Albert-L‡szl— Barab‡si (2002), Linked: The New Science of Networks, Cambridge, Mass.: Perseus.

Scientists have tended to work on the reductionist assumption that by taking something apart we will gain an understanding of how it works from its constituent parts. However, like humpty-dumpty, from knowledge of the pieces it does not necessarily follow that we understand how to put the pieces together or they operate together as a complex system. Nature runs as a self-organising system and only reveals its secrets in part through its parts. It is a network that operates under the laws of complexity - what Barab‡si calls the 'science of networks' (2002). He is drawing partly upon the popular idea of 'six degrees of separation' to describe interconnectivity, a principle introduced by Stanley Milgram in 1967, interestingly developed to understand social networks and the distance between any two people in the United States. Extended to the whole world by John Guare, the suggestion is that in a network of six billion nodes, any pair of nodes are on average only six links from each other (Barab‡si, 2002: 27-30). 

This network is society, but society is stratified. Despite the appearance of randomness in complex systems, clearly there is underlying order - all nodes are not equal by any means. Despite this, there has been a tendency to think of networks as equitable systems, viewed as fundamentally random simply because they are too complex to comprehend how power is distributed (although not discussing power at all, Barab‡si cites the work of Erd—s and RŽnji on the connections between randomness and complexity, 2002: 24). 

Information, however, is generally taken to be value-free. The World-Wide Web was clearly developed with this in mind (nineteen degrees of separation according to Barab‡si, 2002: 34) and with a different articulation of power in mind. Various browsers have also attempted to draw attention to the linking rather than the contents contained within web pages, showing an architecture and a cartography of information wherein distance is reconfigured but not necessarily shrunk as Barab‡si would claim into 'small worlds' (2002: 41; although he later claims there is no intrinsic scale, that networks are scale-free, 2002: 70. note: add some dialectical ideas on distance here). Connectedness and power are articulated through the density of linkages within the system despite the enormous size and apparent complexity of the system. This is deceptively simple as any one node connects to an exponential number of others through this logic (think of a search engine like google and its 'I feel lucky' link; in April 2004, if 'weapons of mass destruction' were sought using this facility a long ironic treatise would result - a story spread around communities on the internet but missed by the authorities that police google. This demonstrates 'value-added' rather than 'value-free' information). 

In social systems, nodes gather together in clusters but this extends to the ways in which other systems operate and is seen to be a generic property of complex networks (Barab‡si citing the work of Watts and Strogatz, 2002: 51) - a distinctly non-random behaviour. 'Connectors,' nodes with an large number of links or connections, are present within diverse systems (from society, to the cell - or cellular system that includes genes, proteins and other molecules) and further account for the redundancy of randomness and points to the lack of democracy within the system (and something that is crucial to an understanding of search engines for instance). The World Wide Web is dominated by highly connected nodes or 'hubs' and thus can be seen to not be an egalitarian space - an extension of the public sphere (a search engine would be a good example of a hub). 
[link to ideas of the public sphere here?] 

Hubs express power not centrally but in a decentralised manner in keeping with contemporary descriptions of power (in Hardt and Negri's Empire for instance) but whether they follow universal rules of nature is clearly debatable. Barab‡si quotes the work of the Italian economist, Vilfredo Pareto, who observed universal laws in that 80 per cent of peas were produced by 20 per cent of pea pods, extended to property in that 80 per cent was seen to be owned by 20 per cent of the population (2002: 66). Although this '80/20 rule' seems rather reductive to say the least, it does offer some truth, perhaps not to universal laws of nature but, to the politics of networked systems. There is such a mathematical expression called 'power law' that stands in opposition to the orthodox bell curve as a means of understanding quantities in nature. A power law does not have a peak and expressed as a histogram would be a continuous decreasing curve, wherein many small events coexist with few large events in keeping with the existence of large numbers of nodes but few hubs (Barab‡si, 2002: 67-70). The mathematical take on power is that few events manifest most of the action. In complexity theory, this is the emergent order within disorder. Clearly, there are similar 'laws behind complex networks' (Barab‡si paraphrasing Pareto, 2002: 73). 

