Chapter XVII

Automated Systems

Automated systems can be said to be a general class of artificial or human made system, or any other alternative system that functions according to certain laws and theory of automata, which can be most generally defined as an device that is self regulated and relatively independent in its control functions. In a more theoretical vain, we refer to any device into which input may be affected internally by certain state transitions, leading to some form of predictable or logical output. By inference, at the center of automatons, especially very advanced systems, are computer-based systems. A completely independent automata would be akin to a very sophisticated, anthropomorphic robot that was capable of functioning in every manner as a normal human being in the world. This analogue of automata is a very anthropocentric one. It is in my opinion that this anthropocentrism of our conception of automata is more of a hindrance than a critical insight into their potential development. Vast, super-sophisticated systems that are completely distributed upon a number of levels, and achieve maximum integration, appear to me to be more interesting kinds of models to pursue.

At some point, an integrated system can be said to be one in which all automatic functions occur within the same machine framework, or device modulator. This may in fact be a relative issue of a device is nothing more than a plastic case disguising a great deal of interconnected wiring and multiple modular components of such systems. Miniaturization of circuitry has led the revolution of digital information processing to the achievement of vastly superior integration of machine automata. But this challenge is met with a corresponding problem of the transference of information from one module or system to another, along with the challenge of creating a universal interface by which all systems can achieve effective talk. With distributed systems, the greater the distribution, the greater the function of communication and problem of transmission of information becomes a problem over the question of the integration of function of information processing on a reducible scale.

There occurs invariably a kind of trade-off between systems between the integrative miniaturization and distributive communication between systems. We cannot have a perfectly integrated system without some measure of distribution, and we cannot have a fully distributed system that achieves complete communicative efficacy. In other words, distribution must be gained at the expense of possible integration into a single system, and integration into a single system must come at the cost of distribution of such a system to a larger range of possibilities. A system can be powerful and highly integrated, but its potential is severely circumscribed if it functions by itself and is not able to share its information within a larger network.

I would add to the challenge of developing sophisticated automata a few other related challenges:

1. The challenge of a machine system being capable of producing or acquiring its own energy independently of any other system.

2. The challenge of a machine system being capable of performing other kinds of work or productive/purposive activity beyond its own informational processing, implying manual articulation of such a system in the environment and regulation of function in the environment.

3. The challenge of a machine system being capable of both reproducing and repairing itself, both as hardware and as software.

4. The challenge of a machine system to learn and develop as a system, and to integrate changes into itself.

5. The challenge of a machine system to evolve as an integrated population of modular components, such that we can speak potentially of multiple successive generations of such systems.

What is described in these goals is a completely automated metasystem that would require as few human influences or inputs, beyond the original design and construction, as would be necessary for its overall function. Such automation would be achieved gradually by small steps, one step at a time, rather than all at once. It is the case that most metasystems are not achieved overnight, but are complex solutions arrived at over many cycles of trial and error.

A number of other ancillary functions can be associated with these as well. We can include for instance the capacity of different modular components of distributed systems to share and interact and coordinate functions to problem solving in an independent manner, and the ability of such modular components to effectively recognize themselves and one another in a way that allows them to make some kinds of decisions regarding interaction.

Implied in these kinds of challenges would be a certain openness of design of such systems, as well as a certain basic plasticity or flexibility of structure that would enable such designs to be modified and adapted to the widest range of functions and circumstances that are possible for these systems to achieve their fullest distribution and coordination. A system therefore cannot be overdetermined or even fully determined, but must be complexly chaotic in structure and therefore remain only partially determined and relationally connected to the world in which it is situated.

The suggestion of a fully connected and distributed system implied above is that of an informationally based automated infrastructure that coordinates and produces energy for work of various forms, and even provides the working robotics that would be the basis of such production. In other words, I am speaking of a form of systems integration that combines the different aspects of basic social infrastructure into a single coordinated system of automation. In this, would could not clearly draw the line where one form of processing or integration leaves off and another begins. Communication would be a form of processing, and processing a form of communication that is extended and elaborated.

