|
p.5 When we seek to examine these urban systems and their development, we confront a series of choices between the different ways they may be visualized... What we see depends so much on the ways we choose to look at things. Alternative ways of seeing and representing urban systems affect our concepts, the data we utilize, the methods we apply to them, the nature of the descriptive results, and the degree to which we can move beyond description to explanation.
p.5 To observe an urban system (or any cognitive entity), we must choose between modes of cognition and levels of behavior. Two observational modes can be employed in perceiving and representing urban systems (or any cognitive entity). The first one is classificational and differentiating. It identifies urban entities by describing their physical object and activity characteristics. The second mode is relational. It is concerned with the way the elements of urban systems are linked to form holistic patterns and coherent processes.
p.5-6 When we examine an urban system as a behavioral entity, both observational modes can be practiced at each of several levels... [JLJ - text below formatted for readability]
- The first is the object landscape of the urban system. This consists of physical artifacts and other physical manifestations of human populations...
- The second level has to do with the physical activities of urban households and establishments as they transform and transfer the material and information base of an urban society.
- The third level involves the programs or "control modules" of urban system behavior. Its elements are the rule structures governing those information processes that shape second-level activity patterns.
- The fourth level of observation deals with developmental information processes that "reprogram" urban system behavior.
...At each of these levels of observation we may practice either the classificational or relational mode of cognition characterized above, or both. [JLJ - importantly, Dunn implies that there are eight ways of observation we can apply to the study of urban systems, perhaps even to systems of similar complexity as well]
p.8 Sometimes behavioral systems are unable to cope with environmental disturbances by drawing on their preprogrammed managerial information-processing responses. They are then forced to engage in developmental information processing at the fourth level of behavior... Adaptation requires more than a modification of activities mediated by the firm's established purposes and ways of behaving. It requires a restructuring of purposes, ways of behaving, or both. Disturbances such as these are not resolved by routine problem solving. The essential response requires creative or developmental problem solving... The range of creative responses available may be substantial. The enterprise may institute a new production process... it may reorganize its marketing procedures... it may modify its standards to accept a lower rate of return; it may reorganize its performance processes... In all cases, a major transformation takes place in its established ways of behaving... As suggested in the previous example, the practice of using the classificational and relational modes of observation at this level of behavior is an underdeveloped frontier of the behavioral sciences.
p.8 The natural boundaries of an individual... are usefully defined by the limits of autonomous control. There is always some ambiguity, of course, because all behavioral systems are open systems engaging in cross-boundary transactions that, in turn, modify the character of their environments... The big problem with social entities is that they are skinless and we are forced to conceptualize boundaries in terms of degrees of autonomy in functional control.
p.9 In the relational mode... we are more concerned with the spatial patterns formed by the human population and physical structures as they have adapted to the activity pattern they serve, as well as to the natural landscape they inhabit. The traditional way of forming such a relational image has been to develop a schematic map of the position of these structures in space, demonstrating the way they are both linked and separated... When we move to the second level of observation (physical activities), we begin to encounter fresh difficulties, particularly in the relational mode of observation... The difficulties are of two sorts. First we have to negotiate an enormous jump in the scale of observation. We are faced with a substantial increase in the complexity of the activity patterns, as well as a major enlargement of the geographical span over which these activities spread... Assuming that such a complex transaction network can be conceptualized, observed, and represented, the image it offers raises a second difficulty, more disturbing than the first... From the point of view of a relational network of activities, these subentities of the national economy are revealed to be "open systems."
p.9 We enjoy some success in developing activity classifications because observing the typical activity of a behavioral entity is not excessively difficult. Such descriptions can be generated by standard statistical reporting techniques.
p.9 The purposes of this study are best served by conceptualizing urban structure as a complex set of overlapping and intersecting transaction networks formed by the linked transformation and transfer activities of people and their social organizations.
p.10 Whenever we designate a meaningful field of observation we are confronted with the task of representing the contents of that field in a symbolic or schematic form. This is necessary to make it intelligible to ourselves and to others... I wish to... use graphics to represent schematically the classificational and relational modes of defining the field of observation.
p.10 An exercise in classification is equivalent to drawing a boundary around a field of observation... graphic boundaries can be used as representational devices.
