Morowitz (2002) uses the origin of agriculture as an example of emergence. He argues that, if one considers the possibility space of human-environment interaction, agriculture is clearly an attractor, i.e., a conjuncture or genus of conjunctures, that has been colonized several times. If the tape were played again, the result would probably be agriculture.

It is perhaps pedantic to observe that the domestication of animals and plants occurred much more often than he suggests. The instances that have come to popular attention did so because domesticates were moved beyond their endemic range by people who left qualitatively new types of trace in the landscape. It is probable that animals were domesticated in the Pleistocene, for example. Plants and animals have certainly been husbanded far beyond the cradles of civilization, but there was no explosive complexification to draw attention.

More than 99% of human history has been nonurban, with very little technological specialization, no spectacular pockets of local order (POLOs) such as great buildings, and no evidence of conspicuous wealth, writing, or social stratification. Village-based agricultural societies and hunter-gatherers have different types of technology, but there is little to distinguish them in terms of social complexity. Villages are a little larger and more visible than hunting camps, but pastoralists often leave almost no trace in the landscape. Compared to the Upper Paleolithic art sites of southern France, the subsistence agriculture of prehistoric Europe and sub-Saharan Africa seems bland and unstructured. The really interesting topic, as Morowitz’ book amply demonstrates, is not agriculture, but the complexification with which it is sometimes associated.

Sir Herbert Spencer believed that primitive humans were solitary, asocial beasts. Competition for limited resources brought Malthusian stress that created a natural selection pressure. This selection forced us to compete with each other, leading to conflict and war. More civilized populations that could aggregate in ever larger, more co-operative units were better able to win wars. According to Spencer (1985), “...this formation of larger societies by the union of smaller ones in war, and this destruction or absorption of smaller ununited societies by the united, larger ones, is an inevitable process through which the varieties of men most adapted for social life supplant the less adapted varieties.”

One of the most striking features of Spencer’s model is that the synergetic constraints restricting adaptive potential are inexorable and universal. Evolution becomes a deterministic process that drives “progress,” turning apes into savages and savages into gentlemen. This progressive model remains an important structuring principle in contemporary society, especially in advanced technocracies. Whenever a science fiction story portrays a being that spontaneously “evolves” into a higher life-form, we see Spencer’s cultural legacy.

Discursive scientists are reluctant to give Spencer’s theory serious attention because it came to be associated with the laissez-faire economics of the empires of the late 19th century and the genocidal policies of the Third Reich. In this respect, physics is a more generous science than sociology. Physicists do not repudiate Einstein because he was wrong about physical determinism or because his work made it possible to develop atomic weapons. They retain those elements of his work that remain useful and add new ones of their own.

Spencer’s hypothesis was not ridiculous. He could have known very little about the biology of great apes. Malthusian stress was then a universal feature of urban life and, indeed, of life in the rural hinterland that supplied urban markets. The best reason for rejecting Spencer’s model is that it cannot be reconciled to evidence gathered after Spencer’s death. Ethnographic research suggests that hunter-gatherers are much more predisposed to share goods and resources than urbanites are. Population pressure can be reduced by sanctioning alternative routes to sexual fulfillment and by practicing celibacy, infanticide, and other birth control methods.

There are very good material reasons why hunter-gatherers should avoid urban life. Sahlins (1972) described hunter-gatherers as the “original affluent society” because they were able to produce most of their subsistence needs in about 4 hr a day. Producing enough to meet domestic needs using agriculture increases that workload, but the burden of supporting a primitive urban society is almost unbearable. Intensive mixed agriculture triples that workload, creates malnutrition, and aggravates infant and obstetric mortality rates. The urban revolutions also brought slavery and engaged the Malthusian ratchet. These were not developments that seemed like a good idea at the time. It probably took a major co-evolutionary catastrophe to drive humans out of stable, extensive conjunctures into an urban configuration (Binford 1983, Chapter 8).

The simplest way to salvage Spencer’s theory is to allow that it is not universally but only locally true. We divide the possibility space into two attractors. One is demographically extensive, with no or only weak Malthusian stress; the other is demographically intensive with severe Malthusian stress. The analogy with physics is rather clear: The extensive attractor is like a cloud of unconnected atoms floating in the void, whereas the intensive attractor is like a highly connected mass whose behavior is made predictable by synergetic constraints.

This small adjustment gives us a model in which socio-natural dynamics are metastable. We invoke a pivotal event or perturbation that flips the system from a demographically extensive attractor to a demographically intensive one. People must have been driven into these aggregations by the need for patchily distributed resources in times of environmental stress. In some regions it may have been stands of wild cereals; in others it could have been water for irrigation.

Once the transition was made, Spencer’s model would apply without loss of generality, although his belief that sociability was the product of the Malthusian ratchet can no longer be sustained. Humans, like all the other great ape species, are innately sociable. Where hunter-gatherers tend to share subsistence resources and villages compete with other villages, we would now see people competing with their immediate neighbors. Rapid reproduction rates and co-evolutionary stress would put ruthless delinquents at a small selective advantage. Spencer's argument is turned on its head; cities actually made people less sociable.