Integration is implicit to the concept of sophisticated automation. The more integrated a system becomes across multiple alternative functions, the greater can be said to be its degree of automation. Automation itself though remains a relative measure, short of an absolute standard. We must even inquire, for instance, how fully autonomous we are ourselves as human creatures, or if possibly some of our sense of autonomy might but be an illusion of our own false hubris disguising the fact that we are perhaps like marionettes to certain complex forces beyond our control or ability to manipulate. We are not always in as much self-control as we might believe or want to believe ourselves to be.

However automated a machine-based system might become, it must be kept in mind that this automation remains essentially mechanistic and in a real sense blind and "dumb." In other words, automation of function of operation of such machines is not the same thing as the simulation or especially the realization of "true intelligence" as this is at least anthropomorphically defined. I doubt whether machines can ever become capable of true self-awareness as if an independent living being, no matter how sophisticated or "intelligent" their design may become.

In a sense, such an emergent supersystem would be a logical outcome of the development of natural metasystems, and it would represent a new level of integration of such systems that can be characterized by its own uniquely emergent properties. These emergent properties would entail perhaps macro-level social phenomena and reorganization of human systems in a manner unprecedented in traditional human social organization. It would not be that humans become puppets of these new automated systems. Rather this becomes increasingly their function and their work, to service and maintain this metasystem, as it becomes in a way a symbolic and metabiotic aspect of human social organization, muct has cells of a multi-cellular organism become functionally differentiated and contextualized within various tissue matrices of the organism. Instead of like single cell organisms that we are now like, we become organized socially and functionally into a multi-cellular kind of body. The kinds of state systems we have now are in a sense precursors of this, but they are more like guild-colonies formed by unicellular organisms than they are like a truly integrated organic metasystem.

I believe this stage will be reached, if it is reached at all, when truly autonomous and distributed systems are developed on a global basis, and as a result human social systems become so functionally and structurally and symbolically interdependent that the traditional or conventional boundaries separating people will tend to break down and give way to a complete new system of social organization. I believe it will entail the gradual development of a "metacultural" orientation as well, one that is focused around the manipulation of complex information. This metacultural orientation will not displace traditional cultures all at once, but merely come to share space and overlay these other orientations to the point that there is a global streamlining and a disappearance of cleavages and vast differences between different peoples. Language boundaries may in time come to make little difference if a common web system exists that can automatically translate inputs from any known language to reasonably reliable outputs in any other language. Then people can achieve communication in a virtually instantaneous manner on a global scale.

The alternative model is a superorganic form of artificial life, that, though it might be artificial, represents a form of living system that meets certain minimal criteria of such systems. It will serve its own purposes and will have its own goals in its state-path trajectory. It may be the case that human beings come to need such a system, if they are the completely escape the contradictions of their own predicament on earth, whatever tyrannical implications it may seem to have. We should not see such a possible system as a single conscious entity, a kind of Hal from Space Odyssey. It is something more than this--it would be a billion conscious entities both focused at one point in place and time, and simultaneously spread out around the entire globe. We would not fear its over-control, ans there would be no centralized control function. Control, inherent to its design, would be as distributed as the information upon which it is based.

I see this relationship as being ultimately totally symbiotic and mutualistic between human beings and their metamachine. In other words, the metamachine would serve humans as much as it is served by humans, and would provide virtually every aspect of human needs and interests that can be practically defined. At the same time, its own functions would be non-human in nature, except to the extent that these were defined by humans originally to serve human interests in some exclusive manner.

I believe that because we can never classify such systems as truly intelligent, however integrated they may be, we do not ultimately need to fear losing ultimate control over their function and destiny. At some point, human beings would retain the ability to switch such a metamachine on and of at its own will, or at least we might be well advised in the future to retain this kind of final control. This consideration leads to the following proposition:

Sense of machine autonomy can never be extended beyond the bounds of human control factors and human-based design, therefore autonomous systems can never achieve what can be called as ultimately arbitrary autonomy. They remain autonomous in two senses therefore, because they are relative independent of human influence, and because they are fully "automatic" as a mechanistic system.

