- Email: [email protected]
Operations research education for forgotten populations
European Journal of Operational Research 140 (2002) 225–231 www.elsevier.com/locate/dsw
Operations research education for forgotten populations ibej Joze Andrej C Faculty of Economics, Ljubljana University, Kardeljeva pl. 17, SI-1000 Ljubljana, Slovenia
Abstract Typical OR education programmes are focused on those who should professionally act as ‘‘performers’’ (specialists, modellers, etc.) in the process of business problems solving through quantitative modelling and similar formalized procedures. The problem of unsatisfactory level of understanding between ‘‘specialists’’ and ‘‘generalists’’ (managers, e.g.) has been known for decades; yet we feel that at least the latter (‘‘users’’ and ‘‘sponsors’’ as their subspecies) are still not adequately educated and this fact helps the gap to survive. We discuss different views of this problem and advocate for some approaches, mainly based upon our own pedagogical experiences. Ó 2002 Elsevier Science B.V. All rights reserved. Keywords: OR education
1. Brutal facts President of one of the most important professional OR organizations recently mentioned (see [7]) that close-to-extinction of operations research in some parts of the world was mainly caused by the academics who had paid too much emphasis on technicalities, thus allowing managers and other potential users to interpret OR as pure ‘‘mathematical masturbation’’. Business student proverb ‘‘Magical Science is a waste of time’’, cited in [11] as a negation of our faith in the usefulness of management science, somehow completes the picture. Since Ackoff’s masterpiece [1], these are probably two of the most direct warnings to the OR/ MS community that its main task should rather be co-operation with users on ‘‘grand plan’’ than
ibej). E-mail address: [email protected] (J.A. C
(merely) production of sophisticated mathematical models and algorithms, overloaded with details which can usually not be properly supported with reliable data. During the last years, there have been many other calls for resurrection of the more traditional, i.e. integral, contents-oriented, on the whole less technical and in the first place strictly interdisciplinary approach to OR/MS. The revival of classical sources (like [5,15,23], and many others) is obvious too. Yet there is probably a long way to go to bring OR/MS specialists and managers closer again. Some basic obstacles could be the following: (a) A crucial part of the problem is the fact that the roles each of the parts should play in efficient problem solving are often not precisely defined in advance; sometimes it is not even clear how different the roles and corresponding responsibilities actually are.
0377-2217/02/$ - see front matter Ó 2002 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 7 - 2 2 1 7 ( 0 2 ) 0 0 0 6 5 - 6
226
ibej / European Journal of Operational Research 140 (2002) 225–231 J.A. C
(b) Players are not properly trained for their roles, which is especially true if we consider the users’ inability to participate as active members of interdisciplinary teams. (c) The current level of communication and mutual understanding between the two groups is generally rather low. Discussing the evidence in detail could mean reinventing hot water. Churchman and Schainblatt (compare [6]) were among the first to recognize the dichotomy between researchers and managers. Although more focused on the implementation phase, they have already unveiled the most important common reasons for confrontation between them (see [8,9,13,21]), and also initiated a more general discussion, which sometimes goes all the way down to the basic differences in cognitive characteristics of both groups [14]. One of the most striking differences, namely the ‘‘language problem’’, has been evident in the OR literature for more than 40 years (and in OR practice, probably always), and some authors have systematically tried to overcome it. Vazsony’s classical OR textbook [23] even exposes it in its ‘‘mission statement’’: ‘‘It is the first attempt to develop a mathematical language which can be understood by businessmen using scientific techniques (OR, LP; statistical decision theory, etc.). The book is written for management personnel, with scientific techniques explained, not in terms of mathematics, but in terms of business’’. I believe that adequate – and in some sense more balanced – OR education for users could help a lot in achieving the necessary condition for success: enabling users to act as providers of an important part of the interdisciplinary knowledge. This does not mean that the education of OR (MS, IS, etc.) specialists is perfect, yet the users’ side still seems to be the weakest link in the chain. Situation may be different from country to country, but recent publications in INFORMS Transactions on Education [10,11,17] and other sources confirm that there is a common core of problems connected with the education of non-specialists. The crucial part of such education takes place as OR/ MS courses within (undergraduate) business school curricula, and in the first place in postgraduate MBA programmes. The importance of
OR/MS courses for technical professions and functional education for people from practice should also not be underestimated, but the main emphasis has always been on students of business. This paper focuses on the question how to design and implement OR/MS education for all those we like to call forgotten populations, i.e. users as a whole and sponsors as their subspecies. To be on the safe side, let us say that we understand the word ‘‘sponsors’’ as it usually applies in the context of IT projects, mainly in information systems development. Sponsors can be general managers, CEOs, members of the board and other persons responsible in the first place for the strategic level of decision-making. Compared to the involvement of other users, sponsors’ controls are more or less of indirect and strategic type. On the other hand, sponsors are directly involved in the formalized problem solving procedures at least at the very beginning (they have to pull the trigger and allow the process to start) and during their final phase, thus sharing the responsibility for the project. It was empirically shown that more than 90% of projects remained unaccomplished without active support of the top management in the firm. Last, but not least, sponsors must do what their name promises: pay. Typically, potential ‘‘direct users’’ of OR in firms are either decision-makers at the operational level, or planners, controllers, accountants and other ‘‘cost-cutters’’ at the tactical level. They are usually the first to encounter the symptoms of a problem situation and are supposed to be able to define the problem in exact terms. If OR/MS methods are used in the firm, these people are generally directly involved in the modelling process, data providing and/or practical implementation of solutions. We would not go that far to imply different cognitive characteristics for these groups, but there is another functional difference that makes us sometimes distinguish between sponsors and ‘‘direct users’’. While the latter may also achieve instantaneous tangible direct profits from the application of OR/MS methods through increased productivity of their individual work and consequentially higher earnings, the former can only get enthusiastic if one can make them believe that the
ibej / European Journal of Operational Research 140 (2002) 225–231 J.A. C
benefits of quantitative methods applications fit in the general managerial framework and contribute to the overall ‘‘bottom line results’’ of the firm. Therefore the motivation structure can surely be significantly different, causing possibly important differences in propensity towards OR/MS projects. However, the main gap remains between ‘‘users’’ as a whole and ‘‘providers’’, or simply between generalists and specialists in the classical meaning of these terms.
