The Power and Pervasiveness of Spontaneous Order

Spontaneous order is crucial for understanding human institutions as fundamental as language and the law, and as pervasive as morals, markets and money. It is also central to a key defense of individual liberty. Nevertheless, many people find the notion of spontaneous order counter-intuitive, seeing nothing between planning and anarchy. Some respectable academics have even claimed that the notion is incoherent.

This piece will therefore aim to clarify spontaneous order as it relates to human institutions, and make it more comprehensible.1 It will suggest a characterization of spontaneous order that may help make it easier to identify. And it will highlight the operation of spontaneous order in often unsuspected fields, ranging from natural resource management to academia and artificial intelligence.

Meaning

What is spontaneous order? When people act spontaneously, they do so impulsively, typically without forethought or concern for the consequences of their actions. Similarly, spontaneous orders feature neither planning nor intention, and just emerge; they are self-generating and self-organizing.

Commentators often refer to the ‘invisible hand’ mentioned by Adam Smith. Or they adopt a variation of a formula that was originated by Adam Ferguson in the eighteenth century, and popularized by F.A. Hayek in the twentieth. Spontaneous order, they say, is ‘the result of human action, but not of human design’. Although strictly true for institutional examples, that shorthand description is unfortunately open to misunderstanding, especially if taken in isolation. Notably, it does not adequately distinguish between orders and their components.2

An order is a state of affairs that consists of items which exhibit some regularity or pattern, and the relations among them. Orders have conventionally been considered to be either natural or artificial. They are wholly natural if they and their constituent related items exist independent of human action, e.g., crystals, cats. They are artificial if they are man-made, e.g., sonnets, skyscrapers. Artificial orders are always constructed: they result from a conscious agent intentionally imposing some pattern or arrangement on the constituent items, typically to achieve a goal. Thus, parts are assembled to build a car; records are alphabetized to facilitate retrieval.

Orders can, however, also be of third kind: they can be spontaneous. Unlike wholly natural orders, spontaneous orders can have human actions as their components. Unlike artificial orders, spontaneous orders are not constructed: they involve no intentional coordination of the constituent items. They arise when the items fall into a pattern without themselves intending to do so, or their being arranged by an external agent. Natural examples include snowflakes, and iron filings’ reaction to magnetism. Spontaneous orders of human actions exist when, without the intervening agency of any coordinator, the actions of multiple dispersed individuals give rise to an overall pattern. The order was not intended by anyone; each participant was simply acting to achieve his own, particular objectives. A spontaneous order of human actions can thus be characterised concisely as ‘an unintended order of intentional action’. More generally, spontaneous orders are self-organizing, complex adaptive systems: a spontaneous order exists when a pattern that has not been arranged by any coordinator emerges from the behaviours of multiple, dispersed individual components. Core examples of spontaneous order include language, the common law, and the outcomes of evolution.

Obstacles to understanding

Spontaneous orders are often not recognised. Organisation commonly results from the deliberate imposition of a pattern or arrangement. Deliberate design thus tends to be the default when explanations are sought. The existence of a designer—however elusive—is typically presumed: recall Smith’s metaphor of the ‘invisible hand’…. Even people who eschew theological Intelligent Design in favour of biological evolution, often assume design in respect of human institutions: how could law or money exist without it, they wonder.

But consider language: who or what could have designed it? Is it plausible that prior to the existence of language, any human agent could have authoritatively associated meanings with particular sounds or symbols throughout a community? Language probably evolved from many separate individuals seeking to solve particular communication problems. They assigned meaning to sounds or symbols in specific situations; some of those associations were sufficiently accepted over time by a community to constitute its language.

Recognising the price system as an example of spontaneous order may be at least as challenging, especially to non-economists. Extreme complexity3 is commonly thought to exacerbate the need for conscious direction. Arguably, however, deliberately constructed orders are less capable than spontaneous orders of handling complexity. Constructed orders are inflexibly limited by what the designing mind can encompass, and much of the information needed is ‘never so given to a single mind…’ (Hayek 1945, p.530).

The limitation is not one of computing power. Even if technological advances could provide enough to deal with any number of given, static inputs, the relevant inputs for economic coordination are neither static nor simply given. Dispersed among innumerable human actors, they are instead a) constantly subject to being adjusted in response to changing circumstances, including the other inputs; and b) reflect knowledge that is both tacit and privileged. Individuals have knowledge of how to do things, and of local particularities of time and place that include their own personal preferences. Their knowledge is not directly available to other people, cannot reliably be predicted, and both affects and is affected by others’ preferences.

