This page describes the organizational forces that limit change. It explains how to overcome them when necessary.
This page uses an example to illustrate how:
This page uses the example of HP's printer organization freeing itself from its organizational constraints to sell a printer targeted at the IBM pc user.
The constraints are described.
The techniques to overcome them are implied.
New knowledge can emerge from representations of localized perceptions
SummaryThis page discusses the interdependence of perception and representation in a complex adaptive system (
This page introduces the complex adaptive system (CAS) theory frame. The theory is positioned relative to the natural sciences. It catalogs the laws and strategies which underpin the operation of systems that are based on the interaction of emergent agents.CAS). Hofstadter and Mitchell's research with Copycat is reviewed.
John Holland's framework for representing complexity is outlined. Links to other key aspects of CAS theory discussed at the site are presented.
IntroductionCompetition in a
This page introduces the complex adaptive system (CAS) theory frame. The theory is positioned relative to the natural sciences. It catalogs the laws and strategies which underpin the operation of systems that are based on the interaction of emergent agents.complex adaptive system (CAS) can take many forms. But each of the forms
John Holland's framework for representing complexity is outlined. Links to other key aspects of CAS theory discussed at the site are presented.
This page discusses the mechanisms and effects of emergence underpinning any complex adaptive system (CAS). Key research is reviewed.emerges from the actions of an
This page discusses the mechanisms and effects of emergence underpinning any complex adaptive system (CAS). Key research is reviewed.emergent
Plans are interpreted and implemented by agents. This page discusses the properties of agents in a complex adaptive system (CAS).agent. One major challenge is to be able to respond to a limited number of
It then presents examples of agents in different CAS. The examples include a computer program where modeling and actions are performed by software agents. These software agents are aggregates.
The participation of agents in flows is introduced and some implications of this are outlined.
Agents use sensors to detect events in their environment. This page reviews how these events become signals associated with beneficial responses in a complex adaptive system (CAS). CAS signals emerge from the Darwinian information model. Signals can indicate decision summaries and level of uncertainty.signals
To benefit from shifts in the environment agents must be flexible. Being sensitive to environmental signals agents who adjust strategic priorities can constrain their competitors.flexibly and effectively by correctly identifying the appropriate situation from a potentially wide variety.
Agents are sensitive to particular signals. The detection of a specific signal results from a sensitive agent interacting with the signal. Some agents perform
The agents in complex adaptive systems (CAS) must model their environment to respond effectively to it. Samuel modeling is described as an approach.modeling of the signal. The model could be as simple as checking a value of the signal and responding with one of two outputs.
While some agents aggregate to better cope with the signals they must act on, others remain focused on some simple aspect of the overall situation, relying on prior selection pressure to improve the likelihood that the situation they depend on recurs frequently. Aggregation is when a number of actions become coordinated and operate together. In the adaptive web framework's Smiley, codelets become coordinated by their relative position in the deployment cascade. The cascade's dynamics are dependent on the situation, the operating codelets responses to that situation and the grouping of schematic strings they are associated with. The aggregate affect is a phenotype the adaptive agent.
allows for parallel processing of complex, rapidly changing inputs exemplified by visual processing is the main part of the cerebral cortex in mammals. It was originally thought to exist only in mammals but is also present in reptiles and birds buried behind other areas of the for-brain. The for-brain develops based on a genetic plan consistent across all vertebrates. The neocortex processes vision in the visual hierarchy V1, V2, V3 .. V5 ... V20; and language with areas including Wernicke's and Broca's with sensors in the inner ear. Primate species with bigger social groups have larger cortices. Human cortex size suggests traditional human cultures had an average size of 150 people.
. Aggregation enables the complex changing state of the environment to be represented within the
This page discusses the effect of the network on the agents participating in a complex adaptive system (CAS). Small world and scale free networks are considered.network of components of the aggregate. However, it requires organization - self-organization.
Douglas Hofstadter writes that the perceptual process 'is essentially one of constructing larger units out of smaller ones, with temporary structures at various levels and permanent mental categories trying to accommodate each other. ' .... 'In any type of perception, much back-and-forth motion must occur -- that is, an intimate mixture of construction, destruction, regrouping, and rearrangement of tentative structures. Any architecture for a system to carry out this type of process is the result of many subtle decisions about how independent processes should interact, how structures should be put together or broken apart, what kinds of things form stable structures, what easy ways are of making new possible structures when old ones are seen to be inadequate, and so on.' Hofstadter outlined such an architecture which he called Copycat.
Fluid Concepts and Creative Analogies. What follows is a brief overview of the major components extracted from their detailed discussion.
