Lindblom's 1959 article on incrementalism is one of the most cited works in social science. 1 This article examines and probes five of the theory's basic, underlying assumptions in light of current empirical and theoretical approaches. The five assumptions are: (1) the limited nature of rationality and the weak powers of human cognition, (2) the emphasis on practical reason and applied knowledge, (3) partisan mutual adjustment conceived as interactions and conflicts among singular, distinct, and disconnected entities, (4) policy decisions made incrementally can be remediated and reversed, and (5) the United States political system is taken as the model for incremental politics. In our assessment, incrementalism holds up quite well with more recent thinking on public policy process with respect to the first and second assumptions. The third assumption has been confronted by new work on networks. The fourth has been challenged by theories on path dependency. The fifth assumption betrayed a national focus and strong value consensus – globalization and “culture wars” have rendered it untenable. The article therefore gives only qualified support for the continued relevance of incrementalism as a theory of the policy process.
A celebration of the influence, impact, inspiration and limitations of an article first published a half century ago would be challenging in itself, but is complicated in the case of Lindblom's “The Science of Muddling Through” ( Lindblom, 1959). First, the article is a moving target. It was preceded by a somewhat more nuanced version published in the American Economic Review in 1953 ( Dahl & Lindblom, 1953) and in 1958 ( Lindblom, 1958), and succeeded by another version ( Lindblom, 1979). So, there is no single classic source, nor a fully consistent version of incrementalism, despite the prominence of the 1959 article. Second, a great deal of the article's fame was due to its claim to offer (or clarify) an alternative model of decision-making. Incrementalism was, according to “The Science of Muddling Through,” a corrective to the synoptic, rational model. Largely due to limitations in human reasoning (though, again, versions varied in emphasis on what the limiting variables were), the real world of policy-making was dominated by decisions that compared closely related and only marginally different alternatives, supplemented by a process of partisan mutual adjustment. Whether he intended to or not, Lindblom seemed to be making both descriptive and normative claims that incrementalism was widely evident in what practitioners did, and moreover, had certain benefits in terms of remediation and reversibility. This was seized upon by earlier critics ( Dror, 1964; Etzioni, 1967) who were concerned about the apparent endorsement of a conservative, piecemeal policy process. In fact, over time, Lindblom made it clear that incrementalism was not necessarily opposed to rationalism, and that indeed the two types of decision-making supported each other. If incrementalism is always a hybrid (similar to Etzioni's “mixed scanning” model), then there really is little to choose between it and rationalism.
This treatment of incrementalism and rationalism leads to a third, and deeper, problem, and that is that Lindblom essentially repudiated any normative claims incrementalism might have had for engaging citizens and a wide variety of policy actors in the policy process. The intimations that elites and the capitalist system actually dominated decision-making was there in the 1979 reply to critics, and echoed the powerful critique of business influence in his 1977 Politics and Markets. By 1990 he was exploring cognition and arguing that it was to the advantage of elites (principally business elites) to encourage and impose a “patterned impairment” of thinking among ordinary citizens.
The fourth problem is the curious tenor of the debate in the literature following the 1959 article. “The Science of Muddling Through” is one of the most widely cited articles in the social science literature, but even Lindblom remarked that he never “understood why incrementalism in its various forms has come to so prominent a place in the policy-making literature” ( Lindblom, 1979, p. 524). Despite this, several authors noted that incrementalism had faded away by the mid-1990s, and that it had failed to serve as a basis for a cumulative empirical and theoretical tradition of work ( Bendor, 1995; Weiss & Woodhouse, 1992). As well, after fifty years, the debate over incrementalism has produced fairly predictable and enduring fault lines. Most worry about its piece-meal, plodding, and anti-synoptic approach, essentially echoing the earliest critiques of incrementalism. Others have tried to rebut the main criticisms ( Weiss & Woodhouse, 1992), conduct empirical tests in experimental conditions to see if incremental decision making under conditions of uncertainty actually holds ( Knott, Miller, & Verkuilen, 2003), or specify the conditions under which incremental decision-making in organizations is likely to happen and when it is not ( Lustick, 1980).
