Looking Back on a Framework for Thinking About Group Support Systems
This chapter is an update to the thinking framework for Group Decision Support Systems (GDSS) proposed by Colin Eden 30 years ago. As the source paper, this chapter is a personal take on the topic; however, it is a personal take rooted in substantial experience in the broad area of decision-making and modelling and in some specific narrow areas of decision support. There have been major developments in the broad context surrounding GDSS, including the improved understanding of decisions on the conceptual side, and many aspects of computer development, such as artificial intelligence and big data on the technical side. Considering the volume of these changes, it is surprising how much the observations, arguments, and conclusions offered in the source paper are still valid today. The most important component of any GDSS is still the facilitator, and the most valuable ingredients of the GDSS process are the participants’ intuitions, creativity, opinions, arguments, agendas, personalities, and networks. The outcome of the GDSS process is only valuable if it is politically feasible. Today we have a better understanding of transitional objects and their role in the GDSS process; their significance is the second after the facilitator. Artificial intelligence can be useful for GDSS in several different ways, but it cannot replace the facilitator.
KeywordsPolitical feasibility Transitional object Artificial intelligence Intuition Creativity Emotions
This chapter has the same title as a paper published by Colin Eden, nearly 30 years ago, in Group Decision and Negotiation (Eden 1992a). In that paper, Eden showcased the framework he developed for thinking about Group Decision Support Systems (GDSS). The purpose of this chapter is to revisit the topic explored by Eden and explore whether the claims made in that paper still make sense. To this end, I examine whether the scope of validity has changed and consider the shifts in trends identified in the original paper. In other words, this chapter offers an updated version of Eden’s framework for thinking about GDSS for the age of big data and artificial intelligence. I pick up where Eden left off, and work my way backwards through Eden’s paper, indicating the changes I have observed in the field of GDSS over the last three decades.
There is no agreement in the broad area of decision support and operational research about whether a particular tool, method, or approach is a GDSS. For instance, Eden noted that while he regarded Soft System Methodology (Checkland 1999; Checkland and Scholes 1999) a GDSS, he was not sure whether Peter Checkland (the originator) would accept the label.
There is no common set of objectives for all GDSSs. However, supportive of pluralism of GDSS, Eden (1992a: 214) emphasized that “it is not important to agree the purposes of GDSSs but rather that the designer be explicit about them in each individual case.”
The conceptual underpinnings of different GDSS are different and are worked out at varying levels of sophistication. Some, such as SSM (Soft System Methodology) or SODA (Strategic Options Development and Analysis), are built from explicit philosophical and conceptual basis, while others, such as Group Systems (see “Group Support Systems: Past, Present and Future”), ignore the conceptual level altogether and focus on technicalities instead.
There is no agreement about what GDSS stands for as a technical term. For instance, Eden (see also Eden 1990) notes the typically American idiosyncrasy of the term GDSS being reserved exclusively for systems that support groups with dedicated computer hardware and software, excluding many solutions developed elsewhere.
These reasons also imply that comparing GDSSs to each other in an attempt to figure out which one is generally the best is both hopeless and useless. So, what is the real purpose of a thinking framework for GDSS? As I see it, the primary purpose is to bring some order to a messy field and provide markers for orientation to scholars and practitioners who engage with the field and want to use GDSS. The secondary purpose, not less important, but unsound without the primary one, is to assist GDSS users in figuring out what distinctive benefits a particular GDSS can bring to addressing a specific problem situation – even if it may not be possible to assess whether one GDSS is better than another.
In what follows, I first revisit the three approaches to considering the success of GDSS, which constituted the outcome of Eden’s original exploration of the topic. My conclusion is that the three suggested approaches, namely controlled experiments, comparing GDSS to their underlying conceptual backgrounds and asking the user, are as valid now as they were at the time when they were introduced. Next, I get back to Eden’s consideration of the roles and significance of computers in GDSS. There is one, relatively small change in this area, the development of virtual reality, which may have significant consequences in the future. Then, I look into the conceptualizations of decision-making that underline GDSS. Subsequently, I revisit the dimensions along which Eden conducted his analysis. In each of these dimensions, I will look into what has changed and how. The source article was one of the earliest mentions of “transitional objects” in the GDSS context; I assign a more prominent role to this concept, as its significance has substantially increased since the publication of the source article. Finally, I offer a personal view on what role(s) big data (BD) and artificial intelligence (AI) may play in the future of GDSSs.