They write 'there are three major components to the architecture: the Slipnet, the Workspace, and the Coderack. In very quick strokes, they can be described as follows. (1) The Slipnet is the site of all permanent Platonic concepts. It can be thought of, roughly, as Copycat's long-term memory. As such, it contains only concept types, and no instances of them. The distances between concepts in the Slipnet can change over the course of a run, and it is these distances that determine, at any given moment, what slippages are likely and unlikely. (2) The Workspace is the locus of perceptual activity. As such, it contains instances of various concepts from the Slipnet, combined into temporary perceptual structures (e.g., raw letters, descriptions, bonds, groups, and bridges). It can be thought of, roughly, as Copycat's short-term memory or working memory, and resembles the global "blackboard" data-structure of Hearsay II. (3) Finally, the Coderack can be thought of as a "stochastic waiting room", in which small agents that wish to carry out tasks in the Workspace wait to be called. It has no close counterpart in other architectures, but one can liken it somewhat to an agenda (a queue containing tasks to be executed in a specific order). The critical difference is that agents are selected stochastically from the Coderack, rather than in a determinate order'.
This page introduces the programs that the Adaptive Web Framework (AWF) develops and uses to deploy Rob's Strategy Studio (RSS).Perl frame includes a Copycat like framework (
The programs are structured to obey complex adaptive system (CAS) principles. That allows AWF to experiment and examine the effects.
A production program generates the web pages.
A testing system tests the production program. It uses a framework to support the test programs. This is AWF's agent programming framework as described in the agent-based programming presentation.
An example of the other AWF agent-based programs that are also described in the frame is the virtual robot.
Finally a strength, weaknesses, opportunities and threats assessment is presented.
This page describes the Adaptive Web framework (AWF) test system and the agent programming framework (Smiley) that supports its operation.Smiley) for integrating perceptions and representations. It includes implementations of the three inter-dependent aspects of Copycat: the
Example test system statements are included. To begin a test a test statement is loaded into Smiley while Smiley executes on the Perl interpreter.
Part of Smiley's Perl code focused on setting up the infrastructure is included bellow.
The setup includes:
This page describes the Copycat Coderack.Coderack,
The details of the codelet architecture are described.
The specialized use of the Coderack by the adaptive web framework's (AWF) Smiley is discussed.
The codelet scheduling mechanism is discussed.
A variety of Smiley extensions to the Coderack are reviewed.
The Coderack infrastructure functions are included.
This page describes the Copycat Workspace.Workspace and
The specialized use of the Workspace by the adaptive web framework's (AWF) Smiley is discussed.
How text and XML are imported into the Smiley Workspace is described.
Telomeric aging of schematic structures is introduced.
The internal data structure used to represent the state of each workspace object is included.
The Workspace infrastructure functions are included.
This page describes the Copycat Slipnet.Slipnet; which support the emergence of codelet based CAS agents.
The goal of the Slipnet is reviewed.
Smiley's specialized use of the Slipnet is introduced.
The initial Slipnet network used by the 'Merge Streams' and 'Virtual Robot' agent-based applications is setup in initchemistry and is included.
The Slipnet infrastructure and initialization functions are included.
The output of the modeling is reflected in other signals issued by the modeling codelet. Other codelets may respond to the initial signals, or the output of the models or both, initiating some action. Due to the modeling a variety of potential actions becomes associated with the original signal and current local environmental state. The different actions correspond to differences in the benefit generated.
The codelets can be
This page describes the adaptive web framework (AWF) Smiley agent progamming infrastructure's codelet based Copycat grouping operation.associated with schematic structures in the Workspace. Since codelet's actions vary depending on the state of the local Workspace particular associated representations become influences of the codelet's actions. With
The requirements needed for a group to complete are described.
The association of group completion with a Slipnet defined operon is described. Either actions or signals result from the association.
How a generated signal is transported to the nucleus of the cell and matched with an operon is described.
A match with an operon can result in deployment of a schematic string to the original Workspace. But eventually the deployed string will be destroyed.
Smiley infrastructure amplification of the group completion operation is introduced. This includes facilities to inhibit crowding out of offspring.
A test file awfart04 is included.
The group codelet and supporting functions are included.
Plans change in complex adaptive systems (CAS) due to the action of genetic operations such as mutation, splitting and recombination. The nature of the operations is described.reproduction of the Workspace structures selecting for valued actions, the Workspace structures become
Plans emerge in complex adaptive systems (CAS) to provide the instructions that agents use to perform actions. The component architecture and structure of the plans is reviewed.memetic plans. A higher level modeling system appears based on the interactions, and a true CAS agent emerges from the interplay of Slipnet, Workspace structures and codelets.
Agents select different actions due to differences in the local environment and the chance encountering of signals. Certain of the agent's selections may be relatively advantageous.