Under these circumstances, a fresh approach is needed. We will assess the core assumptions of incrementalism – five of them – both in terms of the original theory itself and in terms of more recent work in policy studies. How well does the theory hold up, both in its own terms, and in terms of work in the last decade?
If we take the 1959 article as the basic version of theory (though, as we noted above, revised from time to time by Lindblom himself), it consists of the following elements. The rational-comprehensive model of decision-making assumed clarification of values and objectives distinct from comparison of alternative policies, a means-end analysis, and a comprehensive review of every relevant factor to the decision. Lindblom argued that intellectual and informational capacities, as well as time and resources, simply do not permit this style of reasoning most of the time, especially for complex problems. In fact, administrators do not rank values in the abstract, but simply compare a limited number of alternative policies that combine values in different ways. Moreover, the alternatives typically differ only marginally, or incrementally. He called this the method of “successive limited comparisons.” A hallmark of the method was the attempt at simplification (again, largely because of capacity, resources and time). Simplification might seem to ignore important possibly consequences of decisions, but Lindblom argued that every important value or interest has a watchdog, and so in a process of partisan mutual adjustment, “our system often can assure a more comprehensive regard for the values of the whole society than any attempt at intellectual comprehensiveness” ( Lindblom, 1959, p. 85). The final distinctive element of what he called the “branch method” was that policy is made and re-made endlessly in a process of “successive limited comparisons.” “Muddling through” was in fact a system of decision-making, however imperfect, practiced in the real world of administration.
This article examines five key underlying assumptions in the theory, and weighs them against more recent developments in the policy sciences. How well does incrementalism hold up as a theory in light of, for example, recent thinking about networks or on globalization? We have selected five key assumptions, drawn directly from the 1959 article:
Lindblom was, by training, an economist, and his earliest writings accepted the mechanisms of the price system as impersonal and complex, resulting in aggregate results that no one person could have anticipated or planned ( Lindblom, 1949, pp. 7–9). There can be no doubt that the market's anonymity, complexity of interactions, and collectively satisfactory results formed a template for Lindblom's approach to incrementalism. At the time (the 1950s) Bentley's work on pluralism was being rediscovered ( Bentley, 1935), and Truman's opus on interest groups had been published in 1951 ( Truman, 1951), both strongly supplementing the model adopted from price and market mechanisms. It is arguable that the impact of pluralist theory was more strongly felt in the “The Science of Muddling Through” and subsequent emphases on partisan mutual adjustment, whereas the earlier precursors stressed the problem of the limits of human rationality. In Politics, Economics, and Welfare ( Dahl & Lindblom, 1953), Dahl and Lindblom rehearsed some of these limitations. From the point of view of decision-makers (and they situated their discussion against the backdrop of a scenario involving politicians having to come to a decision about a highly complex science-based policy), this involved the complexity and the amount of information that required processing, the limits of effective communication of that information, the sheer number of variables that had to be taken into account, as well as the relationships among those variables (p. 59).
Humans are burdened with “small brains,” limiting the capacity for foresight (we don’t really know what we want until we test it). Moreover, many of the objectives that we seek as human beings (love, affection, respect, spontaneity) cannot be achieved through rational calculation, or at least not entirely (pp. 60–61). In this work, they argued that various stratagems have been developed to deal with these limitations, and incrementalism is one of a suite of “comprehensive” techniques: “Incrementalism is a method of social action that takes existing reality as one alternative and compares the probable gains and losses of closely related alternatives by making relatively small adjustments in existing reality, or making larger adjustments about whose consequences approximately as much is known as about the consequences of existing reality, or both” (p. 82). The advantages of incrementalism that they list fall into two categories, rational/cognitive and process/institutional. In the first category, incrementalism limits alternatives, and makes it easier to predict the consequences of a decision. In the second, there is better control by superiors in an organization, reversibility, organizational stability (since fundamentals are not being attacked), and more clarity to an electorate and the processes of polyarchy (pp. 82–83).