According to Hayek, only a spontaneous order based on general rules is capable of integrating that sort of dynamic, interactive information into a simple, real-time signalling system. Full central planning is not only unnecessary, but theoretically impossible: it is incapable of solving the ‘knowledge problem’. This limitation is shared by all general command and control economic systems; it fatally undermines socialism.

The extent to which spontaneous orders are optimal is controversial. The environment that supports the optimality of spontaneous market order has several special features. They include not only well-defined and enforceable private property rights, freedom of contract and freedom of exchange, but also the rule of law, and a moral code of behaviour that legitimises those conditions.

Whether or not spontaneous orders are generally optimal, spontaneous orders are better than constructed ones insofar as they respect individual liberty. They do so in three ways. First, spontaneous orders are essentially non-coercive: by their very nature, they involve no imposition, and a fortiori, no forcible imposition. Second, the mere existence of spontaneous order proves that planning is not the only way that order can be established. Since alternatives are available, imposition cannot be assumed to be necessary, and a basic presumption of coercive government is thereby refuted.

The third way that spontaneous orders support freedom, is that they require it in order to function. For the privileged knowledge that only individuals possess to be used optimally, those individuals must be free to act on it. If they are not, the order ceases to be self-adjusting and self-correcting. Interference with spontaneous orders impedes the orders’ operation and typically makes things worse. To the extent that the benefits of spontaneous order are valued, this constitutes a strong argument for freedom and against coercive intervention of any sort, however well-intentioned.

Correcting Confusions

Despite its significant role in supporting individual liberty, and in explaining key human institutions, the notion of spontaneous order may still seem dubious: how, after all, can something be ‘the result of human action, but not of human design’? The confusion often results from not recognising that orders are neither the same as, nor reducible to, either their constituent items or the relations connecting them: ‘spontaneous’ modifies ‘orders’, not the constituent parts. Insofar as the components of the order are human actions, and actions presuppose intention, design is necessarily present in institutional orders. What is not intended or deliberately designed, is the emergent, meta-phenomenon—the order—that arises from those components’ complex interactions. The integration of the components into a coherent system ensues automatically, without any arranger or planner actively coordinating them… not even the human members themselves intend or seek that ordered outcome.

The absence of a coordinator is the definitive feature of spontaneous order. Commentators offer many illuminating characterisations of spontaneous order, but typically fail to identify or emphasise this crucial element. When determining whether an order is spontaneous, it does not matter whether the coordinator consists of a single mind (a curator arranging an exhibition) or a group (a government drafting an army), or whether its decisions may subsequently be subject to revision (those of a trial court jury): the coordinating agency is ‘final’ only in respect of designating an authoritative4 outcome.

Equally irrelevant is whether the items being coordinated might themselves be constructed orders: consider the interplay of constructed corporations in the spontaneously ordered economy. Nor does it matter whether the coordination is imposed by one of the constituent items (the officer of a private club) or some outside agency (the university administration). It’s also irrelevant whether the imposed pattern was wholly or partly planned in advance, or was the subject of explicit deliberation: even if adopted on the spot (a salad improvised from leftovers), being constructed is what counts. Finally, the absence of coercion and being unintended are each necessary but not sufficient conditions for constituting a spontaneous order. Constructed orders (e.g., partnerships) can be formed cooperatively and accepted voluntarily, and not all unintended consequences are orderly.

Unlike the Ferguson formula of ‘human action but not human design’, the alternate formula—’the unintended coordination of intentional action’—clearly differentiates orders from their elements. It therefore also helps clarify the relation of spontaneous order and reason. Orders that have not been deliberately planned or intended are sometimes considered irrational. But even by that very restrictive characterisation of rationality, if the order’s components are intended human actions, those actions may be at least partly rational, and the order itself will simply be non-rational. It may also be that the order emerged (and survived) because it served some important human purpose. If so, spontaneous orders might be recognised as satisfying a more substantial understanding of rationality.