2. Teaching OR/MS: Two different directions We can only agree with Erkut’s editorial [10], that there can be a wide variety of personal research interests, but almost everybody involved in OR/MS does some teaching, either on a regular basis or at least occasionally and informally, through seminars and consultancy. Even the most enthusiastic teachers are generally keener to present methods, algorithms, and tricks of the trade to once-to-be specialists in OR/MS than to educate potential users, people whose attitude towards mathematics, statistics, OR and quantitative methods in common is as a rule rather hostile than friendly (Slovenia is not an exception, see [20,25]). The underestimation of this fact seems to be just one of the most frequent causes of misunderstandings and non-optimal performance of OR/ MS, both in practice and education. Several authors have already noticed very concrete educational problems (problems from practice are as old as OR/MS itself). One of the best papers on this subject is Grossman’s alarming report [11]. Our point of view – that there is a need for specific approaches if users should be properly educated to share our faith in OR – is best supported by author’s recognition of the fact that even the cases presented in the classroom as ‘‘management science success stories’’, often reduce to something, practised rather by OR/MS specialists than by managers. Therefore it is very naive to expect that students will be able to find connections between what we want them to learn and the general management practice they are motivated to learn through such examples. Grossman also unveils a long list of typical pedagogical ‘‘sins’’ in which
227
every OR teacher can definitely find some from his or her own ‘‘black magic box’’. So it is rather obvious that although the role of ‘‘users’’ as participants in projects is at least briefly mentioned in modern OR/MS textbooks (like [2,12,19,24], . . .), the OR professional community has so far underestimated the need for a clear separation of two types of education for two (or even three) groups. Even the (over) simplified description of the goals that should be attained through education process adds much argument for a complete restructuring of educational programmes: sponsors should understand the needs of users, having in mind the power and range of OR/MS methods. Users should be in the first place taught how to express the needs and how to co-operate with specialists. Specialists should be able to fulfill the needs. Of course, the extent of OR/MS knowledge needed to accomplish these three tasks increases in the direction from sponsors towards specialists. But there are other elements in which a specialist will never be able to compete with neither sponsors nor users. One does not have to go back to Ackoff and other classics of the OR literature to realize what makes teamwork not just important, but practically inevitable: one failure in practice will do the trick. Some kind of blending of qualitative knowledge (mostly held by users) with the more quantitatively oriented segments of knowledge (which specialists in OR/MS must provide) must be assured.
3. OR literacy The fact that users contribute their specific problem-oriented knowledge does not mean that they can be completely OR-illiterate if we hope for teamwork to be successful. My direct answer to a rather provoking question why I advocate for ‘‘OR literacy’’ for users instead of ‘‘problem literacy’’ for OR professionals is very simple. I would of course expect that an OR/MS specialist who has been working for several years in one industry has gradually acquired a fair share of ‘‘problem literacy’’. But at the same time, I believe that – especially during the phase of pre-career
228
ibej / European Journal of Operational Research 140 (2002) 225–231 J.A. C
education – it is still easier to achieve some general ‘‘OR literacy’’ in the population of potential users than to teach potential specialists a lot of different disciplines. Larger ‘‘OR communities’’ may perhaps afford an earlier problem-oriented specialization, but this is generally not the case that the rest of the world is dealing with. Speaking in terms of practice it is clear and generally recognized that a successful project cannot occur otherwise but as a net of people contributing specific quanta of knowledge. On the other hand, attempts to teach everybody everything have not disappeared from our education process. The results of such an approach are usually just the opposite of the expected ones. My usual saying that students then prefer ‘‘nothing’’ to ‘‘too-muchand-unnecessary’’ may seem exaggerated, but is more and more supported by practical experiences and other evidence (compare [7,11,17] and other sources). Defining exact meaning of the word ‘‘literacy’’ has never been a simple task. In my opinion, ORliteracy, as pleaded for in this paper, should primarily mean that both direct users and sponsors are capable of communicating their needs to specialists and understand the jargon of the latter to the extent that blending of the two types of knowledge may occur, especially in the modelling phase of problem-solving procedures. Mastering the basics of a common dictionary is the first necessary condition to achieve superior goals such as ‘‘the art of reasoning logically with formal models’’ (compare [17, p. 62]). Let us sketch some partial – and perhaps also biased by our own experiences – answers to the question, what could be other core elements of OR-literacy. I did not even try to be exhaustive and the order of the following items does not necessarily correspond to the chronological one, both in the sense of consecutive phases in practical problem-solving procedures and the necessary structure of underlying education. First of all, anybody involved in application of quantitative methods should understand the complexity of business problems. Responsibility for achieving this goal is partially out of our ORcommunity, but there still remains the understanding of systems thinking, interdisciplinary