A fundamental, single language for computers, that accomplished a universal interface and that permitted both evaluative and performative functions of machine processing, would go a long way to achieving, I believe, both communication and integration. We have a bewildering variety of language systems today that are like a tower of babel, or rather really an electronically multi-lingual system. No single language system seems suitable for all purposes. We have HTML markup code for internet communication and AI languages for building complex puzzle solving solutions.

It is important to inquire into the extensive and intensive limitations of any particular computer programming language to fulfill the objectives of a fully integrated system. Not only must such a system accomplish multiple processing and communication functions across different computer architectures, but it must be capable of interfacing with human language in the most facile and natural manner possible. The lack of a single integrated language system has been filled to some extent by major computer software companies who offer integrated programs at a higher level of code, hiding the source code of their programs from open access for proprietary purposes. Thus, certain large companies have come to monopolize this aspect of computing to an exclusive and wonderfully profitable advantage in the world.

What is intelligence, and what seems possibly intelligent about an integrated grid or matrix system?

Intelligence can be said to be a form of active awareness, awareness that sees beyond surface pattern to apprehend the underlying pattern of order through time and across space. Intelligence also implies the capacity to achieve appropriate response that meshes with larger intentional structures. It therefore implies a form of applied rationalism that includes the assessment and derivation of long-term goals themselves.

A complex grid structure, or matrix or array, can be said to constitute a field of interrelations that, in a large size and scale, is incredibly complex. Such a grid system permits the mapping of a number of variables within a larger distribution that can be considered to be total and all encompassing as a system. In constructing alternative automated systems, I have attempted to employ the grid structure as the underlying paradigm of these systems that is recurrent throughout their design and function. It is therefore a central question to try to answer as to what is important and significant about such grid systems that make their use in intelligent machine designs so appropriate. The answer to this question is not obvious.

Such systems have several attributes that are significant for the construction of intelligent systems and that relate direction to operational metasystems. These attributes are:

1. The capacity to systematically organize, compare and manipulate an almost infinite number of different attributes or very large sets.

2. The capacity to represent a total language system and grammar in terms of the total paradigm of strings generatable by the language.

3. The capacity to construct complex list processing structures that can be analyzed and parsed in an unlimited number of dimensions.

4. The capacity to represent the total search-solution space for a paradigm of possibilities represented by a particular problem set.

5. The capacity to interrelate different grid structures on the basis of their size, dimensions and cardinal properties associated with these systems.

In short, I believe arrays or grids have a very generalizable form (rows to columns) that allow us a relatively facile way of organizing and manipulating a very broad range of data and very complex and large data sets. They allow us a means of managing complex information in such a way as to be assured of representing knowledge and information of all kinds within a similiarly structured framework. Grids represent what can be considered to be a systematic means of measuring and representing data across the total range of their distribution.

On an even deeper level, I believe that such grids reflect something fundamental about human information processing that is tied to human language and symbolization. The string structure of human language and the symbolic frames of reference and substrate of meaning upon which cognition and language depend permit people to organize their experience in a certain manner, and this kind of organization of experience is reflected in the structure of grids.

Grids by themselves can be thought of as nothing but empty frameworks needing to be filled with useful information. To be meaningful, data sets must be ordered in some manner, and this sense of order may as often as not be an arbitrary sense of order we bring to such systems. Further, the more active and intelligent the individual frameworks can be made, the more dynamic and interesting the resulting grid table and dynamics will become. Cells of grid structures are not just slot-and filler to be filled with passive information--they are made dynamic as sub-functional or variable entities that interact with other cells and with the system as a whole. A distributed and interconnected computer network represents an array of very integrated and "intelligent" components. Each is capable of processing in an independent manner, but within an array framework each would yield some measure of this independence of function for a specialized role within the larger framework. Role specialization of such systems can be made flexible through modularized assignments of different tasks to different systems.