The alternate formula can also help to resolve other questions. Are works created by committees examples of spontaneous order? The outcome is often a compromise that was not intended, designed or even wanted by any of the participants. And it may be difficult to identify any specific agent or coordinating intelligence responsible for the result of the committee’s work—recall the old joke about camels being committee constructs. Committee outcomes may therefore seem to qualify as ‘the result of human action, but not of human design’. But regardless of their quality, the outcomes hardly seem spontaneous. Committee members are—at least nominally—presumed to be aiming at a common goal. Moreover, the official outcome gets determined by some agreed procedure of collective choice. Accordingly, the results of committee actions—like those resulting from negotiations—would seem not to be examples of spontaneous order. Insofar as unintended consequences are coordinated, they are not spontaneous.

Where might genuine instances of ‘the unintended coordination of intentional action’ be found?

Extended Example 1: Natural Resource Management

A perhaps surprising example is in the management of shared natural resources. Its operation there has confounded the expectations of conventional wisdom and the implications of basic game theory. Economists used to hold that there was a ‘tragedy of the commons’ that necessarily affected non-excludable and rivalrous natural resources. These are resources whose use cannot readily be prevented, and whose consumption reduces the amount available for others. It was believed that such natural resources would inevitably be over-exploited and destroyed unless they were coercively managed by governments.

Nobel Laureate in economics Elinor Ostrom conclusively disproved this notion. She showed that complex adaptive systems can and do emerge that allow common-pool resources to be cared for and used sustainably. Moreover, those emergent systems conserve the resources better than state regulation does. Empirical evidence from locations world-wide detailed such systems, and related not only to pastures, fishing waters, and forests, but also to groundwater basins and even public safety services provided by metropolitan police forces.

Extended Example 2: Literature and the Arts

Where else might spontaneous orders be found? Literature initially seems an unlikely source: literary works are intimately associated with and identified by reference to their authors. ‘Literary works’ in this context includes all writing in a genre usually intended for publication in print or other media, both fiction and non-fiction; they need not be ‘literary’ in the sense of high culture or achievement.

Literary critic and professor of English Paul A. Cantor has nevertheless argued forcefully that:

• …the serialization of novels as it developed in the nineteenth century offers a good example of spontaneous order—of a self-regulating or self-correcting mechanism. Novelists could experiment with different characters, situations, and plot developments, and see what worked with their audience, thus allowing for midcourse corrections in the composition of a novel. (2002, pp.53-4)

Cantor cites the abbreviated Ferguson formula to support his assertion, stating that the serialized novel is ‘the result of human action but not of human design.’

Serialized novels were produced not only (famously) by Dickens, but also by Trollope and Thackeray, Thomas Hardy and George Eliot, and even Dostoevsky and Tolstoy. The authors released a few chapters at a time, typically in popular periodicals; material was distributed before the entire novel was written, and was often produced to tight deadlines. Serialized novels frequently contained inconsistencies, as characters and plot lines got changed in the course of the novel’s development. Cantor offers many examples of how production of the nineteenth century novel did not fit either the Romantic notion of the autonomous artist, or the conventional New Criticism picture of the solitary author creating the perfectly planned work.

He acknowledges that:

• …conscious human minds are involved at every stage of the evolution of a novel as we have described it. Authors consciously write their novels installment by installment, readers consciously make decisions as to which parts of the novels they like, the authors in turn consciously decide how to respond to the feedback they get from their audience, and so on. But this process can still be regarded as a form of spontaneous order because no single mind controls it from start to finish. (Cantor 2002, pp.63-4)

This is an intriguing claim, but ultimately an implausible one. Neither Cantor’s dual test here, nor the mechanisms mentioned in the previous quotation, capture what is essential for being a spontaneous order.

First, being produced by multiple minds is no obstacle to a literary work’s being constructed. Examples abound. Some are deliberately co-authored: the novels of Ellery Queen, philosophy texts by Rasmussen & Den Uyl. Some result from committee agreements: the US Constitution. And some have been generated collectively by groups intentionally assembled for that purpose: TV scripts of series (e.g., Friends) employing writers’ rooms. There are also works in which fictional characters are deliberately perpetuated by authors who did not create them. These include not just the many pastiches involving Sherlock Holmes (e.g., novels of Laurie King) or Elizabeth Bennett (by, e.g., P.D. James), but cases in which trustees of the original creator license new works featuring the characters (as Jill Paton Walsh has been licensed by the Dorothy Sayers estate). Finally, popular genre series are sometimes ‘branded’ with the name of the initial volumes’ author (e.g., James Patterson) even though subsequent volumes are written by other people. In each case, even though more than one author is involved, the participants were deliberately involved in crafting a literary work. Care is needed properly to identify the nature and value of the various contributions, but in every case they were coordinated by some person(s) to form the outcome. Consequently, the outcomes in these examples—the published literary works—are not spontaneous.