Plans emerge in complex adaptive systems (CAS) to provide the instructions that agents use to perform actions. The component architecture and structure of the plans is reviewed.Schematic structures are essentially sequences of signals which the agents detect and respond to. The inclusion of these schemata both expands the local environment of the agents and shapes and constrains it.
The alteration of schematic structures to record the results of the models and actions allows a direct integration of the agent's plan, its actions and its state. The changes to the schemata can be targeted to particular labeled regions. Such changes can affect the schematic signals the agent detects and responds too. CAS agents naturally utilize
Walter Shewhart's iterative development process is found in many complex adaptive systems (CAS). The mechanism is reviewed and its value in coping with random events is explained.Shewhart cycles.
Schematic strings are also easy to copy, and can be subject to
Plans change in complex adaptive systems (CAS) due to the action of genetic operations such as mutation, splitting and recombination. The nature of the operations is described.genetic operators. Alternative strategies correspond to alternative sets of schematic strings. The resulting
This page reviews the implications of selection, variation and heredity in a complex adaptive system (CAS). The mechanism and its emergence are discussed.evolutionary pressures will alter the mix of
This page reviews the implications of reproduction initially generating a single child cell. The mechanism and resulting strategic options are discussed.replicators available for development is a phase during the operation of a CAS agent. It allows for schematic strategies to be iteratively blended with environmental signals to solve the logistical issues of migrating newly built and transformed sub-agents. That is needed to achieve the adult configuration of the agent and optimize it for the proximate environment. Smiley includes examples of the developmental phase agents required in an emergent CAS. In situations where parents invest in the growth and memetic learning of their offspring the schematic grab bag can support optimizations to develop models, structures and actions to construct an adept adult. In humans, adolescence leverages neural plasticity, elder sibling advice and adult coaching to help prepare the deploying neuronal network and body to successfully compete.
and implementation of the future plans specifying the models and actions that drive the agents.
Schematic agents can aggregate into perception and representation structuresMore complex signals, such as those captured by the primate visual system supports processing of visual data into what and how. To do this it has two distinct paths: The ventral path and the dorsal path.
, have highly parallel aspects which are initially captured by
This page describes a schematic system about abstracted neurons operating in a circuit.arrays of photo-receptors and then neuronal state. Having layers of interconnected neurons, specialized eukaryotic cells include channels which control flows of sodium and potassium ions across the massively extended cell membrane supporting an electro-chemical wave which is then converted into an outgoing chemical signal transmission from synapses which target nearby neuron or muscle cell receptors. Neurons are supported by glial cells. Neurons include a:
The neuronal system was designed to focus in on the cellular nature of a schematically defined neuron.
The goals include:
The codelets and infrastructure are included.
Representing state in emergent entities is essential but difficult. Various structures are used to enhance the rate and scope of state transitions. Examples are discussed.represent shared state, and issue signals. As Hofstadter described above multiple structures, each represented by inter neuron, specialized eukaryotic cells include channels which control flows of sodium and potassium ions across the massively extended cell membrane supporting an electro-chemical wave which is then converted into an outgoing chemical signal transmission from synapses which target nearby neuron or muscle cell receptors. Neurons are supported by glial cells. Neurons include a:
growth, are built in parallel by neuron agents adapting to the stream of signals from the sensors and each other. Schematic plans specify the pre-deployment of valuable neuron network agents and associated body cells.
Bodies, capable of executing actions for neuron network agents, demand significant supplies of resources. Emergent infrastructure supports the deployment and operation of the neuron network and body cell agents. Additional leverage from the emergent formation of
This page reviews the catalytic impact of infrastructure on the expression of phenotypic effects by an agent. The infrastructure reduces the cost the agent must pay to perform the selected action. The catalysis is enhanced by positive returns.infrastructure and
This page reviews the strategy of setting up an arms race. At its core this strategy depends on being able to alter, or take advantage of an alteration in, the genome or equivalent. The situation is illustrated with examples from biology, high tech and politics.evolved amplifiers enhances the operational benefits of the system, but requires regulation to limit positive returns, W Brian Arthur's conception of how high tech products have positive economic feedback as they deploy. Classical products such as foods have negative returns to scale since they take increasing amounts of land, and distribution infrastructure to support getting them to market. High tech products typically become easier to produce or gain from network effects of being connected together overcoming the negative effects of scale. effects of the amplifiers.
Competitive parasitic is a long term relationship between the parasite and its host where the resources of the host are utilized by the parasite without reciprocity. Often parasites include schematic adaptations allowing the parasite to use the hosts modeling and control systems to divert resources to them. strategies can subsequently target the model and regulatory environment to leverage the infrastructure and control systems while avoiding the costs involved in developing and operating a perception and representation infrastructure.