Knowledge that Transforms

The development of the Even-Flow system was initiated by Mr. Gregory M. Boni, who observed that the elapsed time to complete a job in job shops was often a function of the number of operations performed rather than the time spent actually performing the operations, and that the time between operations was relatively constant. When this is true, the job shop can be thought of as an assembly line in which the machines are the stations, and the work to be done at each station consists of all of the operations which must be performed if the jobs are to flow evenly from machine to machine through the shop. The existence of a predictable time between the successive operations of each job gives one the ability to schedule the completion of parts at desired due-dates with reasonable assurance. This ability is crucial in the production of parts for assembly lines; the shortage of one part could result in shut-down of the line or expensive out-of-station work. The Even-Flow system described below is an attempt to maximize the number of on-time completions by integrating the scheduling, loading, and dispatching functions in a manner which capitalizes on the “natural” movement of work through the shop. The simulation of this system in comparison with other scheduling rules, described below, was conducted by Professors Conway and Maxwell on the Cornell University Simulator, a program written for the IBM-650. Management Technology, ISSN 0542-4917, was published as a separate journal from 1960 to 1964. In 1965 it was merged into Management Science.

A multiplicity of techniques must be used in systems analysis in order to answer the variety of questions that arise in research on management systems. A research organization devoted to solving management science problems will inevitably use mixtures of these techniques. The purpose of this paper is to examine the nature of systems analysis, and past experiences with it, to help develop a strategy for selecting appropriate techniques. To facilitate the development of this strategy, several categorizations are developed: (1) of the characteristics of the kinds of systems studied by systems analysts; (2) of the many purposes of systems analysis; and (3) of the attributes of many systems analysis techniques. Then the simulation category is examined further with a more detailed breakdown presented on the spectrum of simulation techniques. Finally, an illustration is given of the way in which the various systems analysis techniques have been used to evolve a greater understanding of supply and inventory policies and systems. There seems to be a natural way in which the use of the several techniques evolved and were fitted together. It is believed that this pattern has occurred in other large systems studies, and that it would be applicable to those large and complex systems in which there would not be a significant pool of prior research available. Management Technology, ISSN 0542-4917, was published as a separate journal from 1960 to 1964. In 1965 it was merged into Management Science.

A Simulation Process was developed at the El Segundo Division of Hughes Aircraft company whereby actual shop operating data contained in an existing data processing system is used as input to an IBM 7090 computer for simulation of the El Segundo Fabrication Shop. The Simulation Process was used as a study tool to evaluate the effects of various management decisions on the operation of the Fabrication Shop. Management Technology, ISSN 0542-4917, was published as a separate journal from 1960 to 1964. In 1965 it was merged into Management Science.

This paper applies, to ballistic missile systems, replacement policies for equipment subject to stochastic failures in which the actual state of the system—good or failed—is not known with certainty. Although it applies these policies only to the missile itself, optimal maintenance policies for the system as a whole are conceptually similar. The criterion employed for comparison of alternative maintenance policies is to minimize the total cost per expected piece of equipment good per period of time. Maximization of system reliability is a special case of this criterion which ignores the financial costs of maintenance actions. The range of maintenance policies examined may be subdivided in two ways. First, it may be assumed that the distribution of times to failure for parts not subject to inspection is known. Secondly, it may be assumed that the distribution is known but that one or more parameters characterizing the distribution are unknown. The central feature of these latter policies is that the interval between replacements is adapted to new information as it accumulates. A second subdivision encompasses those maintenance policies in which the time to replacement for the non-monitored part depends on the state of the remaining parts of the system, and those in which only the failure rate of the uninspected part and the cost of maintenance actions are taken into account in determining the replacement interval. The first class of policies is called “opportunistic,” since optimal policies are designed to take advantage of opportunities arising through failures of the remaining parts. Opportunistic policies have a structure which may be described by a critical number N and a set of such critical numbers n 1 ,n 2 , …, each corresponding to one of the parts under inspection. The optimal policy may be described as follows: For 0 ≦ x ≦ n i , the non-monitored part is not replaced if the i-th monitored part fails; for n i ≦ x ≦ N, the non-monitored part and the i-th monitored part are replaced whenever the i-th monitored part fails. For illustrative purposes, optimal policies of each type are computed for a hypothetical ballistic missile system. They are then compared with each other and with a rule of thumb frequently proposed for replacement policy: Replace at the mean time to failure.

In general, it is non-optimal to estimate, on the basis of past experience and as a basis for present action, an a posteriori statistical parameter other than the optimal action itself.

A new efficiency criterion is proposed for the Markowitz portfolio selection approach. It is shown that the use of standard deviation as a measure of risk in the original Markowitz analysis and elsewhere in economic theory is sometimes unreasonable. An Investment with a relatively high standard deviation (σ) will be relatively safe if its expected value (E) is sufficiently high. For the net result may be a high expected floor, E − Kσ, beneath the future value of the investment (where K is some constant). The revised efficient set is shown to eliminate paradoxical cases arising from the Markowitz calculation. It also simplifies the task left to the investor. For it yields a smaller efficient set (which is a subset of the Markowitz efficient set) and therefore reduces the range of alternatives from among which the investor must still select his portfolio. The proposed criterion may also be somewhat more easily understood by the nonprofessional.

If certain information is given about the “psychology” of the players who participate in an n-person cooperative game, concerning their bargaining abilities, their moral codes, their roles in the various coalitions and their a priori expectations, then it is possible to define a measure for the power of each coalition which, perhaps, is a better description of the game than the usual characteristic function. The required information, called the standard of fairness of the players is a Thrall partition function which satisfies certain requirements. Its determination is discussed both from an experimental and from a theoretical point of view. In terms of the power, every game becomes a constant-sum game. Applications to the von Neumann and Morgenstern solutions and to the bargaining set are discussed.

It is extremely dangerous to extrapolate a polynomial trend. Although there are other sources of danger, this study concentrates on the statistical error, which is very large for end values and increases alarmingly with extrapolations. Relative standard errors of polynomial trend values are summarized, and generalizations are made concerning the relative magnitude of the standard errors with changes in: (1) the degree of equation; (2) the number of observations; (3) the position of the trend value in time. The most interesting generalization is that the relative variance of a polynomial of degree m, as the position of the trend value departs farther and farther from the central position in time, may be described by a polynomial equation of degree 2m.

Simple rate comparisons, the use of biographical characteristics of the individual, and the use of a formal definition of resignation are all parts of the method of resignation analysis in common use among students of administration, personnel technicians, and executives. Tests of the assumptions on which these common tools are based indicate that they are open to considerable objection, both for practical organization purposes and for purposes of research in organizational behavior.

This paper gives an account of an experiment in the use of the so-called DELPHI method, which was devised in order to obtain the most reliable opinion consensus of a group of experts by subjecting them to a series of questionnaires in depth interspersed with controlled opinion feedback.