approach and all the other ‘‘global notions’’, preserving the genuine characteristics of OR/MS (see [1,3,7]). Both (all) parts included should be aware of the structure of formalized problem-solving processes (compare classical graphical presentations, used to describe it, e.g. in [2,19], and other textbooks). The emphases can be different for different ‘‘pupils’’, but always built upon the general structure, which must remain unique in its entirety. The roles and responsibilities of the three different types of participants must be defined clearly, but with an anticipation of the possible changes in their structure. The already mentioned dictionary problem with certainty appears within populations with different professional backgrounds and such ‘‘mixtures’’ are most typical in post-graduate managerial education. The lecturer must be ready to interpret from one meta-language to another if necessary. Since we cannot master the secrets of many different professions, examples and cases should be used which are accessible at the general education bases. Sticking to an appropriate form of the well-known ‘‘KISS principle’’ is therefore generally recommended. Connected to the former question is the classical dilemma whether to work on simplified ‘‘school cases’’ or ‘‘real-life cases’’. The answer cannot be completely unambiguous and perhaps the best way is to take a ‘‘school case’’ to show how things should work, than use a real-life case and show what problems we may encounter in practice. The additional advantage of such an approach is that a fruitful debate about the relation ‘‘real-life’’ – model – data can be started afterwards. Some ‘‘early warning system’’ during education process is necessary if we want to get rid of disappointments due to the data quality problem in practice. In my opinion, real-life problems with acquisition of reliable, unbiased data should be discussed in any OR course, both for professionals and users: the former will learn about the dirty details far beyond the beautifully prepared data in textbook exercises, the latter will get (more) ready to take care of the data themselves. Grossman [11] speaks about ‘‘acquisition, management, and cleaning of the data’’ and repeats O’Keefe’s
ibej / European Journal of Operational Research 140 (2002) 225–231 J.A. C
warning (‘‘MS/OR as portrayed in textbooks assumes that any data requirements can be met; in practice this is patently ridiculous’’); Powell [17] mentions modelling of the data in the framework of (advanced) modelling skills. All this opens questions about interdisciplinary contacts in the circle, encompassing at least OR/MS, statistics, accounting and information systems, both in specific education programmes and on the large scale. User should be able to articulate a clear picture of their goals and possibly a set of appropriate criteria for evaluation. Profit optimization is without doubt the crucial term included in many objective functions and the most common ‘‘driver’’ of the majority of people’s endeavours. This is especially important for the managers’ part of participation and becomes therefore immanent to their cognitive structure. Generally, the term ‘‘profit’’ should be understood in the broadest sense, starting close to ‘‘survival’’ and (e.g. ecologically) sustainable growth, finishing at the standard Pareto optimality concept, and encompassing categories that are classical cases of the so-called second-best solutions or satisfactory behaviour approach in the terms of Herbert Simon. Speaking of him, it is always useful to raise a question about the general approach to optimization, or rather, whether we are ready and capable of infecting users with a clear awareness of the fact that multicriteria approach is as a starting point more natural (taking into account the possibility of collision of criteria, of course) than the classical mono-criteria optimization. One part of the problem is rather technical, but at least in some countries one of the important obstacles is that multi-criteria approach can collide with the basics of the traditional one-way reasoning and decision-making. Both direct users and sponsors should have a general overview of the available OR tools; users would probably care to have a better understanding of those tools, specific for their field of work. We will not go into details about the contents, since the structure of textbooks is almost standardized (compare the widely used ones like [2,12,19,24], etc.), but rather emphasize that the organization of the existing textbooks – no matter how good they are in our own opinion – around different groups of methods better suits education of specialists than
229
that of users. After all, the information for the latter should focus on what can be done by intelligent application of formalized modelling, not on how it is actually done. An advice from [4], i.e. ‘‘Save greater technological expertise for those students who choose to pursue it.’’, seems quite reasonable (and close to what I try to do as a teacher . . . not always successfully). To recapitulate, no matter how it is actually presented, the review of methods must be sufficiently exhaustive, but also non-technical, and first of all – very honest: we should not promise anything that cannot be achieved in practical problem-solving situations. In the long run, such void promises can ruin OR. One crucial question in OR education (not only for users!) is how to introduce stochastic phenomena. Reading reports on different types of ‘‘innumeracy’’ in the USA and other countries strengthens the impression that problems in understanding any approach beyond classical causal models are quite common. We can agree that a sound course on probability theory and stochastic processes is essential and inevitable for specialists. What to do with users and sponsors? Two most important elements must be provided: (a) understanding of probability in terms of an ‘‘idealized relative frequency’’; (b) understanding the notion of expected values and the subjectivity of decisionmaking based on them, when the number of repetitions of the experiment is not large enough (which means almost always in practical business situations). I had very good results in increasing the level of understanding if computer simulations were used to compare the ‘‘order’’ in large series of experiments and possible ‘‘surprises’’ when only some repetitions were produced. Spreadsheet models are nowadays perhaps the most popular software tools in OR, or rather in OR teaching, since at least some add-ins are often needed for professional applications. There are some beautiful teaching cases available on the Internet and a special issue of INFORMS Transactions on Education was devoted to them (compare [16], then [4,18,22]). In spite of some hesitations concerning possible decay of algebraic methods, spreadsheets seem to be in the first place the common language all managers can understand and appreciate; even if calculations are done in the