In A Strategy of Decision ( Braybrooke & Lindblom, 1963, p. 113), at least four of the rationales for incrementalism were cognitive: limited intellectual capacities, limited knowledge, the inability to construct a consistent welfare function, and the interdependence of facts and values. In The Policy Making Process ( Lindblom, 1968, p. 27) Lindblom defended the various strategies or “dodges” for getting around the complexity of decision-making by arguing that they are “useful devices for stretching man's analytical capacities. Man has had to be devilishly inventive to cope with the staggering difficulties he faces. His analytical methods cannot be restricted to tidy scholarly procedures.” The same themes of complexity, information overload (or underload in some cases), the challenges of the fact/value distinction and clarity and stability of goals, remained themes throughout the rest of his work, though as we noted, with the publication of Inquiry and Change ( Lindblom, 1990), his critique of elite political domination through a capitalist system induced him to suggest that incrementalism was less a matter of inherent limits to human cognition that it was to “patterned impairment.”
The original emphasis on the limits of human cognition has stood up well. It relied to some extent on Herbert Simon's model of satisficing (which in itself was a powerful critique of synoptic rationality), and that model has more or less retained strong support in the social sciences. More recently, work in behavioural economics has clarified some of the limits on human rationality as well as reinforcing many of Lindblom's original arguments that led to a defence of incrementalism. Schelling, for example, arrived at the notion of the “divided self” (a literary phrase that has been refined conceptually in economics as “hyperbolic discounting”) in order to understand the gap between rational intentions and actual behaviour ( Schelling, 1984). He also contributed to the development of experimental economics. The limits of rationality has now moved beyond academe: Thaler and Sunstein's Nudge (2008) has been a best seller and makes an explicit argument about policy makers as “choice architects” dealing with the foibles of ordinary people: “…we show that in many cases people make pretty bad decisions – decisions they would not have made if they had paid full attention and possessed complete information, unlimited cognitive abilities, and complete self-control” (p. 5). Thaler and Sunstein are solidly in the tradition of Schelling's “divided self,” supplemented with other psychological insights about behaviour and choice, and focus more on the impediments to rational decision-making at the individual rather than the policy level (indeed, their analysis of the limits to reason is the basis for interventions that nudge citizens and consumers in the direction of their own self-acknowledged interests). Nonetheless, the point about the general limits of human cognition is a strong echo of Lindblom's early arguments about incrementalism as a technique or strategy to deal with those limitations.
The importance of limits to rationality has perhaps been acknowledged more in political science and the policy sciences than economics, but the insights of behavioural economics have seeped into these other disciplines. Jones argues that people are “intendedly rational” but fail for a variety of reasons, many of them cognitive. His list of those limitations – forty years after “The Science of Muddling Through” and almost fifty years after Lindblom's first arguments about the limits to human rationality – contain the following: the incrementality of search behaviour, the difficulties of calculation, framing effects, self-control, difficult choices among incommensurable attributes, challenges of decision designs, reluctance to update decisions with new information, and status quo biases ( Jones, 1999, pp. 307–308). The notion of bounded rationality has been extended to understanding policy-making as information processing systems in a context where there is an overwhelming amount of information and a consequent need to focus attention on policy problems and prioritize information ( Jones & Baumgartner, 2005). This theory of disproportionate information processing seeks to rescue what Baumgartner and Jones call the “discredited” theory of incrementalism by making it a special case or phase in a policy process characterized by punctuated equilibrium.
However, why should cognitive limits induce caution? Without reviewing Baumgartner and Jones here in detail, their argument about punctuated equilibrium and disproportionate information processing opens the door precisely to the possibility of overreactive decision-making. Lindblom's model of limited cognition was of course complemented by references to a policy process populated with many competing partisans, and so satisficing (incremental) behaviour might be due to the saw-offs among many, competing interests. But for the moment we are considering only limits to reason. Once we consider this, and think about the logic of incrementalism, we see that a key if hidden assumption underpinning incremental reasoning (or incrementalism as policy analysis) is that actors are what Jones calls “intendedly rational.” This means that they wish to satisfy objectives, however unclear they are about those objectives, and however limited the knowledge at their disposal. But if actors’ objectives conflict, or include multiple, non-policy goal specific objectives, such as speedy resolution, popularity and re-election, or a reputation for incisiveness, there is no reason why these actors would cling to small steps.
