|
p.2 the complex systems approach has done away with the old philosophy of dualism, which sees the world as made out of two distinct substances: matter, as described by the natural sciences, and mind, as described by the social sciences and humanities.
p.2-3 The components of a complex system are commonly called agents... Usually, agents are assumed to be goal-directed... When no explicit goal can be distinguished, their activity still follows a simple cause-and-effect or condition-action logic: an agent will react to a specific condition perceived in the environment (cause) by producing an appropriate action (effect). However, this causal perspective is essentially equivalent to the intentional perspective (which Dennett [1989] calls the intentional stance), because irreversible actions eventually lead to a so-called "attractor" of the agent's dynamics, and an attractor behaves indistinguishably from a goal or intention... causal (material) and intentional (mental) models are essentially equivalent - even though the one may be more easily applicable in a certain context than the other.
p.3 Processes in complex systems... are usually non-linear: their effects are not proportional to their causes... Interactions with positive feedback are very sensitive to their initial conditions: a change in that condition may be so small that it is intrinsically undetectable, yet results in a drastically altered outcome.
p.3 simple interactions at the local level give rise to complex patterns at the global level. This phenomenon is called emergence.
p.4 the concept of self-organization has diffused into virtually all scientific disciplines, as an explanation for previously mysterious phenomena in which complex structures arise from the interaction between simpler components.
p.4-5 In the following, I will try to formalize a general conceptual foundation for the study of self-organization, and apply this to the emergence of collective intelligence in groups.
p.5 Organization can be defined as structure with function: the components (agents) of the system are arranged in an orderly way (structure) so as to achieve a certain goal (function)... The relation between structure and function becomes clearer when we introduce the notion of coordination [Crowston et al., 2006]: what counts is not so much how individual agents are arranged (e.g. in some kind of hierarchy or network), but how their actions work together in a harmonic way towards their collective goals. At the very least, these actions should not hinder, obstruct, or oppose each other. This is what I have called the avoidance of friction [Heylighen, 2007b, 2008, 2011]. At best, they will smoothly complement each other... they can solve problems together that they cannot solve individually. This... may be called synergy [Corning, 1998; Heylighen, 2007, 2008]. Coordination can then be defined as : the structuring of actions in time and (social) space so as to minimize friction and maximize synergy between these actions.
p.5 Coordination can be subdivided in four elementary processes or mechanisms: alignment, division of labor, workflow, and aggregation, which I will now discuss in turn.
p.5-6 Alignment is in general easy to achieve by self-organization... With a little bit of trial-and-error they [two individuals simultaneously pushing against a heavy object to get it out of the way] may further discover that by pushing in one precise direction, their push is not only not hindered, but actually fully reinforced by the other one who is pushing in the same direction. Once they discover this shared direction, their actions are fully aligned, and their effort is maximally productive.
p.6 the evolutionary mechanism of the blind-variation-and-natural-selection of actions will sooner or later produce an interaction that is more synergetic and less frictional. This mechanism does not require any planning, knowledge, or intelligence on the part of the agents. The only assumption is that the agents obey a logic of trial-and-error or variation-and-selection, producing a variety of actions until they find one that is maximally productive and stick to that one, irrespective of what causes that increase in productivity.
p.6 The more are already aligned, the larger the force in the direction of their alignment, the more difficult it will be for others to oppose that movement, but the easier it will be for them to join in with that movement. Therefore, trial-and-error on the part of those others will settle more quickly on the direction of the initial alignment. This is a process of positive feedback: the more alignment there already is, the faster others will join it. As a result, all agents will quickly align, making the collective homogeneous in its direction of action... If the agents are spread across an extended region of space, it may take a while for the alignment to propagate across that space, as more remote agents will initially not sense that in another region an alignment has started to form.
p.8 in a market economy... the "calculation problem" of establishing exactly how many goods of which type need to be delivered by whom is much too complex to be solved by any planner. Only self-organization can produce robust solutions to problems of such complexity and variability.
p.9 As long as a large enough number of agents with sufficiently diverse or broad skills is available, such a self-organizing solution to the problem of workflow can be quite efficient. There is no need to plan... as long as enough agents are available so that a sufficiently skilled one is ready to take over soon after the previous task is finished.
p.9 Like division of labor, aggregation is a parallel process: different streams of activity come together simultaneously.
p.12 "groupthink" [Janis, 1972; for a review, see: Esser, 1998] ...is the phenomenon where people in a group all start to think the same, because a slight initial preference for one approach rather than another becomes amplified via positive feedback. The underlying dynamic is that if someone hears an opinion expressed by someone else, s/he will be more inclined to express a similar opinion... Groupthink is an example of self-organization gone wrong, where non-linear interaction has led to premature alignment on a suboptimal solution, and where the positive contributions of diversity and division of labor have been neglected... The simplest way to avoid groupthink is to disallow direct communication between the group members, so that the one cannot influence the other one until everyone has been able to make a full contribution.
p.14 We have defined alignment as the unification or merging of the targets of different actions or agents. As long as there is no "director" agent to impose a target on the others, this problem can only be solved through self-organization, that is, spontaneous, reciprocal adaptation of the agents' targets.
p.18 As a member of the editorial board of the Journal of Happiness Studies [JLJ - ok, so you submit an article to this editorial board for consideration for publication, and it gets rejected. How exactly does this make you feel? Seems like the Journal of Happiness Studies is producing a lot of unhappy people.]
p.22 This paper has reviewed the mechanism of self-organization, conceived as the spontaneous coordination of actions performed by different agents. Such coordination helps to make the actions more synergetic, while reducing the friction between them. The result is that coordinated actions achieve their intended goals more easily and more effectively... The underlying dynamic of self-organization is local trial-and-error or variation-and-selection, in which two interacting agents try to mutually adapt their actions, until they hit on a "coordinated" pattern that is acceptable to both, and thus is selectively retained.
|