230
ibej / European Journal of Operational Research 140 (2002) 225–231 J.A. C
‘‘back-office’’, spreadsheets can still be applied as user-friendly interfaces. Calculation by hand has of course become an obsolete teaching tool, at least for students who do not need to know the details of algorithms. Computer is indispensable especially for methods, requiring a lot of iterations (like Monte Carlo simulation methods, e.g.); in such cases it is not just a fast calculation machine, but also enables dynamic presentation of convergence to the solution. There is, however, always a tactical question whether to show a live simulation – risking to get something that will not demonstrate what we wanted to show – or (good, best) results of our former endeavours. ‘‘Honesty is the best policy’’: use live simulations, in the case of an ‘‘unfortunate’’ sample explain the statistical background of this event. . . and have some nice ‘‘static’’ pictures ready. Multimedia approach in presentations is generally very useful and attractive, especially for the non-technical parts devoted to sponsors and direct users. In terms of ‘‘honesty’’ mentioned before we should add ‘‘Yes, but not too smoothly, please!’’. The beauty of the presentation should not hide the weak points of the methods themselves. Last, but not least: if some kind of mimicry does the trick, just use it to motivate people. Sometimes it is enough to carefully select the name of the course or seminar to attract people. Most people hate ‘‘mathematics’’, ‘‘quantitative methods’’ is a lot better name, labelling it ‘‘profitable decision-making’’ is a direct hit.
4. Conclusion There are people who will probably find some of my remarks too ‘‘commercial’’ to be accepted by the scientific community. It may be so, but we live in a world of smart marketing. The analyses of the current state (like [7,11]) indicate that we are losing our customers, so far mainly in the area of education, but delayed consequences can appear sooner or later. One of the main reasons for this is obviously bad positioning of our product: we still try to sell T-bone steaks to vegetarians. Either the product or the customers must change. Anyway, the customer is always right.
Acknowledgements An earlier version of this paper was presented as a semiplenary lecture at the 17th European Conference on Operations Research, Budapest 2000; I wish to thank the chairman (Professor J.P. Brans) and other participants for their encouraging comments. Two anonymous referees substantially helped me in improving the paper and I wish to express my sincere gratitude for their valuable suggestions. References [1] R.L. Ackoff, Science in the systems age: Beyond IE, OR, and MS, Operations Research 21 (3) (1973) 661–671. [2] D.R. Anderson, D.J. Sweeney, T.A. Williams, An Introduction to Management Science, sixth ed., West Publishing Co, St. Paul, 1991. [3] B.A. Bayraktar et al. (Eds.), Education in Systems Science: Report and Proceedings of the NATO Advanced Research Institute, Taylor & Francis, London, 1979. [4] R.L. Carraway, D.R. Clyman, Integrating spreadsheets into a case-based MBA quantitative methods course: Real managers make real decisions, INFORMS Transactions on Education 1 (1) (2000) 38–46. Available from. [5] C.W. Churchman, R.L. Ackoff, E.L. Arnoff, in: Introduction to Operations Research, Wiley, New York, 1957, 645 pp. [6] C.W. Churchman, A.H. Scheinblatt, The researcher and the manager. A dialectic of implementation, Management Science 11 (4) (1965) B69–B87. [7] N. Cummings, In depth report on OR 42. http://www.orsoc.org.uk/conf/or42/OR42fullreport.htm. [8] K.S. Dhir, A judgment-analytical approach to understanding models. Abstracts. OR in the 21st Century: Communications & Knowledge Management, 12–14 September 2000, University of Wales, Swansea. http://www.orsoc.org.uk/conf/or42/streams/simulation.htm. [9] S. Eom, The contributions of systems science to the development of the decision support system research subspecialties: An empirical investigation, Systems Research and Behavioral Science 17 (2000) 117–134. [10] E. Erkut, Editorial. INFORMS Transactions on Education 1 (1) (2000). http://ite.informs.org/Vol1No1/others/ Editorial.html. [11] T.A. Grossman, Causes of the decline of the business school management science course, INFORMS Transactions on Education 1 (2) (2001). Available from . [12] R.I. Levin, D.S. Rubin, J.P. Stinson, E.S. Gardner Jr., Quantitative Approaches to Management, eight ed., McGraw-Hill, New York, 1992.
ibej / European Journal of Operational Research 140 (2002) 225–231 J.A. C [13] G.M. Marakas, J.J. Elam, Semantic structuring in analyst acquisition and representation of facts in requirements analysis, Information Systems Research 9 (1) (1998) 37–63. [14] H. Mintzberg, Planning on the left side and managing on the right, Harvard Business Review 4 (July–August) (1976) 49–58. [15] P.M. Morse, G.E. Kimball, Methods of Operations Research. First Edition Revised. New York, Published jointly by the Technology Press of MIT and Wiley, 1951, p. 158. [16] S.G. Powell, Introduction to the special issue: teaching management science with spreadsheets workshop, INFORMS Transactions on Education 1 (1) (2001) 2–3. Available from. [17] S.G. Powell, Teaching modeling in management science, INFORMS Transactions on Education 1 (2) (2001) 62–67. Available from . [18] C.T. Ragsdale, Teaching management science with spreadsheets: from decision models to decision support, INFORMS Transactions on Education 1 (2) (2001) 68–74.