A more pragmatic demonstration of power at work is that water is the most common substance on Earth and yet still some people have difficulty in gaining access to enough to sustain their health. Liquids strike a delicate balance between states - the forces that keep the molecules together are not strong enough to solidify them. In a sense, liquid is trapped between order and disorder, what Barab‡si calls a 'majestic dance' (2002: 73) the water molecules come together, form small groups, and then break apart to form other groups. Ice does alter this process into an ordered solid state - a 'phase transition' from disorder to order. Similar transitions in response to changes in energy take place when liquids like water are turned into gas once heated. The revolutionary transition of 'all that is [ordered] melts into [disorder]' is evoked. The properties of liquids is a complex phenomena demonstrating a self-organising 'BŽnard cell pattern' where a 'series of bifurcations describe greater and greater complexity of spatial pattern that is the precise spatial of the "march to chaos" via period doublings in the logistic equation, ending in turbulence as the liquid analogue of chaos. Here liquids obey the deterministic chaos of a strange attractor...' (SolŽ & Goodwin, 2000: 16-7). Perhaps another example is required of power laws in relation to fractal patterns. SolŽ & Goodwin describe the sandpile experiment: a grain of sand is added one by one; at first, the grains are seemingly independent of each other following the laws of gravity and friction; however, eventually a point of 'self-organised criticality' (coined by Bak) is reached and the pile collapses, first a few grains and eventually large amounts. A new collective behaviour or 'phase transition' has emerged that has no relation to the individual behaviour (2000: 54). This principle was modelled by Per Bak using 'cellular automaton' following the same basic rules thus demonstrating power laws and fractal behaviour. 

Note: In a similar way, the mathematician John Conway invented his 'Game of Life' in which each lightbulb or pixel on a compute screen is treated as a simple machine. It is able to sense whether its immediate neighbours are on or off, and executes simple rules. At regular intervals, each pixel examines the state of its neighbours and switches on or off accordingly. The process is repeated and produces startling effects, sometimes flickering briefly and other times continuing as a stable pattern. Steve grand describes his amazement at seeing a simple version of this called 'glider' where the pixels give the impression of movement from an emergent behaviour (2001: 47).    

The issue for Barab‡si is that these phase transitions from disorder to order demonstrate a consistency even within different systems and follow power laws. Power laws are 'patent signatures of self-organization in complex systems' (Barab‡si, 2002: 77). Clearly a politics is required, suggesting that the power laws themselves require transition. Interestingly, 'transition' in the Marxist lexicon, describes the transition in terms of a socialisation of property, a replacement of the relations of the market with relations of the organisation of social property, of social work (Negri 1991; 211; Negri would go further still and say that property and the law of work-value itself should be challenged). 

Barab‡si worryingly characterises the power law as the 'rich get richer' in his analysis of the evolution of networks in nature (2002: 80). They grow exponentially but this is not random but a highly selective process, demonstrating 'preferential attachment' that together with 'growth' generates a free-scale network (2002: 86) and conception of 'fitness'. No matter the size of the network, it will maintain this hub-dominated distribution of power in neo-Darwinian terms of without intervention. The intervention on a theoretical level, through quantum mechanics, is a better understanding of complexity itself as a critique of the deterministic logic of Darwinian evolutionary theory. Barab‡si explains that every network has its fitness distribution, more or less egalitarian, into two categories: 'winner-takes-all' and 'fit-get-rich' competitive networks. These are distinguishable in his terms, but might as well be described as totalitarian or neo-liberal to my mind - so a rather limited choice. The example of the success of Microsoft is mentioned in this connection. Installing Windows adds a link to Microsoft, binding together the operating system and the hardware as a very 'fit' market strategy (2002: 104). Barab‡si says: 'In the most complex networks, the power laws and the fight for links thus are not antagonistic but can exist peacefully' (2002: 103). But what other networks might emerge? Can we begin to imagine antagonist networks, with nodes that aim to contradict others, where weaker nodes combine forces to become a hub and then gain collective power. Viruses also clearly demonstrate these principles both in terms of biology and computing. In 2000, the 'I Love You' virus was a case in point: opening the message 'love letter for you' would erase documents from your hard drive and then propagate itself by sending new copies of itself through the address book of your mail program. 
On a biological level, parasites both contribute to the maintenance and generation of biodiversity in ecosystems. Often they act upon eachother as well as on their hosts. The metaphors abound when politics is added to the mix.    