p.18 Our observational scheme tells us that explaining physical activity networks requires the we consider information-processing activities at the programmatic level of behavior. Here we find decision rules that help explain the spatial agglomeration of physical activities.
p.18 Since all activity processes incorporate transfers in space, establishments modify the geographical pattern of their nodal locations and relational linkages to avoid unnecessary costs and to increase revenues. The resulting spatial patterns are formed by the mutual coadaptation of each establishment to the operating environment formed by all establishments and their mutual symbiotic relationships.
p.19 Given the right conditions, this process of urban agglomeration may feed itself.
p.20 resource sites are not exempt from the coadaptive process that shape urban agglomerations and networks.
p.24 The creation of an urban network is an evolutionary, developmental, historical process. No total-system social planner... ever has the opportunity to create an optimum social process de novo. All social establishments are constrained to the choices inherent in an operating environment created by history. And the choices they make in solving the problems of their existing situations change the nature of existing operating environments (that is, patterns of the relational urban network) in ways that create new problems and choices.
p.24 As we have seen, boundary nets may be drawn for any classificational purpose. The kinds of geographical networks employed in central place theory can be useful for many purposes... Here I assert a principle. If we wish to designate boundary networks with meaningful real world (empirical) referents, we would best start with relational networks rather than the other way around... We can employ relational networks as a tool in drawing meaningful boundary networks, but designated boundaries can tell us nothing definitive about internalized activity relationships.
p.26 The conceptual metaphor offered in this section views the urban activity network as the result of two opposing forces, internalization and externalization.
p.29 The human individual is, of course, the information processor par excellence.
p.31 The activity systems we know as urban systems display a labyrinthian set of linkages. They form a complex of interconnecting, overlapping, overlaying, relational transfers and transformations of the kinds we have been discussing.
p.34 adaptive information processes that initiate, or react to, changes in network structure are themselves an important part of that structure.
p.34 Our image of urban structure is complex. Therefore, any image of the dynamic transformation of structure cannot escape being complex as well. Once again we progress by identifying generic sources of change and generalizing about their impact.
p.34 As... establishments, enterprises... encounter changing environmental circumstances, they modify their interactions with other establishments and enterprises, thus forming their operating environment. In doing so they initiate further changes in the operating environments of the related entities.
p.34 When organized, social information-processing activities encounter environmental disturbances, they have a choice of two types of problem-solving, adaptive responses. The first is programmatic, and the second is developmental... As social entities grow or develop, their relational linkages with each other at urban scale become reordered.
p.35 the structure of urban activity networks is likely to be transformed by simple growth.
p.36 ultimately, growth depends on development. Were it not for creative technological and organizational changes in behavior, growth would ultimately grind to a halt. Unless the environmental restrictions encountered by growth are displaced by developmental adaptations, slack in the system will disappear and activity networks will settle into a steady-state equilibrium... Thus, growth is a lagged consequence of creative development. It rests on increases in the productivity of social activity born of more effective ways of doing things... In any adaptive situation, one initially, and most naturally, turns to pre-programmed, rote adjustments... But these growth options would not exist were it not for slack generated by antecedent developmental adaptations. More important, growth options are often insufficient to resolve problems without bringing additional behavioral changes into play. Thus the explanations of change associated with growth are meaningful but only proximal. We must look to the developmental sources of urban change for more fundamental explanations.
p.36-37 developmental adaptations tend to transform the logic of the entire system. They radically alter activity relationships and exert a much more direct and dramatic effect on urban network patterns... The essential point is that these changes bring about direct and dramatic changes in the network logic of the urban system, not just marginal adjustments in the density and range of its characteristic activity configurations... Thus, if we wish to explain the sources of urban change, we must devote special attention to developmental adaptations. They are the prime sources of all changes
p.39 Wider response options can be introduced by constructing machines that can do more things
p.44 Do the changing geographical patterns of urban activity systems form recognizable sequences in time? If so, can we generalize about them in a meaningful way? In short, can we derive a theory of urban growth and development based on sequences of urban change? ...Actually, from the viewpoint of the total urban system there is little to be said about growth sequence patterns. Sequences of change are primarily unrepresented as growth trends of undifferentiated aggregates. Even if we visualize the comprehensive spatial network of urban activities, there is not much to be said.
p.47 the growth and development experiences of individuals, establishments, and enterprises... are intraregional sequences formed by the interdependent, coadaptive responses of establishments whose linkages are internalized by the urban region.