[19] [20]
[21]
[22]
[23] [24] [25]
231
Available from. B. Render, R.M. Stair Jr., Quantitative Analysis for Management, Allyn and Bacon, Boston, 1988. V. Rupnik, On the 30th Anniversary of OR in Slovenia, in: V. Rupnik, L. Bogataj (Eds.), Proceedings of SOR ’93. SDI – SOR, Ljubljana, 1993, pp. 3–16 (Slovenian). S. Sarker, Toward a methodology for managing information systems implementation: A social constructivist perspective, Informing Science 3 (4) (2000) 195–205. S. Savage, Blitzograms – interactive histograms, INFORMS Transactions on Education 1 (2) (2001) 77–87. Available from . A. Vazsony, in: Scientific Programming in Business and Industry, Wiley, New York, 1958, 474 pp. W.L. Winston, Operations Research. Applications and Algorithms, Duxbury Press, Belmont, 1994. J. Zohil, Some problems of OR in Slovenia, in: V. Rupnik, M. Bogataj (Eds.), Proceedings of SOR ’94. SDI – SOR, Ljubljana, 1994, pp. 243–248 (Slovenian).
Operations research education for forgotten populations ibej Joze Andrej C Faculty of Economics, Ljubljana University, Kardeljeva pl. 17, SI-1000 Ljubljana, Slovenia
Abstract Typical OR education programmes are focused on those who should professionally act as ‘‘performers’’ (specialists, modellers, etc.) in the process of business problems solving through quantitative modelling and similar formalized procedures. The problem of unsatisfactory level of understanding between ‘‘specialists’’ and ‘‘generalists’’ (managers, e.g.) has been known for decades; yet we feel that at least the latter (‘‘users’’ and ‘‘sponsors’’ as their subspecies) are still not adequately educated and this fact helps the gap to survive. We discuss different views of this problem and advocate for some approaches, mainly based upon our own pedagogical experiences. Ó 2002 Elsevier Science B.V. All rights reserved. Keywords: OR education
1. Brutal facts President of one of the most important professional OR organizations recently mentioned (see [7]) that close-to-extinction of operations research in some parts of the world was mainly caused by the academics who had paid too much emphasis on technicalities, thus allowing managers and other potential users to interpret OR as pure ‘‘mathematical masturbation’’. Business student proverb ‘‘Magical Science is a waste of time’’, cited in [11] as a negation of our faith in the usefulness of management science, somehow completes the picture. Since Ackoff’s masterpiece [1], these are probably two of the most direct warnings to the OR/ MS community that its main task should rather be co-operation with users on ‘‘grand plan’’ than
ibej). E-mail address: [email protected] (J.A. C
(merely) production of sophisticated mathematical models and algorithms, overloaded with details which can usually not be properly supported with reliable data. During the last years, there have been many other calls for resurrection of the more traditional, i.e. integral, contents-oriented, on the whole less technical and in the first place strictly interdisciplinary approach to OR/MS. The revival of classical sources (like [5,15,23], and many others) is obvious too. Yet there is probably a long way to go to bring OR/MS specialists and managers closer again. Some basic obstacles could be the following: (a) A crucial part of the problem is the fact that the roles each of the parts should play in efficient problem solving are often not precisely defined in advance; sometimes it is not even clear how different the roles and corresponding responsibilities actually are.
0377-2217/02/$ - see front matter Ó 2002 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 7 - 2 2 1 7 ( 0 2 ) 0 0 0 6 5 - 6
226
ibej / European Journal of Operational Research 140 (2002) 225–231 J.A. C
(b) Players are not properly trained for their roles, which is especially true if we consider the users’ inability to participate as active members of interdisciplinary teams. (c) The current level of communication and mutual understanding between the two groups is generally rather low. Discussing the evidence in detail could mean reinventing hot water. Churchman and Schainblatt (compare [6]) were among the first to recognize the dichotomy between researchers and managers. Although more focused on the implementation phase, they have already unveiled the most important common reasons for confrontation between them (see [8,9,13,21]), and also initiated a more general discussion, which sometimes goes all the way down to the basic differences in cognitive characteristics of both groups [14]. One of the most striking differences, namely the ‘‘language problem’’, has been evident in the OR literature for more than 40 years (and in OR practice, probably always), and some authors have systematically tried to overcome it. Vazsony’s classical OR textbook [23] even exposes it in its ‘‘mission statement’’: ‘‘It is the first attempt to develop a mathematical language which can be understood by businessmen using scientific techniques (OR, LP; statistical decision theory, etc.). The book is written for management personnel, with scientific techniques explained, not in terms of mathematics, but in terms of business’’. I believe that adequate – and in some sense more balanced – OR education for users could help a lot in achieving the necessary condition for success: enabling users to act as providers of an important part of the interdisciplinary knowledge. This does not mean that the education of OR (MS, IS, etc.) specialists is perfect, yet the users’ side still seems to be the weakest link in the chain. Situation may be different from country to country, but recent publications in INFORMS Transactions on Education [10,11,17] and other sources confirm that there is a common core of problems connected with the education of non-specialists. The crucial part of such education takes place as OR/ MS courses within (undergraduate) business school curricula, and in the first place in postgraduate MBA programmes. The importance of