Second, insofar as ‘start to finish’ control involves total end-to-end planning, it is often absent even from the carefully constructed works of a single author. Trial-and-error, sequential revision in response to self-editing and/or external criticism, and even serendipity can all play a part without the resultant product lacking a unifying organizer. So long as some person has designated an authoritative version, a constructing mind is evident: the hand is visible.

The earlier features of spontaneous order that Cantor cited—’being a self-regulating or self-correcting mechanism’—also do not suffice. HVAC systems with thermostats are the product of deliberate human design; so are self-policing trade associations. Moreover, serialized novels were not in fact either self-regulating or self-correcting. At each stage of every particular work, some mind (even if not always that of the nominal author) decided how to coordinate the various inputs provided by the publisher, printer, sales outlets, etc..

The feature of the serialized novel that most plausibly supports Cantor’s claim is its reflecting reactions from the reading public. Readers individually seeking their own entertainment incidentally provided valuable information about the acceptability of instalments’ contents. The information was provided independently, without any coordination, and typically without any intention of directly shaping the literary work. Nevertheless, though the feedback was spontaneous, it does not suffice for the resulting work to be. Readers’ responses were only efficacious if they were taken into account by whoever was responsible for composing subsequent chapters, and for integrating the total work. Serialized novels were constructed, albeit in a complicated way.

Do other art forms provide any examples of spontaneous order? One candidate might be the unmediated output of a jazz ‘jam’ session. Although the participating musicians presumably intend to make music together, their improvised interaction resembles a conversation, with no predetermined content, shape or direction. Having neither score nor conductor, the ensuing musical piece might well seem an example of spontaneous order. In contrast, despite their suggestive names, improvisational theatre and improvisational comedy seem less plausible examples, insofar as a situation or a prop is specified as the basis for the actors’ contributions, and the result is moderated. Similarly, although also suggestive nominally, ‘found art’ is considered art because some artist declares it to be so, imposing the category on some ordinary object.

Extended Example 3: Genres, Professions, Academic Disciplines

Nevertheless there is something about literature and the art that does seem to qualify as spontaneous order: the overall practices that they constitute. ‘Practice’ here refers to a loosely organised but identifiable activity or institution, often but not always rule-governed. The core examples of spontaneous order—language, money, law—are practices. So, arguably, are artistic genres, the professions, and academic disciplines.

Consider the novel as a literary genre. There is no reason to suppose that the authors ordinarily recognised as contributing to its development were intending to establish a new literary form for use by other writers. They were simply seeking to communicate and have their fictional prose narratives read. Even if they consciously experimented with new techniques, it was in an attempt to better express themselves and reach audiences. As a literary genre, the novel had no deliberate designers. Nor is there any authority that can definitively determine what counts as an example of the novel: the genre has just evolved over time. The same might be said of other forms of literary and musical expression, and of the visual and plastic arts. The sonnet, the symphony and sculpture have evolved from the actions of diverse artists’ creating their own art works, without any coordinating agent imposing order on their attempts. The genres do appear to be instances of spontaneous order.

The reasoning that applies to artistic genres, seems to apply equally to the professions and to academic disciplines. Both seem to be examples of what, following Hayek, might be termed a polycentric order: ‘The order results… from the separate responses of the different elements to the particular circumstances which act on them….” (Hayek 1964, p.6)5 Typically, in the uncoordinated course of addressing particular situations as they arise, individual practitioners independently exercise their personal judgements, and experimentally seek solutions. Operating within an existing practice, each practitioner’s initiative both takes into account and influences the activities of other practitioners. This ‘self-coordination of independent initiatives leads to a joint result which is unpremeditated by any of those who bring it about.’ (Polanyi 1962, p.3) Over time, some approaches come to be recognized as a part of what it is to engage in a practice—to practice medicine or engineering, to be an accountant or an architect.