Our point is simply that Lindblom's original emphasis on the limits to rationality and cognition in decision-making were well-founded, and have indeed been supported and extended in the recent literature. But he embedded those limits within a pluralist universe of competing interests, a universe that induced caution and hence incremental decision-making. But, if we concentrate only on rationality and cognitive incapacity, and couple that with the other human limits to reason, we might come to a different conclusion. We could surmise that the minimum conditions for this situation as follows:
Admittedly, these conditions amount to a thought experiment, but not one that is divorced from the reality of politics. Political actors often operate in situations where they do not have full information, but nonetheless face pressures to appear decisive in the face of major problems. Under these conditions, we would argue that non-incremental decision-making or “calculated risk” is quite likely. We will not get sidelined into a discussion of what counts as “big” or “small.” The point is simply that under these conditions, high-risk “big” decisions would be just as likely as incremental ones. Of course, this would be unimportant if these conditions are rare or only theoretical, but it may well be that the last fifty years has seen an increase in these types of decisions. President Obama's first 100 days might be a model for this situation, though it should be noted that Dror, shortly after the publication of “The Science of Muddling Through,” argued that innovative policy could not rely on incrementalism, but that nonetheless, many industrialized states emphasized the importance of innovation ( Dror, 1964). Dror's later work continued to explore the theme of crisis decision-making, not as an aberration, but as a normal part of public policy-making ( Dror, 1986).
Another aspect of incrementalism that has rarely been debated is that the contrast between incremental decision-making and synoptic decision-making was also a distinction between practical reason used by practitioners in the field and abstract reason used by researchers or social scientists. “The Science of Muddling Through” was published in Public Administration Review, which includes a large readership of practitioners (the American Economic Review article was only slightly more technical, but it placed greater stress on the model and less on its application to the real world). Again, the observations made in the 1959 article were foreshadowed in earlier work. In 1953, for example, Dahl and Lindblom wrote: “The fact is, the successful policy maker operates with good theory that is inexplicit, inarticulate, and mostly unconscious; he draws heavily on intuition or on experiences stored away in memory, fragmentary, disorderly. The theorist who fails at policy making operates with a bad theory, and it is bad partly because he has tried to keep it explicit, articulate, conscious, and orderly” ( Dahl & Lindblom, 1953, pp. 75–76). To this they added that “much factual knowledge even about consequences is inarticulate; like the knowledge of successful administrators it is based on personal or group experience and remains uncodified” (p. 77). In “The Science of Muddling Through” he wrote: “Theorists often ask the administrator to go the long way round to the solution of his problems, in effect ask him to follow the best canons of the scientific method, when the administrator knows that the best available theory will work less well than more modest incremental comparisons. Theorists do not realize that the administrator is often in fact practicing a systematic method. It would be foolish to push this explanation too far, for sometimes practical decision-makers are pursuing neither a theoretical approach nor successive comparisons, nor any other systematic method” ( Lindblom, 1959, p. 87). This emphasis on practicality and the various streams and types of knowledge that come into and can contribute to the policy process was later evident in Lindblom's notion of “usable knowledge” ( Lindblom & Cohen, 1979).
The debate over the role of analysis in incrementalism has overshadowed Lindblom's early emphasis on practical reason, application, and even intuition and judgement. He did not himself develop these ideas, except to explore the deeper issues of social inquiry. Interestingly, like the assumption about limits to cognitive capacity, this assumption does not necessarily lead to incremental decision-making. Stereotypically, the academic analyst (the theorist) will have large-scale models or theories that are too broad to be applied to specific instances or problems. He/she will also lack what Lindblom called “policy chains” or practical agency and program experience that sets a deep backdrop to decision-making ( Lindblom, 1959, p. 89). The practitioner, on the other hand, needs to deal with specific issues, lacks the time to collect all the facts or the intellectual ability to process all of them, feels the pressure of time, and needs practical results. It might be expected that practitioners (i.e., agency officials) are in for the “long game” and so will prefer incremental steps that can be easily reversed and corrected if things go wrong. Practitioners, while lacking a “rational” grasp of all the facts and variables, might actually be able to draw on tacit knowledge ( Vickers, 1965) or “rapid cognition” ( Gladwell, 2005; Klein, 1998) that can accomplish subconsciously much of what might be attempted through synoptic decision-making. This is not to say that it will either be correct or incremental – some of the worst “large” decisions have been made by decision-makers rely on their “instincts” or “experience.”