OR/MS courses for technical professions and functional education for people from practice should also not be underestimated, but the main emphasis has always been on students of business. This paper focuses on the question how to design and implement OR/MS education for all those we like to call forgotten populations, i.e. users as a whole and sponsors as their subspecies. To be on the safe side, let us say that we understand the word ‘‘sponsors’’ as it usually applies in the context of IT projects, mainly in information systems development. Sponsors can be general managers, CEOs, members of the board and other persons responsible in the first place for the strategic level of decision-making. Compared to the involvement of other users, sponsors’ controls are more or less of indirect and strategic type. On the other hand, sponsors are directly involved in the formalized problem solving procedures at least at the very beginning (they have to pull the trigger and allow the process to start) and during their final phase, thus sharing the responsibility for the project. It was empirically shown that more than 90% of projects remained unaccomplished without active support of the top management in the firm. Last, but not least, sponsors must do what their name promises: pay. Typically, potential ‘‘direct users’’ of OR in firms are either decision-makers at the operational level, or planners, controllers, accountants and other ‘‘cost-cutters’’ at the tactical level. They are usually the first to encounter the symptoms of a problem situation and are supposed to be able to define the problem in exact terms. If OR/MS methods are used in the firm, these people are generally directly involved in the modelling process, data providing and/or practical implementation of solutions. We would not go that far to imply different cognitive characteristics for these groups, but there is another functional difference that makes us sometimes distinguish between sponsors and ‘‘direct users’’. While the latter may also achieve instantaneous tangible direct profits from the application of OR/MS methods through increased productivity of their individual work and consequentially higher earnings, the former can only get enthusiastic if one can make them believe that the
ibej / European Journal of Operational Research 140 (2002) 225–231 J.A. C
benefits of quantitative methods applications fit in the general managerial framework and contribute to the overall ‘‘bottom line results’’ of the firm. Therefore the motivation structure can surely be significantly different, causing possibly important differences in propensity towards OR/MS projects. However, the main gap remains between ‘‘users’’ as a whole and ‘‘providers’’, or simply between generalists and specialists in the classical meaning of these terms.
2. Teaching OR/MS: Two different directions We can only agree with Erkut’s editorial [10], that there can be a wide variety of personal research interests, but almost everybody involved in OR/MS does some teaching, either on a regular basis or at least occasionally and informally, through seminars and consultancy. Even the most enthusiastic teachers are generally keener to present methods, algorithms, and tricks of the trade to once-to-be specialists in OR/MS than to educate potential users, people whose attitude towards mathematics, statistics, OR and quantitative methods in common is as a rule rather hostile than friendly (Slovenia is not an exception, see [20,25]). The underestimation of this fact seems to be just one of the most frequent causes of misunderstandings and non-optimal performance of OR/ MS, both in practice and education. Several authors have already noticed very concrete educational problems (problems from practice are as old as OR/MS itself). One of the best papers on this subject is Grossman’s alarming report [11]. Our point of view – that there is a need for specific approaches if users should be properly educated to share our faith in OR – is best supported by author’s recognition of the fact that even the cases presented in the classroom as ‘‘management science success stories’’, often reduce to something, practised rather by OR/MS specialists than by managers. Therefore it is very naive to expect that students will be able to find connections between what we want them to learn and the general management practice they are motivated to learn through such examples. Grossman also unveils a long list of typical pedagogical ‘‘sins’’ in which
227
every OR teacher can definitely find some from his or her own ‘‘black magic box’’. So it is rather obvious that although the role of ‘‘users’’ as participants in projects is at least briefly mentioned in modern OR/MS textbooks (like [2,12,19,24], . . .), the OR professional community has so far underestimated the need for a clear separation of two types of education for two (or even three) groups. Even the (over) simplified description of the goals that should be attained through education process adds much argument for a complete restructuring of educational programmes: sponsors should understand the needs of users, having in mind the power and range of OR/MS methods. Users should be in the first place taught how to express the needs and how to co-operate with specialists. Specialists should be able to fulfill the needs. Of course, the extent of OR/MS knowledge needed to accomplish these three tasks increases in the direction from sponsors towards specialists. But there are other elements in which a specialist will never be able to compete with neither sponsors nor users. One does not have to go back to Ackoff and other classics of the OR literature to realize what makes teamwork not just important, but practically inevitable: one failure in practice will do the trick. Some kind of blending of qualitative knowledge (mostly held by users) with the more quantitatively oriented segments of knowledge (which specialists in OR/MS must provide) must be assured.
