 Implicit in our traditional design engineering paradigm is the assumption that we are dealing with a well-bounded system over which we have almost complete control and knowledge. This is true when we are developing most physical systems such as chairs, bridges, cars and so on. But part of the definition for complex systems is that the components within the system have some degree of autonomy, which means as designers, we can only have a partial influence over the system as a whole. The degree to which we can define the system will depend upon the degree of autonomy of the elements. So, in, say, a transportation network, we can have a relatively high level of control by constraining the actions of the cars. But in designing, say, a social network, people value their autonomy highly. No one is in control of the networks that are spawned out of Facebook or Twitter. They are created out of the self-organization of the users on the local level. What we are describing then is essentially a spectrum, on the one side of which we can have top-down control, allowing us to design a well-defined system that will thus be relatively orderly. On the other end, we have what we call self-organizing systems that are less designed but are instead created from the actions and interactions of their users. Thus they will likely be less orderly. Our traditional design engineering approach is of course on the left-hand side of the spectrum. It is a linear model based on the assumption that the end-user is a passive recipient or consumer. Within this industrial model, end-user variation and engagement is dumbed down so as to fit in with pre-designed procedures and systems of mass production. The obvious result of this is disengagement, alienation, and a world where end-users are constrained by systems that are created by a few designers and engineers in large centralized organizations. This is a model we should all be very familiar with. Although user-generated systems have always been there on the fringes of the mainstream, the rise of IT and the Internet has put powerful tools for self-organization and collaboration in the hands of many. Today, many of the most innovative, dynamic, and fast-growing businesses and services are harnessing this by creating platforms for technologies and people to interact, adapt, and self-organize. These are what we call co-creation platforms. Instead of the traditional divide between producer and consumer of a technology or service, innovation harnesses the relatively untapped and potentially vast resource of end-user engagement. Although finished products are sometimes what people want, it is also true that when we give people the capacity to be part of the design and production process, they feel more engaged, are more likely to value the end product, and can be a valuable source of innovation amongst other things. So how do we design these platforms of co-creation to be productive if we can't actually control what happens? If our company crowdsources the designs for the production of our next pair of sports shoes, how do we know they're going to be what we want? Well, we can't. That's part of giving over control. One person will use a social network for saving the planet, the next for avoiding work. What we can do, though, is create attractor states. That is, when we build a platform, we set desired default positions that users can change, but will be attractors for most as they are the default. To give an example of this, when building a video sharing site, if we want the site to be open and sharing, we could set the default copyrights for an uploaded video to Creative Commons. They can change it, but this is an example of an attractor state. Similarly, when we are designing an urban transport system, if we build lots of greenways, psychopaths, and pedestrian streets, people still have a choice as to what mode of transport they use, but walking and cycling increasingly become attractors in the system towards which people will naturally gravitate as it becomes the course of least resistance. Thus we can see how for every choice we have, there is a default, because it is the cheapest, nearest, or easiest. Leveraging these default positions is a powerful method of design. Building subsidies for renewable energy or open workspaces for collaboration make these desired outcome attractors. If at one end of our spectrum we have our traditional, fully controlled systems, and in the middle co-creation platforms, then at the other end we have what are called peer networks that are truly self-organizing systems, examples being Bitcoin, mesh networks, and swarm robotics. There is very little in the way of centralized platform here. Elements in the system are almost fully autonomous. Each node contributes to providing the system's infrastructure and maintaining the system. Thus they require more engagement and responsibility from each node in the network, but can result in exceptionally robust systems. In complex adaptive systems there is always a dynamic between agents and structure. That is the elements in the system and the system itself. Understanding this dynamic and the trade-off between being able to control the system versus harnessing the uncontrollable resources of the users is a key consideration. When we impose a top-down formal design on the system we will create barriers to entry and thus exclude elements either intentionally or unintentionally. They will then inevitably self-organize on the local level to create a two-tier parallel system, one formal, the other informal. Shantytowns are good examples of this. Because the requirements to enter the formal urban system were too high for the migrants, they created an informal self-organized system. This is an undesirable state that will create chronic systems integration problems. By understanding the relationship between the formal and informal, the centralized design and the distributed self-organization, we can design a multi-layered co-creation platform to integrate the two, with an integrated transition from informal low constraints and requirements to the more highly constrained formal level. Top-down formal systems can be very austere, abstract, and impersonal. They are designed to be universal one-size-fits-all. Bureaucratic systems are paradigms of this. They are designed to be impersonal and standardized for all. McDonald's is often cited as an icon of this paradigm with the exact same procedures and processes for making a Big Mac, whether in Bahrain to Santiago. But there is a good reason why many large corporations who are icons of globalization have regional offices and localized offerings. If you want to really engage people, you have to engage them on their own terms, and co-creation is one of the most effective ways to achieve this. When it came to translating Facebook into French, all they had to do was open it up for the users to translate, and passionate users had it completed within a few days. Thus, an impersonal system was adapted to local needs, and this is what co-creation is all about, creating synergies between producers and consumers, between systems and their constituent agents. Harnessing the vast resources of the end user through co-creation is one of the great sources of untapped potential that we are only just beginning to discover in post-industrial economies. It requires us to be aware of and better understand the dynamics of self-organization and just as importantly, the interplay between agents within the system, that is, the users and the structure of the system, the producer of the product, or service.