Nonetheless, Lindblom was on to something, even if he did not develop it. There are clear differences in the ways in which academic experts and practitioners approach policy issues and decision-making. The importance of experience and judgement (a more acceptable term than instinct or intuition) is well-recognized in human resource practices within governments through mentoring programs, leadership training, “experiential learning” through exchanges and “stretch assignments.” How to blend the practical and theoretical is a continuing challenge in all public and private management academic programs (the skepticism of synoptic strategic planning and the type of management education that comes with it is nicely expressed in Mintzberg's work, for example, Mintzberg, 1990; Mintzberg, 1994; Mintzberg, 2005; Mintzberg, Ahlstrand, & Lampel, 1998) where he consistently supported the idea of “learning” in an organization as part of strategic development, though he did not use the term incrementalism (and did not cite Lindblom). Quinn (1978); Quinn (1980), on the other hand, did deliberately try to balance strategy and practice under the rubric of “logical incrementalism,” a process similar to Etzioni's idea of “mixed scanning” wherein organizations adopt strategies, but in a step-by-step fashion.
As we noted above, Lindblom's original formulations of actors and partisan mutual adjustment in incremental decision-making clearly relied on inspiration both from market/price systems and interest group pluralism as it was then known. The key features of partisan mutual adjustment were its fluidity, its anonymity in the systemic sense (partisans would be mutually adjusting to those in their immediate space, but these adjustments would ripple out through a much larger and anonymous system that would reach some sort of equilibrium [a policy decision] without anyone having deliberately designed it), and its recombinant nature (issues being constantly revisited as coalitions of interests change). A somewhat more obscure assumption was that these actors were relatively disconnected from each other, much as billiard balls might bang into each other to create a resultant pattern, but without any interaction of relationships except purely reactive ones. We can see this from Lindblom's 1979 definition: “Partisan mutual adjustment, found in varying degrees in all political systems, takes the form of fragmented or greatly decentralized political decision making in which the various somewhat autonomous participants mutually affect one another (as they always do), with the result that policy making displays certain interesting characteristics. One is that policies are resultants of the mutual adjustment; they are better described as happening than as decided upon” ( Lindblom, 1979, pp. 522–523, emphasis added). Lindblom's arguments about partisan mutual adjustment, either within organizations or in the wider polity, assumed that these groups might dissolve and recombine, but in all cases would be singular entities interacting with other singular entities.
As we noted earlier, a key influence on the 1959 article must have been the theories of interest group pluralism of the day, theories that had embedded in them network concepts but which did not use that term as such. Moreover, network theory did not take off until almost twenty-five years later ( Marin & Mayntz, 1991), though it was adumbrated in Heclo's work on “issue networks” in 1978 ( Heclo, 1978). As with the limits of cognition in policy-making, Lindblom's theory of incrementalism also anticipated in some respects new theories of governance that were to come to prominence much later: complexity theory, networks, and “governance” as a master concept wherein the hierarchical state is less capable of steering and more reliant on a spectrum of groups and non-governmental actors. Klijn, for example, notes that complexity theory has not been incorporated in the public administration field, but cites Lindblom as having raised some elements of this approach in his argument that the accumulation of small steps might still lead to significant change through the accumulation of system instabilities ( Klijn, 2008). Meadowcroft, in a piece on the style of public planning required for sustainable development policy, argues that the nature of the problem actually invites planning processes that are disaggregated and conducted through “flexible networks, and multi-party collaborative interaction” ( Meadowcroft, 1997, p. 443). He specifically refers to Lindblom's incrementalist framework as being friendly to planning for sustainable development: “The adjustive processes to which Lindblom points are potentially of enormous significance to the project of planning for sustainable development. They suggest the possibility of co-ordinating complex changes across a range of social spheres (through radically decentralising analytical responsibility and encouraging practical interaction) and the potential of a reform strategy which emphasises institutional design directed to re-configuring spontaneous adjustive mechanisms to favour sustainable development friendly policy outcomes” (p. 439).
