SYSTEMS AND THINKING, GLOBALIZATION OF KNOWLEDGE, AND COMMUNITARIAN ETHICS INTERNATIONAL SOCIETY FOR THE SYSTEMS SCIENCES
s PROCEEDINGS OF THE FORTY-FIRST ANNUAL MEETING OF THE ISSS SEOUL NATIONAL UNIVERSITY SEOUL, KOREA JULY 22-25, 1997
EDITORS YONG PIL RHEE
KENNETH D. BAILEY
CO-EVOLUTIVE ACTION-DESIGN METHODOLOGY Nagib C. Callaos Dpto. de Procesos y Sistemas, Universidad Simón Bolívar, Apdo. Postal 89000, Caracas, Venezuela e-Mail: Internet:
[email protected]
Abstract Elsewhere [3] we differentiated between systemic and systematic methodologies. Systematic methodologies are usually more efficient and adequate for a stable environment, with a low level of uncertainty. Systemic methodologies tend to be more effective and are more suited to dynamic environment with high level of uncertainty. They are more effective because they are adaptive, while systematic methodologies are usually more rigid. So, systems analysis and synthesis in a dynamic and uncertain environment need a systemic methodology. We attempt in the paper to describe the general outline of such a methodology, which has been tested in and nurtured from more than 100 real life projects. This methodology is based on a combination of the "disjointed incrementalism theory" [1], the paradigm of action-research and action-learning, the total quality approach and some cybernetic concepts. Consequently, we transformed the "disjointed- incrementalism" in a "conjoined" one, inserted into an co-evolutive process of action- analysis and actionsynthesis. The meta-methodology which we are using to design the systems analysis and synthesis methodology we referred to, has also, recursively, the same characteristics and is nurtured inductively from the application of the methodology to specific projects. Keywords: Methodology, Action-Design, Conjoined Incrementalism, Co-Evolutive Development. Introduction Elsewhere [2] we concluded that a methodology is a set of related or relatable methods (ways or "roads" of thinking and/or acting) with their respective tools (means of "transportation" through the "roads"/methods) and techniques (aptitudes and attitudes required for adequate management of the tools), in order to achieve given objectives. If the methods are related in a pre-established way, the methodology will be a systematic one, i.e. the methodological system will be closed, rigid and efficiency-oriented. If the methods are relatable, i.e. they are able to be related in a way or another depending on application domain context, then the methodology will be systemic, i.e. the methodological system will be open, flexible and effectiveness-oriented As systemic methodologies are not predefined, they should be defined a new for each specific application, i.e. they should be "post-defined" What could be predefined in systemic methodologies are (1) non-exclusive sets of possible methods, tools, techniques 2
and alternative relationships between or among methods; (2) non-exclusive set of rules and heuristics for selecting adequate methods and conforming relationships, according to the specific application and its specific context, i.e. a knowledge base, or an open expert system [6]; and (3) a meta-methodology which would support the process of post- definition, project design (planning), project execution and project control by means of feedback and feedforward loops. In this paper we are attempting to give some general metamethodological guidelines. Conjoined Incrementalism Systemic methodologies are used, explicitly or implicitly, when methodological adaptability is required, and this is necessitated when the process of the specific methodological application is going to be inserted in an uncertain context. So, we need methodological decisions under uncertainty, for systems development in a changing environment. Being this the case we thought that managerial theories of decision under uncertainty would be helpful in the meta-methodological design we are attempting. Prescriptive theories of decision under uncertainty showed to be unhelpful in our case because they assume, as given, factors that are not given in our case; as, for example, the set of possible outcomes of the uncertain situation. But, descriptive theories usually make no assumptions not to be found in real world situations, because they are actually trying to describe how decisions are in fact being taken under real, non-theoretical, uncertainty. So, we reviewed this kind of theories and concluded that the Braybrooke and Lindblom' s [1] "disjointed incrementalism" is a suitable starting point for us.