The value therefore of the grid seems to be the multilinear organization of data in series of parallel rows (or, alternatively, columns), and the built in addressing system that permits information anywhere within the grid to be rapidly located and written to or read. Blocking of information into cells in a grid matrix allows for a discrete organization of function and distribution that permits the complex superorganization between different grid frameworks.

Light computing offers an unrealized potential for the extension of conventional digital computing methods utilizing many of the same basic devices. Light amplifying diodes, holographic media, signal modulation of lasers, optical fiber computing and photo-electric devices offers the possibility of a wider range of application of these kinds of technology to computer and information processing systems than has yet been realized. I would add this this list certain kinds of sensitive filter systems and spectro-photometric systems that permitted a continous reading of light signals. Light shares with electricity certain intrinsic properties as a form of energy that makes its consideration in computing attractive.

Brains can be considered to be organic computing systems, pure and simple. Nature arrived at intelligent systems within its cellular framework in the same manner that it invented the first motors and engines and informational processing systems. Brains were thus a solution to the problems of coordination of bodily function or physiological process, mechanical coordination and senory-motor response to environmental stimuli, and the organization of behavior into complex patterns of reaction and relation between different organisms. Biological systems therefore have been based on a wide variety of what can be considered as alternative systems that had no precedence in the natural world. They were not "invented" in the manner of human alternative systems were achieved. Rather they were "evolved" as adaptive mechanisms that solved basic problems of functional adaptation. That these brains evolved through many millenia of trial and error and development goes without remark. There has been a sense of progressive development of brain function, as many dinosaurs were reputed to have had relatively small sized brains, thought this stereotype of slow-witted, slow moving monsters has changed recently.

The evolution of the primate and hominid brain was a remarkable achievement, but we can find other relatively big brained animals in nature. Brains are plastic tissue organs just like any other kind of organ in the body. Inherent variability of genotypical pattern and phenotypical expression of brains results in selective pressures favoring certain specific conformations of brain size, structure and function. Big brains evolved on environmental demand--pointing to the fundamental requirement of solving through coordination or complex response pattern some difficult problem or challenge to survival in the larger framework.

The consideration of input-output feedback loops brings to focus the concern with human user interface designs, and this represents an aspect of autonomous systems that can be said to structurally constrain such autonomy. In short, we can say that there may be more than one kind of input, and not all inputs have to be humanly initiated or ordered ones--computers can be designed to directly monitor environments through a variety of sensory input systems. Similarly, we can say that the more different kinds of outputs a computer based system is capable of, and the more derivative these outputs from the central information processing function of the computer, the more useful and adaptable such a system would potentially be in the world. Not all outputs have to be either human-mediated or human-intended kinds of outputs. They can include outputs to other machine systems or other computer based systems that are capable of functional response or productive work.

At the same time, the concept of the interface as being necessarily that between the human and the machine, in which the inteface is a human mediational device, can be seen as perhaps a product or central premise of the Turing test for artificial intelligence. There is no need for an interface design to be restricted to a human-user framework. Increasingly, the construction of mix and match supercomputer systems demands the use and development of computer-to-computer interfaces that permits talk between different processing architectures and data-structures. Similarly, interfaces can also be considered between sensory systems, on one hand, and motor-control output systems on the other hand, and these do not have to exist at the same place and at the same time.

Ideally, we would desire what can be called a "universal interface" design that would permit the widest range of input-output mediation possible between different kinds of systems, including human beings. This would entail that the interface itself would be an independent intelligent system capable of detecting and translating between different kinds of signals, and generating a wide variety of relevant outputs. Behind this would stand a capacity to coordinate signal relay systems between different components or subsystems, the ability to intermediate effectively between different levels of different systems.