Regrettably, the long debate on network theory in the policy sciences has made scant if any reference to Lindblom or to incrementalism (for a sample, see Dowding, 1995; Dowding, 2001; Klijn & Skelcher, 2007; Marsh & Smith, 1991; Marsh & Smith, 2001; Raab, 2001; Raab, 2002; Rhodes, 2007). This is likely because partisan mutual adjustment did not, Meadowcroft's favourable argument notwithstanding, see actors (even if they were groups) as anything but disconnected monads. Networks operate differently. They are loosely connected and loosely coupled patterned relationships of information exchange and decentralized coordination. Because information is a prime medium of exchange, the role of policy learning, evolution, and adaptation is much more prominent in network theories than in coalitions or interest groups. Network members are usually more variegated than simple coalitions – including a host of different types of actors (e.g., research institutes, advocacy organizations, government agencies, etc.). Institutionalized networks may be relatively stable over time, even as they adapt and evolve, and in this would differ from the portrait of fluidity in partisan mutual adjustment, even though at other times and around other issues, they can coalesce rapidly. Moreover, networks are increasingly global in scope. A random example of all these features is the Climate Action Network (CAN) (www.climatenetwork.org), with 450 NGO members “working to promote government and individual action to limit human-induced climate change to ecologically sustainable levels” and seven regional offices in Africa, Central and Eastern Europe, Europe, Latin America, North America, South Asia, and Southeast Asia. CAN's members are actually the thirteen regional members, who each have their own membership of organizations, cascading down to the national and the sub-national level. This network cannot be adequately grasped in terms of partisan mutual adjustment, is certainly geared towards non-incremental change, and it is global.
These are serious shortcomings for incremental theory. Of course, as Lindblom pointed out in 1979, partisan mutual adjustment at some level and in some sense occurs whenever there is more than one actor in a decision space and when power over final outcomes is distributed to some degree (distributive power does suggest connection among the nodes even if the nodes together do not form a network). Even in cases where one actor has 100% of the power resources in the system, there may still be partisan mutual adjustment to capture that actor's attention. In other cases, power may be more evenly distributed, in which case the scope of partisan mutual adjustment is even greater and more determinative of outcomes. But in this model, actors have to be both risk averse and only interacting with a limited number of contiguous players so that the systemic outcomes are aggregative. Networks are different: actors have the capacity to interact much more fluidly and across a wider range of players and organizations than is captured in the idea of “partisan mutual adjustment.” This implies a very different set of dynamics as well as relationships. At the very least, networks bring a wide range of diverse actors into the policy mix, and might actually increase the chances of non-incremental and bold policy ideas winning the day.
The notion of improvements at the margin implied that these improvements were small and consequently easily reversed, and hence a positive argument in favour of incremental decision-making. Dahl and Lindblom listed reversibility as one of the advantages of incrementalism in 1953 ( Dahl & Lindblom, 1953, pp. 82–85). In “The Science of Muddling Through” Lindblom argued that a decision-maker “often can remedy a past error fairly quickly – more quickly than if policy proceeded through more distinct steps widely spaced in time” ( Lindblom, 1959, p. 86). Remedy and reversibility are two different things, but Lindblom did not take pains to distinguish them. The obvious problem with this sense of the easy reversibility (or remedy) of small decisions is that it ignores path dependency. Decisions engender institutions and practices, and reversing or remedying them, once established, is not that easy. Indeed, the idea of easy reversibility possibly contradicts another part of the incrementalist argument, that is that incremental decisions over time can lead to substantial change as each small step is added to the last one. In his 1979 response to critics, Lindblom acknowledged that big changes rarely occurred in the American system, due in part of constitutional design and as well to concentration of power among business elites, but he thought that incrementalism could sometimes “smuggle” changes into political systems: “Important changes in policy and in the political system often come about quite indirectly and as a surprise to many participants in the system. … Incremental changes add up; often more happens than meets the eye” ( Lindblom, 1979, p. 521). While it is not logically impossible to imagine small changes or steps that are in themselves reversible, accumulating over time to suddenly reach a tipping point or a phase shift to a “big change,” the actual logic in real life would suggest that as one moves along the curve of accumulated decisions, it becomes harder and harder to go back, even in single steps. In The Intelligence of Democracy Lindblom relied on a telling example of reversibility – that of a traffic engineer deciding to deal with congestion by introducing a series of one-way streets ( Lindblom, 1965, p. 150). The outcomes of even small decisions develop their passionate partisans or supportive coalitions, and so reversing those decisions is almost always a matter of dealing with the partisans and the coalitions that congealed around them.