According the "disjointed incrementalism" theory, managers and executives, when faced with uncertainty, they look for no optimal decision based on a global perspective, neither on an exhaustive list of present and possible future events. This would need more time than the executive is allowed to have before making the required decision. Besides, the content of the exhaustive list would very probably change before finishing the process of list identification. This would throw the executive into an infinite and sterile loop. Consequently, executives, when facing uncertainty, they take an incremental, short range, decision, according to a set of criteria; then, after executing the decision, they analyze its impact, verify their decision hypothesis, identify new variables in the dynamic environment, and review the criteria set, according which they take another incremental decision, and so on. Since the criteria and the data supporting the incremental decisions are changing in a non-foreseeable way, these decision are dis-jointed. But, a system methodology is a means for achieving specific pre-established objectives as, for example, a concrete system development, a particular design, a special problematic situation diagnosis or prognosis, etc. Then, we are in a position of decision making under uncertainty, but with concrete objectives to be achieved. Consequently, control mechanism should be established such that the incremental process will surely converge in objectives, achievement. Planned feedback and feedforward control loops could be integrated to the 3
incremental process in order to direct it to the pre-established end. In this way, incremental decisions will be no more disjointed, but con-jointed in a process converging to an end fixed in advance. A chain of alternative concatenated feedback and feedforward loops applied to the sequence of decisions' increments proved to be, in our experience, a good means for conjoining incremental decisions in a thinking/acting process with a high probability of converging in the pre-established objectives. In this way thinking and acting are conducted in parallel, simultaneously, as we conducted elsewhere [4] as being an essential characteristic of a systemic methodology, and not in series as it is usually established in traditional projects management methodologies. After taking an incremental decision, an implementation of this increment follows before the next incremental decision is made. So, project planning and execution are intertwined. In the execution of each increment, clear and explicit negative feedback loops are designed and implemented in order to achieve the increment's micro-objectives. After completing the increment execution, and before deciding and planning the next increment, a feedforward control should be applied, according to the information and the experience gotten in the last increment's execution (action-learning), to the impact of the increment's execution on its environment, and to the possible changes that could have happened in such an environment and in the users' requirements. A next incremental decision will be taken based on the most proximate feedforward loop, and so on, in a conjoined incrementalism, in a process converging on the pre-established ends. In this way, incremental decisions are conjoined by explicitly planned control loops of feedback and feedforward, i.e. what we could call a cybernetic conjoined factor. But, experience shows that a human conjoining factor is also important, it is even necessary in many cases. System analysts/synthetists, system end users and clients (those who take the final decisions on the projects restrictions: time, budget, etc.), should participate conjointly in feedforward control decisions. These decisions, unlike the feedback decisions, are not just technical, and should not be taken just by the analyst/synthetist and/or the project manager. They should be achieved by consensus with the end users and the client. In order to achieve this consensus, tradeoffs should be done among technical, managerial and functional variables, on one side, and between efficiency and effectiveness of the product to be achieved and the process used as a means. This take us to a three dimensional matrix containing 3x2x2 cells, among which tradeoffs should be done. Quantitative and/or qualitative operations research models could be combined with individual and/or collective decisions theory models (in ordinal or cardinal scales) in order to "operationalize" and to rationalize in a coherent procedure/process the achievement of the related tradeoffs. These tradeoffs are not to be done isolated from each others, they should be related systemically in a real total quality, where "total" would not mean just comprehensive (as it is implicit or explicit in most total quality methodologies), but it would also mean global, i.e. systemic. This concept of strong (but not rigid) conjoining is a necessary condition if we are to apply incrementalism to systems analysis and synthesis, and to project planning and implementation. Without it we will be at the risk of falling in Lindblom and Braybrook's disjointed incrementalism, which describe adequately the way the executives take decisions 4
under uncertainty, but it is non-adequate for systems analysis/synthesis (and project planning/implementation) under uncertainty. Elsewhere we made a detailed treatment of the Systemic Total Quality and the system of tradeoff we referred here too [7]. Action-Design As we noted above, an action-learning is at the heart of the methodology we are describing. Action-Research could also be required in most cases. When action-learning and/or action-research are used to produce something (a system or a part, a set or an individual member), and not just to induce and/or reduce knowledge, or to support the achievement of a higher level of understanding of a given situation, then we are in the presence of what we can call action-design. When action-learning and/or action-research support human intentionality for the purpose of making an intention to come true, actiondesign is in progress. Elsewhere we analyzed the epistemological and ontological relations that exist among intention, design and action [5]. Co-Evolutive Development The process of conjoined decision/implementation's increments is, unlike Braybrook and Lindblom's "disjointed incrementalism", not a mere progress, but a teleological development, a purposeful growth, an intentional unfolding. It is based on a systemic notion of evolution which combines, in a dynamic whole, darwinian and lamarkian perspectives. These perspectives are, in our opinion, not contradictories, as frequently are taken, but polarities that require and necessitate each other for a comprehensive understanding of the phenomena of evolution Conceptually, they maintain reciprocal relations, and form a systemic notion of evolution where mutations intrinsic to the evolving system propel the evolutionary process, and extrinsic changes, in the system' s environment, "pro-voke" the system to adapt. By analogical thinking, we can associate these two complementary perspectives to what is known in the literature of "technological development and innovation" (TDI) as "technological-push" and "demand-pull". For a period of time there was a controversy about these two theories of TDI, but at the present, there is a consensus (still non-unanimous yet) that both forces are present (simultaneously or at different times) in the TDI. Likewise, thinking analogically, we could hypothesize that push/pull forces combine to propel and to direct, conjointly, the evolutive process. We are conscious that the analogical thinking supporting the hypothesis formulation is not an epistemological guarantee, but our experience showed us that our "push/pull" (or darwinian/lamarkian) hypothesis, has in fact a praxiological value. The possibility of epistemological value might be addressed in the future, but from the methodological perspective of this paper, the praxiological value would suffice. The conjoined push/pull evolution characterizes, not just the system, but also its environment, what we can call its co-system This is because in the adaptive process, the system react modifying itself, and "proact" modifying its environment or its co-system. 5
After the system's proaction, different parts of the co-system could behave in three different ways: (1) some parts could have no reaction or proaction at all, these could be called the passive parts, but the active parts: (2) could react adapting themselves to the system proaction, and (3) they could in turn proact modifying the system back. The system in turn could behave in the same three ways, and so on go the process in a co-evolutive loop. The system's evolution is cause and effect of its co-system's (environment's) evolution, and viceversa. The conjoined incremental action-design, we briefly described above, is a methodological system "proacting" on its environment, which is not completely passive. Social environments react/proact back almost always, and natural environments frequently do the same thing, even if sometimes there is a time delay produced by ecological positive feedback loops that require time to grow and to generate noticeable effects So, the methodology we are trying to describe briefly, is in fact imbedded in co-evolutive process Therefore an "explicitation" of this characteristic is to be done in order to make the methodology more effective and more efficient. Co-Evolution should be unambiguously specified in planning/implementing process of the conjoined incremental action/design. A fairly good format for assembling the data to be entered in the action-design process, and for supporting the participation of designers, end users and clients in such a process, is a three-dimensional matrix (a cube) of system's reaction/proaction vs. environment action/proaction vs. past facts/future prevision-anticipation Methodology And Meta-Methodology We think that the methodology, we briefly outlined here, is fairly general as to be called a General Systems Methodology. We applied it to software development (more than 120 projects), information systems analysis and/or synthesis (more than 50 projects), executive support systems and decision support systems (12 projects), strategic planning of corporative informatics (2 projects), strategic planning of technological development and innovation (3 projects), educational systems (3 projects), etc. Each one of these systems was designed and implemented with the disjointed incremental co-evolutive design-action methodology roughly described here At the same time, each one of the project represented, in turn, as a whole, an increment at the meta-methodological level, i.e. the meta-methodology by which we designed the methodology has been also, recursively, a conjoined incremental co-evolutive action-design meta-methodology. Each one of the project was one incremental step at the meta-methodological level, and each project had multiple incremental steps at the methodological level Both levels has been supporting each other The methodology is cause and effect of the meta-methodology, and viceversa. [1] D. Braybrooke and C.E. Lindblom. 1970. A Strategy of Decision: Policy Evaluation as Social process. New York. The Free Press. 6
[2] N. Callaos and B. Callaos. 1991. "A Systemic Definition of Methodology". Systems Science in the 21st Century, in Proceedings of the 35th Annual Meeting of the International Society for Systems Science. Ostersund. Sweden. June 14-20. (pp. 7179). [3] N. Callaos. 1992. "A Systemic 'Systems Methodology'" 6th International Conference on Systems Research. Infomatics and Cybernetics: Baden-Baden. Germany, august 1723. [4] N. Callaos and B Callaos, 1992, " A Systemic Methodology for Infomation Systems. Analysis and Synthesis", 36th Annual Meeting of the International Science. Denver. Colorado, USA July 12-17. [5] N. Callaos and B. Callaos, 1993. "Intention, Design and Action", 5th International Conference on Comprehensive Systems Design of Education, International Systems Institute, Asilomar, Monterey, California. [6] N. Callaos, 1994, "Expert and Information Systems", Proceedings of the Second International Conference on Expert Systems for Development; Bangkok, Thailand; Washington: IEEE Computer Society Press. [7] N. Callaos and B. Callaos, 1994, "Designing with Systemic Total Quality". Educational Technology, vol. 34, No. 1 (pp. 29-36).
7