Population growth would have made the process irreversible. The people living around the conurbation would have had strong incentives to discourage an influx of hungry refugees from the city. Trade between the centre and periphery would be favored. But what do you trade for subsistence goods when you live in a city made of mud bricks and stones? The answer seems to be that the city forms an army to protect the fields and raises taxes to pay for it. It builds infrastructure, such as irrigation systems and roads, to raise productivity. It trains people to bake clay and make pretty pots and to melt stone and make pretty metals, and then persuades its neighbors that these goods and services are intrinsically desirable. The amazing human capacity to negotiate nonadaptive cultural norms comes into its own, as they persuade each other that this is “normal.” These geegaws become so valuable that farming communities give their food away to possess them.

Competition within the cities would have driven social stratification. Systems of laws would be developed to keep the peace. Slavery and capital punishment would become normal. As the population grew, the size of the exploited hinterland would grow too. Small trading posts would be established on the periphery to channel food to the centre and craft goods to the hinterland. The urban elites would need to control subsistence resources, and this would require a system of accounting and a way of managing the production of the high-status goods that brought those resources in. Markets would be established and craft specialists recruited to meet this demand. The cities would fill up with craftsmen working for the elites who would feed them with some of the food they brought in. The civilization would begin to expand, driven by the Malthusian ratchet.

Making Spencer’s model metastable is an undoubted improvement. We no longer have to look for the cause of complexification. All that is required is a co-evolutionary perturbation that drives people into a conurbation and holds them there long enough for the ratchet to engage. There were probably different perturbations in different arenas. Some of these complex societies simply collapsed, and the survivors went back to subsistence agriculture or, in some cases, to hunting and gathering. Others had access to the right combination of resources and became self-sustaining.

However, there are problems with the metastable model too. Basically, it gives us only two routes to complexification: spontaneous aggregation or trade and acculturation. In a world without good communications, this would suggest a contagious model. There is ample evidence of contagious development, but the diffusion process was remarkably slow. Urbanization crawled out of the Middle East and crept across Europe like a slow cancer. It reached southeastern Europe in the Bronze Age (about 5000 years ago) and the northwest Atlantic fringe in the Iron Age (about 2000 years ago), but it only engulfed Scandinavia in the 17th century. For metastability to be a complete model, there should be very little complexification beyond the reach of these burgeoning civilizations.

In fact, agriculture spread far beyond the cradle of civilization and may indeed have been discovered independently in many regions. However, extreme Malthusian stress and runaway complexification did not spread with it. The first Middle Eastern civilizations had bronze-working and advanced pottery technology, so European archaeologists tended to see the pyrotechnologies as the fingerprint of urbanization. However, in the 1960s and 1970s, they discovered that pyrotechnology, like agriculture, spread much faster than urbanization.

The Bronze Age settlements of Crete do indeed look a little like urban units, but those of nearby Mycenae look more like fortified villages with warrior elites. No doubt there were famines and infant and obstetric mortality, but nothing like those experienced by Middle Eastern cities, where density-dependent diseases would have claimed many. It is conceivable that Mycenae was a second-order contagious development from the Middle East, but what about the Bronze Age societies on the Atlantic Fringe?

The later Neolithic and Bronze Age communities of these regions produced some beautiful artifacts and impressive POLOs, including rich burials, stone circles, and megalithic structures, that naturally make people think of social complexification. However, there is no evidence here of demographic runaway or, indeed, of any sort of physical crowding. This is not a spontaneous aggregation event. Neither is there much likelihood of contagious development through contact with the Middle East because these sites are rather early, possibly even older than the Mycenaean (Renfrew 1979).

There was little transport infrastructure to facilitate contact between Britain and the Mediterranean before the second century BC. Exchanges of material goods would have been possible over such distances, but could hardly have been intensive enough to aggravate Malthusian stress on the western fringe. We have no reason to believe that the Bronze Age communities around Stonehenge would ever have developed into an urban civilization spontaneously. This is a different type of complexification, one in which most competition is directed between neighboring communities rather than between neighbors in the same settlement.

Metastable models based on the Spencerian mechanism do not explain complexification of this sort or, indeed, the incipient complexification of the European Palaeolithic. However, they capture the core dynamics of urban systems quite well, at least until the 20th century, when we start seeing cities with virtually no Malthusian stress. We therefore need to generalize them in a way that accommodates the dynamics of less severely constrained systems. Building innovation into the model does this nicely. We admit the possibility that the level of synergetic constraint can vary. System behavior is predictable and organized within conjunctures, and more chaotic and unpredictable when constraints are relaxed and adaptive potential is released.

The intensity of those synergetic constraints is often linked to demographic and other co-evolutionary pressures. If food is scarce or the demand for resources high, societies end up playing a “zero-sum game” in which each person’s gain is another’s loss. Structured behaviors become cultural markers and constraints harden, initiating a Phoenix Cycle of entrenchment and collapse. After collapse, adaptive potential is freed up, and people can develop mitigating strategies that reduce conflict, conserve those aspects of old life ways they value, and abandon those they do not. The interaction of memory with anticipation and aspiration gives historical systems their ex post path dependence, which makes narrative possible, but also provides enough adaptive wiggle room to make them unpredictable ex ante.

The history that actually happened represents only one of many possible pathways, and any model that “predicts” that trajectory to the exclusion of others can usually be dismissed on a priori theoretical grounds. Historical systems are potentially innovative, and their behavior is uncomputably complex. They only become metastable, i.e.,computably complex, when synergetic constraints are unremitting and adaptive potential is curtailed.