The example of Linux, the free operating system, suggests some hope. More specifically, Armin Medosch points to the recent initiative exploiting the free 2.4 gigahertz frequency spectrum and 802.11 Wi-Fi technology to provide free networks. People are encouraged to make their own network nodes and to patch them together in a mesh peer-to-peer network making its own shared infrastructure (2003: 18). Thus the bandwidth legally owned is shared amongst a local collective of users. This example of collective action in the spirit of the idea of the public realm has been successfully implemented in many big cities, such as across East London (see http://www.consume.net/). 

Relative weakness, even in nature, is supported by interconnectivity to the system as a whole - and therefore the characterisation of the system is paramount. The historical development of the Internet is a case in point - wherein attack on any node is compensated by the linking structure of the system as a whole. Scale-free networks are relatively robust as a result of their topology, even when under attack by 'crackers' (hackers with malicious intent). But complex systems display elements of vulnerability too if attention is paid to the hubs rather than nodes. Thus any budding terrorist would do well to target the hubs; I could tentatively suggest that this is a class distinction (in targeting the bourgeois hubs rather than the proletarian nodes within the system. Also, interestingly, terrorist groups have known network theory intuitively organised in self-organised flexible networks, able to reorganise when required. In the case of Al Qaeda, Bin Laden remains a significant hub). In a similar way 'cascading failures' are well known explanations of this tendency, where a local failure redistributes responsibilities to linked nodes; this failure is cascaded through the system sometimes to disastrous effect depending on how central the role of the node is within the system as a whole. This can be demonstrated with a wide variety of systems: failed routers within the Internet, or species within an eco-system, or in economics with the collapse of a certain company (Barab‡si, 2002: 120). This is clearly of use to those wanting the system to become more robust or efficient or those wishing to bring about its crisis. Capitalism, Al Qaeda, Microsoft, AIDS, and other complex self-organising networks, are not value-free.   

In describing an optimal system, Paul Baran famously characterised three architectures: centralised, decentralised and distributed. Baran was designing a robust communications infrastructure that could withstand attack (from Soviet nuclear strike - this is 1964; in Barab‡si, 2002: 145). Both the centralised and decentralised model were too vulnerable and the distributed 'mesh' architecture was proposed but not used until much later. This underpins the development of the Internet and charts a developmental evolution that takes account of a growing understanding of complexity, leading to vast scale-free topology of routers and links, like an ecosystem - demonstrating growth and preferential attachment in Barab‡si's terms - and later: 'A scale-free network is a web without a spider' (2002: 221; the absence of the spider is what makes it self-organising). According to Lawrence Lessig, in Code and Other Laws of Cyberspace, (1999) code should be regulated to limit the increasing rule of the 'natural forces' of the market (spiders catching flies). Barab‡si would suggest that code is separate from collective human action and only the first can be regulated and stating that 'Regulations come and go, but the topology and the fundamental natural laws governing it are time invariant' (2002: 175). Clearly there is little sympathy for a more detailed discussion of ideology or the formation of machinic subjectivities here. There's clearly more critical work to be done in this area of network theory (not least in its application to economics) that requires a politics without losing sight of how the network works on a technical level - not simply doing either, but both.