In almost all his work on incrementalism, Lindblom assumed the United States political system, with its checks and balances and multitude of interest groups and other actors, as the model.
Partisan mutual adjustment (incrementalism as a process, according to Lindblom by 1979) requires (a) a reasonably large number of partisans or actors, who are (b) endowed with roughly similar resources as well as access to key decision points, and who (c) are prepared to lose one round in the expectation that they can engage in another series of rounds to accomplish their objectives. Together, of course, these are the classic assumptions of pluralist democracy as described in the 1950s and 1960s. Lustick pointed out that incrementalism is less likely when one set of actors has disproportionate control over the decision-making process ( Lustick, 1980, p. 344). Of course, the degree to which the American system as a whole is marked by both a multiplicity of actors and a wide range of veto points makes it perhaps the incrementalist system par excellence (at least on paper – we should note that a nation's policy process is not uniform across all policy fields; national defence in the US, for example, has never been subject to partisan mutual adjustment in Lindblom's sense). The point is simply that Lindblom made possibly exaggerated claims about the scope of incrementalism by unconsciously using the American model as his benchmark and even then, over time, he came to recognize the dominant role of elites (particularly the business elite). The greater the degree to which entire political systems diverge from the American model, the less prevalent incrementalism should be as a mode of decision-making. This point cannot be rebutted by simply claiming that all systems of decision-making, however centralized and however synoptic (at least in form), depend on interactions. Lindblom himself pointed out that partisan mutual adjustment is not the same as incrementalism ( Lindblom, 1979, p. 523) – the Chinese Communist Party has innumerable factions constantly jockeying for position, but that does not mean that the Party makes incremental decisions.
Saying that most of the world does not resemble the United States is a fairly simple critique. There are two more fundamental points to be made about Lindblom's reliance on the United States political system as his model or benchmark for incrementalism: (1) a focus on national-level politics, and (2) the assertion of value consensus in the US as a backdrop or foundation for incrementalism. On the point of national politics, nowhere in Lindblom's descriptions and explanations of incrementalism can we find any extended reference to the international system. There were, of course, references to decision-making in the Cold war context (à la Graham Allison), and to alternative socialist planning systems, but the clear point of reference for Lindblom in all of his work was national (American) politics. This was entirely appropriate to the period when Lindblom first began writing – at that time, there were sharper divisions than now in the discipline of political science between national (American) politics and comparative and international politics. More recent work has focused on the impact of global systems on national policy-making. The literature is too voluminous to cite here ( Robertson, Scholte, & Cheru, 2007; Scholte, 2005), but we can nevertheless extract two points that are salient to the discussion of incrementalism.
The first is the role of the nation-state in a globalized world. While initially the worry was that states would wither away in the face of some sort of emergent global governance regime or set of institutions, it is now clear that states in many senses have been strengthened through globalization, precisely because of the imperative of global management of disparate yet interconnected global public policy fields. Everything from telecommunications to intellectual property is now a matter of global interest because of global markets, the global reach of corporations and other organizations, and the increasingly important role of international governmental organizations, most prominently the World Bank but also international regulatory agencies (consisting of state representatives).
A second point follows from the first, and that is that global problems and challenges increasingly call for globally coordinated responses that look more like attempts at synoptic decision-making than incrementalism, and states themselves are key institutional actors. We have had two recent and almost simultaneous crises that nicely illustrate the point – the global financial crisis and the swine flu pandemic. The global financial crisis is just that – the first one to originate in the United States, and the most severe downturn since the 1930s. G-20 summits were quickly called in November 2008, April 2009, and November 2010 to coordinate a global response. Whatever the limitations of that response, it was palpably clear that the scope of the financial crisis required a global response with implications for sweeping (not incremental) policy moves both through international institutions and at the national level through strong states.
