Decentralisation (or decentralization) refers to various means of more widely distributing decision-making to bring it closer to the point of service or action. It occurs in a great many contexts in engineering, management science, political science, political economy, sociology and economics - each of which could be said to study mass decision-making by groups too large to consult with each other very directly.
Law and science can also be said to be highly decentralised practices of human beings. There are serious studies of how causality and correlations (respectively) of phenomenon can be determined and agreed across an entire nation, or indeed across the entire human species spread across the planet. While such institutions as the International Criminal Court or Intergovernmental Panel on Climate Change seem highly centralized, in fact they rely so heavily on the underlying legal and scientific processes that they can be said to simply reflect, as opposed to impose, global opinion.
A central theme in all kinds of decentralisation is the difference between a hierarchy, based on authority and two players in an unequal-power relationship, and a market, based on a lateral contract between two players of roughly equal power. The more decentralized a system is, the more it relies on lateral relationships and the less it can rely on command or force. In most branches of engineering and economics, decentralisation is narrowly defined as the study of markets and interfaces between parts of a system - this is most highly developed as general systems theory and neoclassical political economy.
Karl Marx, in Das Kapital, observed that the historical progress of economies from feudalism to capitalism was a classic example of decentralisation: It relied correspondingly less on the authority of a "nobility", and more on flexible systems of control of capital - the markets themselves, which were relatively merciless in driving down the price of labour as one of many factors of production, or punishing poor investment strategy - English nobility could be impoverished by a single bad investment decision, which could not have happened under any feudal system.
However, there are limits to decentralization as a strategy. Any relaxation of direct control or authority introduces the possibility of dissent or division at critical moments, especially if what is being decentralised is decision-making among human beings. Friedrich Engels, in a famous response to Bakunin which refuted the argument of total decentralisation, or anarchism, scoffed "how these people propose to run a factory, operate a railway or steer a ship without having in the last resort one deciding will, without single management, they of course do not tell us".
In "On Authority, Engels wrote of democratic workplaces that "particular questions arise in each room and at every moment concerning the mode of production, distribution of material, etc., which must be settled by decision of a delegate placed at the head of each branch of labour or, if possible, by a majority vote, the will of the single individual will always have to subordinate itself, which means that questions are settled in an authoritarian way. The automatic machinery of the big factory is much more despotic than the small capitalists who employ workers ever have been."
Modern trade unions and management scientists tend to side strongly with Engels in this debate, and generally agree that decentralization is very closely related to standardisation and subordination, e.g. the standard commodity contracts traded on the commodity markets, in which disputes are resolved all according to a jurisdiction and common regulatory system, within the frame of a larger democratic electoral system which can restore any imbalances of power, and which generally retains the support of the population for its authority.
However, a strategy of decentralization is not always so obviously political, even if it relies implicitly on authority delegated via a political system.
For example, engineering standards are a means by which decentralisation of supply inspection and testing can be achieved - a manufacturer adhering to the standard can participate in decentralised systems of bidding, e.g. in a parts market. A building standard, for instance, permits thebuilding trades to train labour and building supply corporations to provide parts, which enables rapid construction of buildings at remote sites. Decentralization of training and inspection, through the standards themselves and related schedules of standardized testing and random spot inspection, achieves a very high statistical reliability of service, i.e. automobiles which rarely stall, cars which rarely leak, and the like.
In most cases an effective decentralisation strategy, and correspondingly robust systems of professional education, vocational education, and trade certification, is critical to creating a modern industrial base. Such robust systems, and commodity markets to accompany them, are a necessary but not sufficient feature of any developed nation. A major goal of the industrial strategy of any developing nation is to safely decentralise decision-making so that central controls, e.g. as in a command economy, are unnecessary to achieving standards and safety. It seems that a very high degree of social capital is required to achieve trust in such standards and systems, and that ethical codes play some significant roles in building up trust in the professions and in the trades.
The consumer product markets, industrial product markets, and service markets that emerge in a mature industrial economy, however, still ultimately rely, like the simpler commodity markets, on complex systems of standardization, regulation, jurisdiction, transport, materials and energy supply. The specification and comparison of these is a major focus of the study of political economy. Political or other decision-making units typically must be large and leveraged enough for economy of scale, but also small enough that centralised authority does not become unaccountable to those performing trades or transactions at its perimeter. Large states, as Benjamin Franklin observed, were prone to becoming tyrannies, while small states, correspondingly, tended to become corrupt.
Finding the appropriate size of political states or other decision-making units, determining their optimal relationship to social capital and to infrastructural capital, is a major focus of political science. In management science there are studies of the ideal size of corporations, and some in anthropology and sociology study ideal size of villages.
All these fields recognize some factors that encourage centralised authority and other factors that encourage decentralised "democracy" - balances between which are the major focus of group dynamics. However, decentralisation is not only a feature of human society - it is also a feature of ecology:
Another objection or limit to political decentralisation, similar in structure to that of Engels, is that terrestrial ecoregions impose a certain fiat by their natural water-circulation, soil, and plant and animal biodiversity which constitutes a form of (what the United Nations calls) "natural capital". Since these natural living systems can be neither changed nor replaced by man, some argue, an ecoregional democracy which follows their borders strictly is the only form of decentralisation of larger political units that will not lead to endless conflict, e.g. gerrymandering, in struggle between social groups.