3. OR literacy The fact that users contribute their specific problem-oriented knowledge does not mean that they can be completely OR-illiterate if we hope for teamwork to be successful. My direct answer to a rather provoking question why I advocate for ‘‘OR literacy’’ for users instead of ‘‘problem literacy’’ for OR professionals is very simple. I would of course expect that an OR/MS specialist who has been working for several years in one industry has gradually acquired a fair share of ‘‘problem literacy’’. But at the same time, I believe that – especially during the phase of pre-career
228
ibej / European Journal of Operational Research 140 (2002) 225–231 J.A. C
education – it is still easier to achieve some general ‘‘OR literacy’’ in the population of potential users than to teach potential specialists a lot of different disciplines. Larger ‘‘OR communities’’ may perhaps afford an earlier problem-oriented specialization, but this is generally not the case that the rest of the world is dealing with. Speaking in terms of practice it is clear and generally recognized that a successful project cannot occur otherwise but as a net of people contributing specific quanta of knowledge. On the other hand, attempts to teach everybody everything have not disappeared from our education process. The results of such an approach are usually just the opposite of the expected ones. My usual saying that students then prefer ‘‘nothing’’ to ‘‘too-muchand-unnecessary’’ may seem exaggerated, but is more and more supported by practical experiences and other evidence (compare [7,11,17] and other sources). Defining exact meaning of the word ‘‘literacy’’ has never been a simple task. In my opinion, ORliteracy, as pleaded for in this paper, should primarily mean that both direct users and sponsors are capable of communicating their needs to specialists and understand the jargon of the latter to the extent that blending of the two types of knowledge may occur, especially in the modelling phase of problem-solving procedures. Mastering the basics of a common dictionary is the first necessary condition to achieve superior goals such as ‘‘the art of reasoning logically with formal models’’ (compare [17, p. 62]). Let us sketch some partial – and perhaps also biased by our own experiences – answers to the question, what could be other core elements of OR-literacy. I did not even try to be exhaustive and the order of the following items does not necessarily correspond to the chronological one, both in the sense of consecutive phases in practical problem-solving procedures and the necessary structure of underlying education. First of all, anybody involved in application of quantitative methods should understand the complexity of business problems. Responsibility for achieving this goal is partially out of our ORcommunity, but there still remains the understanding of systems thinking, interdisciplinary
approach and all the other ‘‘global notions’’, preserving the genuine characteristics of OR/MS (see [1,3,7]). Both (all) parts included should be aware of the structure of formalized problem-solving processes (compare classical graphical presentations, used to describe it, e.g. in [2,19], and other textbooks). The emphases can be different for different ‘‘pupils’’, but always built upon the general structure, which must remain unique in its entirety. The roles and responsibilities of the three different types of participants must be defined clearly, but with an anticipation of the possible changes in their structure. The already mentioned dictionary problem with certainty appears within populations with different professional backgrounds and such ‘‘mixtures’’ are most typical in post-graduate managerial education. The lecturer must be ready to interpret from one meta-language to another if necessary. Since we cannot master the secrets of many different professions, examples and cases should be used which are accessible at the general education bases. Sticking to an appropriate form of the well-known ‘‘KISS principle’’ is therefore generally recommended. Connected to the former question is the classical dilemma whether to work on simplified ‘‘school cases’’ or ‘‘real-life cases’’. The answer cannot be completely unambiguous and perhaps the best way is to take a ‘‘school case’’ to show how things should work, than use a real-life case and show what problems we may encounter in practice. The additional advantage of such an approach is that a fruitful debate about the relation ‘‘real-life’’ – model – data can be started afterwards. Some ‘‘early warning system’’ during education process is necessary if we want to get rid of disappointments due to the data quality problem in practice. In my opinion, real-life problems with acquisition of reliable, unbiased data should be discussed in any OR course, both for professionals and users: the former will learn about the dirty details far beyond the beautifully prepared data in textbook exercises, the latter will get (more) ready to take care of the data themselves. Grossman [11] speaks about ‘‘acquisition, management, and cleaning of the data’’ and repeats O’Keefe’s
ibej / European Journal of Operational Research 140 (2002) 225–231 J.A. C
warning (‘‘MS/OR as portrayed in textbooks assumes that any data requirements can be met; in practice this is patently ridiculous’’); Powell [17] mentions modelling of the data in the framework of (advanced) modelling skills. All this opens questions about interdisciplinary contacts in the circle, encompassing at least OR/MS, statistics, accounting and information systems, both in specific education programmes and on the large scale. User should be able to articulate a clear picture of their goals and possibly a set of appropriate criteria for evaluation. Profit optimization is without doubt the crucial term included in many objective functions and the most common ‘‘driver’’ of the majority of people’s endeavours. This is especially important for the managers’ part of participation and becomes therefore immanent to their cognitive structure. Generally, the term ‘‘profit’’ should be understood in the broadest sense, starting close to ‘‘survival’’ and (e.g. ecologically) sustainable growth, finishing at the standard Pareto optimality concept, and encompassing categories that are classical cases of the so-called second-best solutions or satisfactory behaviour approach in the terms of Herbert Simon. Speaking of him, it is always useful to raise a question about the general approach to optimization, or rather, whether we are ready and capable of infecting users with a clear awareness of the fact that multicriteria approach is as a starting point more natural (taking into account the possibility of collision of criteria, of course) than the classical mono-criteria optimization. One part of the problem is rather technical, but at least in some countries one of the important obstacles is that multi-criteria approach can collide with the basics of the traditional one-way reasoning and decision-making. Both direct users and sponsors should have a general overview of the available OR tools; users would probably care to have a better understanding of those tools, specific for their field of work. We will not go into details about the contents, since the structure of textbooks is almost standardized (compare the widely used ones like [2,12,19,24], etc.), but rather emphasize that the organization of the existing textbooks – no matter how good they are in our own opinion – around different groups of methods better suits education of specialists than