On the point of value consensus, a key liability of a national focus on the United States was a curious portrayal of the distribution of values in the polity that made incrementalism possible. Incrementalism is policy-making at the margins, small changes and small steps, repeated endlessly in small remedial ways as each previous compromise among partisan mutual adjusters gets renegotiated. This was a portrait of a policy system that focused on only marginal things, the froth and bubbles on the system's surface, and indeed Lindblom argued that incrementalism was well-suited to a system with constantly changing tastes. But tastes are not the same as fundamental beliefs, and so a natural question arose as to why most politics was incremental (i.e., small and marginal), and did not address the bigger issues, issues that would invite more synoptic, wide-ranging discussion, debate and conflict. In earlier writings he consistently argued that incrementalism as a general decision-making strategy was only possible in societies where there were many competing and changing values (hence constant mutual partisan adjustment and renegotiation of previous agreements), coupled (paradoxically) with strong and perhaps even unconscious agreement on larger issues that would otherwise divide the polity. In “The Science of Muddling Through” he said that incrementalism was only possible because the “two major political parties agree on fundamentals; they offer alternative policies only on relatively small points of difference” ( Lindblom, 1959, p. 84). Party behaviour is rooted in public attitudes characterized by “fundamental agreement on potentially disruptive issues, with consequent limitation of policy debates to small differences in policy” ( Lindblom, 1959, p. 85). He made the identical point in 1963: “We suggest, first, that much agreement emerges simply because the practitioners of the strategy share a common culture that disposes them to an important degree of consensus on values” ( Braybrooke & Lindblom, 1963, p. 133). By 1979 he acknowledged that power rather than consensus played a role in ruling out debate over some core issues such as income distribution, but still asserted that on “less dominating” issues, the influence of powerful elites was less and the scope for incrementalism greater ( Lindblom, 1979, p. 84).
It seems strange that these assertions about an underlying social consensus in America could have been made during the turbulence of the 1960s’ social movements, the Viet Nam war, Watergate, and then the emergent culture wars that are still with Americans to this day. Perhaps Baumgartner and Jones's punctuated equilibrium model saves incrementalism as a special case between periods of turbulence, but it does not address the broader assumption of value consensus. Lindblom's assertions about this consensus on positive values, and its foundation for incrementalism, were exaggerated – there were indeed strong differences in values and much less consensus than he averred, for example around issues of race and equality. Lindblom may have been referring to fundamental agreement around constitutional principles (e.g., constitutionalism, accepting the outcome of elections) but he never made that clear.
This article has tried to assess incrementalism along two dimensions: first, the viability of the original assumption/arguments in its own terms and at its own time, and second, the viability of that assumption in light of more recent theory. The model's assumptions/assertions about the limits of human reason, and its emphasis on tacit knowledge were insightful at the time and continue to resonate with the most recent work on decision-making. Partisan mutual adjustment was entirely consistent at the time with the prevailing pluralism of American political science, and in some respects foreshadowed modern work on networks and complexity. But as we pointed out, this assumption treated partisans as disconnected monads. Network theory connects them. The notion that incremental decisions are inherently more easily reversed or remedied seemed like common sense, but even at its origins clashed somewhat with Lindblom's observations about congealed partisan coalitions. Today, theories of path dependency provide a richer approach to understanding the remediation and reversal of decisions. The final assumption was possibly wrong at the time (both for the turmoil of the 1960s, but even in terms of America's fractious history on a host of issues), but it depends on what Lindblom meant. The text in the 1959 article seems to suggest consensus over substantive issues rather than just about procedure (though this is important as well). Would such a claim of consensus over fundamental societal values resonate in today's world, either in the United States or in Europe? Probably not.
Incrementalism as a theory of the policy process seems weak, therefore, on three out of five assumptions. It is not adequate to a policy-making system that is globalized, networked, value-driven, and possibly more conflictual (as partisans raise stakes and refuse to “mutually adjust”). That it holds up well on two of its assumptions, however, is a testament to Lindblom's insight into human reasoning and politics, and decision-making more broadly in organizations. The theory itself, however, is limited as a foundation for analyzing contemporary public policy and the policy process. It will continue in the form of a metaphor, or a shorthand term to describe what appear to be “small decisions,” but for real insight into our times we need more powerful tools and models.