconsumption of water could lead to sedimentation of the basin or, in the case of basins closest to the Ocean, to the intrusion of salty seawater. In other words, these were CPR that risked being destroyed by overconsumption if no action was taken – the ‘tragedy of the commons’. By analysing three basins (Raymond, West, and Central), Ostrom (Reference Ostrom1990, chap. 4) shows how participants succeeded in implementing private and public measures to alter the situation, thereby stopping overconsumption and destruction of the basins: creation of local water associations; definition of water rights and appropriation rules ratified by the California court, along with an associated market; creation of a ‘watermaster service’ at the California Department of Water Resources; creation of special water districts; etc.

The main lesson Ostrom draws from this is the ‘incremental’ aspect of the institutional transformation that took place (Ostrom, Reference Ostrom1990: 137). Far from having the final solution in mind from the outset, participants began by creating water associations. This enabled them to establish ‘a forum for face-to-face discussions’ and ‘a mechanism for obtaining information about the physical structure of the basins’ (Ostrom, Reference Ostrom1990: 138). Through this process, participants realised ‘that it was possible to accomplish some joint objectives by voluntary, cooperative action’ and gather information on ‘the condition of their basins and the likelihood that others would commit themselves’ (ibid.). Thus, although ‘[d]iscussion by itself was not sufficient to change the pumping strategies of the participants’, it was what ‘enabled the participants to reach an enforceable agreement to limit their water withdrawals’, later ratified by the California court (ibid.). This aspect fully reflects the Simonian analysis of complex social systems as nearly decomposable systems, composed of agents that are boundedly rational, and in which the main point of analysis is their adaptive evolution. Indeed, the ‘resulting institutional infrastructure’ created by the participants represents ‘a major investment’ that none would have been prepared to make initially, especially considering the risks of defection (Ostrom, Reference Ostrom1990: 141). However, since some rule changes ‘could be undertaken with low transformation costs’, we observe the formation of intermediate states that enable ‘participants to gain some advantages of collective action’ before confronting ‘more costly alternatives’ (Ostrom, Reference Ostrom1990: 141). In other words, ‘[e]ach institutional change became the foundation for the next change’: starting by solving ‘some initial second- and third-order problems can help participants move toward solving first-order problems [the tragedy of the commons], as well as the more difficult second- and third-order problems’ (ibid.), which closely aligns with Simon’s conception of social system evolution.

This particular view guided Ostrom throughout the rest of her career as she sought to understand the institutional ‘dynamics of rules’ (Ostrom, Reference Ostrom2008: 49; see also Ostrom, 1999: 524–525; Reference Ostrom2000: 149). The conclusion of her 1994 game-theoretic book on CPR situations fully encapsulates this idea (Ostrom et al., Reference Ostrom, Gardner and Walker1994: 322–323). And the new methods she proposes afterwards for analysing the evolutions of rules and norms take shape from this incremental aspect, as the evolutionary tables that recapitulate the phylogeny of irrigation systems’ rules and norms in Nepal (Ostrom, Reference Ostrom2008; Ostrom and Basurto, Reference Ostrom and Basurto2011).Footnote ²⁴ Thus, interpreting Ostrom’s work through Simon’s complex approach adds depth to Ostromian institutionalism: the analysis of CPR situations requires an incremental understanding of their evolution, based on their nearly decomposable structure.