229
that of users. After all, the information for the latter should focus on what can be done by intelligent application of formalized modelling, not on how it is actually done. An advice from [4], i.e. ‘‘Save greater technological expertise for those students who choose to pursue it.’’, seems quite reasonable (and close to what I try to do as a teacher . . . not always successfully). To recapitulate, no matter how it is actually presented, the review of methods must be sufficiently exhaustive, but also non-technical, and first of all – very honest: we should not promise anything that cannot be achieved in practical problem-solving situations. In the long run, such void promises can ruin OR. One crucial question in OR education (not only for users!) is how to introduce stochastic phenomena. Reading reports on different types of ‘‘innumeracy’’ in the USA and other countries strengthens the impression that problems in understanding any approach beyond classical causal models are quite common. We can agree that a sound course on probability theory and stochastic processes is essential and inevitable for specialists. What to do with users and sponsors? Two most important elements must be provided: (a) understanding of probability in terms of an ‘‘idealized relative frequency’’; (b) understanding the notion of expected values and the subjectivity of decisionmaking based on them, when the number of repetitions of the experiment is not large enough (which means almost always in practical business situations). I had very good results in increasing the level of understanding if computer simulations were used to compare the ‘‘order’’ in large series of experiments and possible ‘‘surprises’’ when only some repetitions were produced. Spreadsheet models are nowadays perhaps the most popular software tools in OR, or rather in OR teaching, since at least some add-ins are often needed for professional applications. There are some beautiful teaching cases available on the Internet and a special issue of INFORMS Transactions on Education was devoted to them (compare [16], then [4,18,22]). In spite of some hesitations concerning possible decay of algebraic methods, spreadsheets seem to be in the first place the common language all managers can understand and appreciate; even if calculations are done in the
230
ibej / European Journal of Operational Research 140 (2002) 225–231 J.A. C
‘‘back-office’’, spreadsheets can still be applied as user-friendly interfaces. Calculation by hand has of course become an obsolete teaching tool, at least for students who do not need to know the details of algorithms. Computer is indispensable especially for methods, requiring a lot of iterations (like Monte Carlo simulation methods, e.g.); in such cases it is not just a fast calculation machine, but also enables dynamic presentation of convergence to the solution. There is, however, always a tactical question whether to show a live simulation – risking to get something that will not demonstrate what we wanted to show – or (good, best) results of our former endeavours. ‘‘Honesty is the best policy’’: use live simulations, in the case of an ‘‘unfortunate’’ sample explain the statistical background of this event. . . and have some nice ‘‘static’’ pictures ready. Multimedia approach in presentations is generally very useful and attractive, especially for the non-technical parts devoted to sponsors and direct users. In terms of ‘‘honesty’’ mentioned before we should add ‘‘Yes, but not too smoothly, please!’’. The beauty of the presentation should not hide the weak points of the methods themselves. Last, but not least: if some kind of mimicry does the trick, just use it to motivate people. Sometimes it is enough to carefully select the name of the course or seminar to attract people. Most people hate ‘‘mathematics’’, ‘‘quantitative methods’’ is a lot better name, labelling it ‘‘profitable decision-making’’ is a direct hit.
4. Conclusion There are people who will probably find some of my remarks too ‘‘commercial’’ to be accepted by the scientific community. It may be so, but we live in a world of smart marketing. The analyses of the current state (like [7,11]) indicate that we are losing our customers, so far mainly in the area of education, but delayed consequences can appear sooner or later. One of the main reasons for this is obviously bad positioning of our product: we still try to sell T-bone steaks to vegetarians. Either the product or the customers must change. Anyway, the customer is always right.
Acknowledgements An earlier version of this paper was presented as a semiplenary lecture at the 17th European Conference on Operations Research, Budapest 2000; I wish to thank the chairman (Professor J.P. Brans) and other participants for their encouraging comments. Two anonymous referees substantially helped me in improving the paper and I wish to express my sincere gratitude for their valuable suggestions. References [1] R.L. Ackoff, Science in the systems age: Beyond IE, OR, and MS, Operations Research 21 (3) (1973) 661–671. [2] D.R. Anderson, D.J. Sweeney, T.A. Williams, An Introduction to Management Science, sixth ed., West Publishing Co, St. Paul, 1991. [3] B.A. Bayraktar et al. (Eds.), Education in Systems Science: Report and Proceedings of the NATO Advanced Research Institute, Taylor & Francis, London, 1979. [4] R.L. Carraway, D.R. Clyman, Integrating spreadsheets into a case-based MBA quantitative methods course: Real managers make real decisions, INFORMS Transactions on Education 1 (1) (2000) 38–46. Available from
ibej / European Journal of Operational Research 140 (2002) 225–231 J.A. C [13] G.M. Marakas, J.J. Elam, Semantic structuring in analyst acquisition and representation of facts in requirements analysis, Information Systems Research 9 (1) (1998) 37–63. [14] H. Mintzberg, Planning on the left side and managing on the right, Harvard Business Review 4 (July–August) (1976) 49–58. [15] P.M. Morse, G.E. Kimball, Methods of Operations Research. First Edition Revised. New York, Published jointly by the Technology Press of MIT and Wiley, 1951, p. 158. [16] S.G. Powell, Introduction to the special issue: teaching management science with spreadsheets workshop, INFORMS Transactions on Education 1 (1) (2001) 2–3. Available from
[19] [20]
[21]
[22]
[23] [24] [25]
231
Available from