p.47 Every embryonic urban region does not carry the genetic potential for duplicating the functional and spatial structure of its parent region. Its potential for growth depends on the nature of the growth impulse that seeds it and the situational environment in which it is implanted.
p.49 Whatever the locus of the creative spark that invents a new way of doing things, it will be applied throughout the total system wherever the problem it addresses is present.
p.49 the roots of technological change are to be found in obtrusive environmental problems not accommodated by managerial adaptation. Such problems... breed technological solutions involving changes in physical- and information-processing techniques. These changes are designed to extend the behavioral range of social activities and ameliorate the problems addressed. In the process they change the structure of relationships between establishments as well as the relationships society must maintain with host physical environments. The operating environments shared by establishments are transformed. There is a paradox at work. Inevitably these invented environmental changes carry the seeds of new problems. So we do discover a developmental sequence of sorts at work. Problems breed solutions that breed new problems for solution. It is an open-ended, never-terminating process... novel social forms and activities are not responses to selective environmental pressures like those described by population genetics. They are the products of creative cognitive processes employing anticipatory imagination to invent novel behavioral options... All existing social technology has developed as a means of amplifying human physical and information processes. It has evolved tediously over time as one developmental sequence out of many possible outcomes - that is, problem → attempted solution → new problem → solution → and so forth.
p.52 The effect of it all was a dramatic reorganization of the network structure of the urban metasystem. But, true to the evolutionary process, these developments yielded new problems... Evolutionary developments in the behavioral capacities of individuals, establishments, and enterprises became manifest in the transformation of urban system networks. The relational networks that represent one epoch's solutions to antecedent problems breed new access and transformation problems to be resolved in the next epoch.
p.53 Can we generalize about developmental sequences in the same way we identified certain sequence patterns characteristic of a single growth track? The answer is no... when we turn to developmental sequences we are concerned with sequential changes in second-level, relational activity patterns induced by changes in third-level managerial programs further induced by fourth-level, developmental, problem-solving processes. We are barred from generalizing about developmental sequences of third-level technological changes because the fourth-level developmental processes are social learning processes. They are opportunistic and problematic rather than rigidly pre-programmed. We are limited to the heuristic process generalization set forth at the outset of this section: problem → solution → new problem → solution, and so forth. This problem-solving process can generate a seemingly limitless variety of third-level technological-programmatic sequences - each of which, in turn, specifies a different second-level, activity network growth track.
p.53 Growth sequences have to be seen as an integral part of the evolutionary process.
p.56 we can focus attention on the changes in urban transaction network patterns associated with various explanatory sources.
p.56 to generalize about a sequence of change, one must be able to recognize recurrent developmental sequences over some significant historical period. Patterns... take the form of sequences of encountered problems and developmental solutions that, in turn, generate additional environmental disorders requiring further developmental solutions... these historical sequences are unique an nonrecurrent. They are the result of n evolutionary process. We may be able to generalize about the process, but not about the future patterns of its behavioral consequences. [JLJ - sounds like adaptive capacity is required here to prepare for exact conditions that we can generalize about, but cannot foresee.]
p.59 Publicly available general purpose data systems are dominated by classificational data poorly suited to represent relational patterns. It is true that much of the available data carries identifying labels with activity referents... They are of no use in constructing relational activity networks. Activity states are single point-in-time-and-place classifications. Relational data, however, combine several observational elements into a composite datum. [JLJ formatting added below for readability and emphasis]
- A "state-of-origin,"
- a "transformation-transfer process," and
- a resulting "terminal state"
must be specified. Each observation requires three classificational elements instead of one - and the key linking element is a "process" rather than a "state" variable. Moreover, the process... is identified as a relational link in an activity progression.
p.59 Even though we confront the task of representing relational networks, the nature of the data resources restricts us to the conventional classificational-statistical mode of observation. We must, therefore, shape conventional data into boundary networks structured in a way facilitating inferences about the relational networks that most interest us. In effect, we will construct images of urban systems with a "camera" possessing a limited depth of field... Visualizing the explanatory "background" will require a good deal of "image"-ination. Fortunately imagination is not without the support of some meaningful clues.
p.59 Because we must work primarily with conventional statistical classifications, it is important to adopt those that facilitate inferences about underlying relational networks.