Intelligent transmission involves the possibility of signal transmissions being able to be processed in transit time in some meaningful manner. Relay switches in transmission lines would include an intelligent, autonomous function, but this is a nodal transition network model compared to what can be called an in-line processing system that might possibly have combinatorial functions independent of nodal switching points. The capacity to program and process information would therefore have to be built directly into the transmission line itself, as a part of the design of the line. This seems like an alternative strategy, if it were possible, to the concentration of integrative systems into a single unit. Ideally, a computer system would consist of nothing but a variegated mass of transmission lines interconnected in some complex manner without necessarily any junctions occuring between them, or possibly with many junctions embedded within the line itself.

We can imagine different architectures for such lines a few of which follow:

1. segmented structure (string of pearls structure)

2. coiled or helix structure (twisted DNA structure)

3. laminar flow structure

4. parallel linear, interconnected structure (telephone cable structure)

5. branching braided structure (rope structure)

6. needle-point loop structure (stitch & sew structure)

7. inter-looped structure (anchor chain structure)

8. rung-structure (ladder structure)

9. by-pass structures (railroad switching yard structure)

10. axon-synaptic structure (neural structures)

11. sphaghetti structure

12. complex hybrid structure of any permutation of 1-11.

Furthermore, borrowing an analogy from protein folding structures, we may identify possibly different levels or orders of folding structures in such systems:

1. primary structure

2. secondary structure

3. tertiary structure--"Gordian Knot structure"

4. quaternary structure

Furthermore, it seems as if these structures can be further interconnected with one another to form various kinds of network or chain-mail or fabric structures that again can be folded or convoluted in various ways. We can imagine structures that are one-way, switching or reciprocal pulse transmissions or again some combination of these. Considerations of these alternative patterns suggest an entire plethora of possible processing structures, but we must inquire whether or not such a primary structure is feasible and efficacious in the first place. For instance, can transmission of signals be effectively combined with the transition of such signals in a linear and sequential manner? Considerations would also include the cost and ease of manufacture and maintenance of such lines, if they were possible, as well as their relative efficiency in either the communication or processing capacities. At what point does the length of the line trade-off with the necessary breadth requirements associated with expanded processing capabilities?

If we can imagine a steady stream of binary code along a transmission line, and an in-line tape reading device that was capable of matching sequences of a certain order, signaling some transition to occur. A separate relay line might result in a series of down-line switches to occur that would channel information in some variable manner.

Another aspect of autonomous integration and distribution of such systems is in terms of what can be called the sensory-motor, or stimulus response, feedback loop, and again, this process does not necessarily have to be mediated by means of human inputs and responses. An entirely automated system would be independent of human mediation, and therefore would be capable of directly monitoring and collecting information from the environment by various forms of mechanical sensing devices or "organs." A device for instance capable of descrambling and converting sound waves into signal code, allowing for its transmission and reproduction in some other form, would be an example of such a device. This device must be capable of meaningfully reading the signals in some patterned way in order to avoide incorrect interpreation or translation. Light or some wavelenght sensig would be a most important type of sensory input to be capable of processing. Sophisticated systems should be capable of monitoring and defining imagery and complex fields that are variagated and continuously changing.

On the other end of this kind of loop, some form of functional feedback or behavioral state modification would be an expected outcome of certain kinds of input signals that were autonomously gathered from the environment. Much of this kind of output cycle would be connected to industrial processes or alternatively to equipment or machines that, for instance, modified ambient environments or lighting conditions in a building, etc. The use of extensible, folding armatures and mechanical hands that demonstrated a degree of "hand-eye" coordination and dexterity comparable to a pair (or a set) of well trained human hands would offer a very wide range of potential applications. Autonomy of function would have to be built increasingly into machine systems themselves, such as with modern motor vehicles that are capable of monitoring and adjusting their own operating conditions.

Mosaic Processing structures refer to the specialized division of labor within integrated systems and between distributed systems, such that no single system performed only one function exclusively, and all component systems were capable of potentially peforming the full range or variety of functions as part of the overall system. Mosaic Processing Structures refers to the variegated and irregular character of this division or stratification of function within complex systems, and to the cooperative and interactive nature of such systems between the various components and subcomponents. We would expect within mosaic processing structures considerable degree of overlap between different processing structures in order to prevent reduplication of effort.