A similar process operates with respect to academic disciplines. What it is to be an academic philosopher, physicist, or historian has developed spontaneously over time. Academics typically have their own, uncoordinated, personal reasons for choosing which topics to explore. In some subjects, the results of their investigations may be cumulative and inform subsequent research. Even in the contested social sciences and liberal arts, the results of a strong trend may shape the direction of professional activity. Consider the ‘linguistic turn’ away from the world taken by academic philosophy…. The standards employed by academic departments and professional associations typically reflect the interaction of multiple, often unidentifiable inputs that were directed at their own ends, rather than at constructing an overall order.

Extended Example 4: Diverse

Where else might examples of spontaneous order be found? Almost anywhere, according to Michael Polanyi (1941, p.438):

• The social legacies of language, writing, literature and of the various arts, pictorial and musical; of practical crafts, including medicine, agriculture, manufacture and the technique of communications; of sets of conventional units and measures, and of customs of intercourse; of religious, social and political thought; all these are systems of dynamic order which were developed by the method of direct individual adjustment….

To illustrate just how widespread but unrecognized spontaneous order is, consider a few short examples, all prompted by unrelated broadcasts on BBC Radio 4, a popular British radio station.

The first concerns ‘street furniture’. A commonly used illustration of spontaneous order is that pedestrians seldom collide—even on a crowded airport concourse—because each makes the necessary adjustments to avoid the others. The emergence of a comparable spontaneous order was expected to result from removal of physical traffic barriers from a busy central London thoroughfare, the location of several major museums. According to the local council,

• The new single surface design is kerb-free with the minimum of street furniture and barriers. Having a less distinct ‘track’ for through traffic encourages motorists to drive more cautiously and slowly, with greater awareness and consideration for pedestrians. It also provides greater flexibility in the way Exhibition Road can be used in the future.6

And so it proved. No traffic accidents were observed during the subsequent survey, even though car speeds increased.

The second example is equally mundane. How do people learn to cook a poisonous plant safely? Cassava (a.k.a. yuca) contains cyanide, but is nevertheless the source of tapioca, and is also a staple source of carbohydrates for much of the developing world. It can be so, because of cultural transmission. By trial and error, some groups discovered ways of preparing cassava that preserved life and nutrition. Traditions that favoured the safe methods developed; they enabled people to survive and thrive, and helped their knowledge to spread.

The next examples also involve trial and error, albeit by extremely sophisticated artificial intelligence programs developed by DeepMind. Instead of using ‘thousands of rules and heuristics handcrafted by strong human players that try to account for every eventuality in a game’, the computer program AlphaZero replaces

• these hand-crafted rules with a deep neural network and general purpose algorithms that know nothing about the game beyond the basic rules. To learn each game, an untrained neural network plays millions of games against itself via a process of trial and error called reinforcement learning. At first, it plays completely randomly, but over time the system learns from wins, losses, and draws to adjust the parameters of the neural network, making it more likely to choose advantageous moves in the future. The amount of training the network needs depends on the style and complexity of the game, taking approximately 9 hours for chess, 12 hours for shogi, and 13 days for Go.7

That’s how little time it took for it to ‘become the strongest player in history’ for each game.

Even more significantly, another DeepMind program, AlphaFold2, has used deep learning to make a fundamental biological breakthrough, accurately predicting how proteins will fold to within the width of an atom. Since proteins are involved in catalysing chemical reactions (enzymes), fighting disease (antibodies) and acting as chemical messengers (hormones such as insulin), this major development in discerning their structure may revolutionise the discovery of new treatments.

None of these examples of spontaneous order was recognized as an example of it, even though the outcomes emerged without any constructing or coordinating agent arranging the components.

Conclusion

Why is it important to identify spontaneous orders? Primarily, because where they operate, coercive interference is not just unnecessary, but positively counterproductive: freedom is needed to obtain the benefits of scope, self-adjustment and self-correction. Spontaneous orders are likely to be more effective than stifling constructed ones in making best use of diffuse knowledge, and allowing improvements to emerge.

As Hayek said of the spontaneous order of the price system:

• … if it were the result of deliberate human design, and if the people guided by the price changes understood that their decisions have significance far beyond their immediate aim, this mechanism would have been acclaimed as one of the greatest triumphs of the human mind. (Hayek 1945, p.527)