p.60 activity relationships can be employed to designate more meaningful boundary networks. What I propose to do, therefore, is to draw upon those smidgeons of activity process data available to define geographical boundary networks that we then employ to generate conventional statistical constructs composed of classificational data.
p.64 To engage in the relational mode of observation is to conceptualize and represent an open system of relationships. Any attempt to name and specify the characteristics of such a system requires that some classifying boundary be imposed on the system. By its nature, this is always arbitrary... An activity network does not end at some object surface... It is an open web of activity tentacles or transaction linkages... if we had access to good relational data, we would not need to rely on such crude methods.
p.64 An entity is identified by specifying those characteristics that differentiate it from other entities.
p.64 Many behavioral patterns have short life spans because of the influence of evolutionary social development. This is a generic problem for all social science
p.113 We have noted that the difference between two normalized cross-sectional industrial patterns at two points in time presents problems of interpretation because the sector differences can be traced to any one of three elements embodied in the computation... a difference vector that describes the sectoral structure of the change for each region... reveals nothing about the dynamic structure of the regional change for each industrial sector. If we could come up with a profile of the elements of such change, we could then contribute a great deal to an inferential explanation of urban-region change... To accomplish this, we turn to a procedure now widely known as the shift-share method. This method has the explicit objective of revealing aspects of this dynamic structure. [JLJ - any technique to turn statistical data that can be counted, into action-type data that reveals dynamics, is worth looking into.]
p.116 The principal objective of the shift-share technique is the designation of a set of summary components that help "explain" changes in the total employment of a region.
p.117 One of the great attractions of this method [JLJ - the shift-share technique] is the fact that the three components of the change in total regional employment provide an "explanation" of regional change at the summary line that takes into account both the structure and dynamics of thirty-one disaggregated industrial sectors.
p.118 We earlier set for ourselves the difficult task of trying to shape traditional classificational-statistical data in ways that help us visualize underlying relational activities. The shift-share technique, just outlined, carries us another step along this path by identifying three components of changing activity structure.
p.120 The shift-share technique is offered as a method for representing these complex structural changes and suggesting preliminary, although limited, explanations.
p.163 The above leaves no doubt that the shift-share components are dramatically affected by underlying transaction networks and their pattern changes.
p.165 It is one of the inherent limitations of the human cognitive apparatus that each conceptual classification hides as much as it reveals (although such mental "image"-inations are our only path to formal knowledge). Thus, one cannot clearly differentiate... sources of change... Nor can one unambiguously separate "mix" effects... from "competitive" effects... Preliminary explanations gloss over the fact that the active sources of all change, including growth, are developmental in origin. They are reflections of the changing modes of behavior of people, establishments, enterprises, and governments. Thus, more complete explanations... must be traced to higher order developmental activities.
p.175 initial structure is itself the product of the region-specific developmental effects of an earlier period. In short, the final explanation of the national growth and industry mix components also resides in the full array of evolutionary, problem-solving adaptations to environmental challenges exhibited by specific behavioral entities in specific urban regions.
p.175 The shift-share technique permits us to extract and measure the national growth and industrial mix components. The preliminary, proximate explanations take advantage of this characteristic so as to begin explanation with the growth sources of change. One now sees this as a way of sweeping complexity "under the rug" until we can get a preliminary grasp on the nature of change. In the end we must understand that developmental sources are the prime movers of all change and their initial impacts are always localized... To lose sight of this is to risk reifying the artifact shift-share components of change. To do so could become an obstacle to meaningful policy analysis.
p.181 The demonstration that developmental sources of change are the driving force and dominant influence behind urban-region transformations is perhaps the most important single finding of this study... Meaningful explanations of urban system development are clearly demonstrated to take root in those higher levels of observation dealing with programmatic changes - changes born of the social adaptive responses evoked by each epoch's distinctive matrix of problems.
p.181 The inadequacy of traditional prediction models. Formal approaches to prediction are virtually useless in the developmental context revealed by this study. Formal statistical models designed to project specified empirical parameters lack the capacity to anticipate and incorporate emerging developmental factors - sources of change that modify the underlying process logic of urban systems. I will suggest that the "pseudo-rigor" of such formal models would be better replaced by a more eclectic means of anticipation guided by evolutionary developmental concepts.
|