Approaches for Considering Success of GDSS
It is perhaps trivial, today, to say that GDSSs are necessarily complex. This was less obvious 30 years ago, but Eden (1992a: 212) offered a sound argument, based on the assumption that “a GDSS is only likely to be economically viable when used to support ill-structured, complex, and probably strategic1 decision making.” This assumption, in turn, derives from the author’s experience and that of others using GDSS in real-world settings with real clients. I believe that this assumption is as valid today as it was 30 years ago. Then, drawing on Ashby’s Law of Requisite Variety, Eden suggested that in order to provide an adequate support in such complex situations, “GDSSs are, and will be, a complex system of computer hardware, computer software, procedures, environments, and facilitation in a mix of proportions” (ibid.: 212). I believe that they still are and they will be for the foreseeable future. This is one of the trends that has kept the same direction, perhaps has become even more forceful. I make a few comments later on how the recent developments in big data (BD) and artificial intelligence (AI) relate to this and what role they may play in GDSS.
When I went to school, they asked me what I wanted to be when I grew up.I wrote down ‘happy’. They told me I didn’t understand the assignment,and I told them they didn’t understand life.John Lennon (1940–1980)
This does not mean that such controlled experiments cannot be useful, only that this usefulness is limited to a specific aspect(s) of GDSS, namely to understand better the micro-characteristics of the designed GDSS and of the conceptual models underpinning the design. With the development of technology, much of validation through controlled experiments can be automated using simulations. However, the judgment cannot be fully automated, as it requires an understanding of the conceptual background and of the decision situations which, however simplified, may still involve a degree of complexity beyond the machines’ capability. Regardless of the degree of automation, this approach will only help to make sure that the GDSS is consistent.
If the system is designed specifically to address real groups (with a history and a future) working on complex issues, then it is no use taking out those very characteristics that make it complex in order to control experiments. Research with students using structured problems will say absolutely nothing about the performance of a GDSS in relation to its designed aims.
The second (not in order) type of evaluation discussed in the source paper is comparing the designed GDSS with the conceptual background that was used to determine the design. As explored later in more detail, there are many different, mutually incompatible conceptualizations of decisions. These different conceptualizations do not even work with the same concepts as building elements; what is central to one may be regarded non-existing by another, and the same concepts may carry different meanings in different models. Therefore, one of the meaningful questions to ask about GDSS is how well it reflects the conceptual model(s) it is based on. (If there is more than one model, these also need to be compatible with each other; ultimately it should be possible, at least in principle, to synthesize these into a single model.) To achieve some clarity in this regard, Eden (1992a: 213–214) urged GDSS designers to be explicit about “the nature of group decision making as a process; the nature of decision making in organizations (including the nature of problems, problem solving/alleviation/finishing, and of implementation); and the nature of support and intervention by a ‘system,’ be it facilitator/chauffeur/consultant/ software tool in relation to a group.” If these were not made explicit, we would not know what to compare the designed GDSS to at this level of validation. I believe that this level of evaluation gained in significance over the past 30 years, particularly as the conceptualizations of decision-making have multiplied. Furthermore, I believe that the significance of this type of validation is not limited to GDSS field but would be sorely needed in all areas of modelling. In all areas of management and organization studies, we find a multitude of mutually incompatible models (and unexamined compatibility does not imply compatibility), and these are often used without critical examination. But why would we question whether the designed GDSS (or any model or solution) reflects the conceptual model that determined that design? The short answer would be experience. Because we have all seen models, including GDSS, that, by the time they were ready, got in contradiction with the conceptual underpinnings on which they were based. Daniel Dennett (1995: 21) said once that “there is no such thing as philosophy-free science; there is only science whose philosophical baggage is taken on board without examination.” In a similar manner, there is a significant danger that some conceptual baggage is taken on, without examination, resulting in a model that is self-contradictory. Although a particular GDSS may not account for every single feature of its conceptual background, it must be fully in harmony with it, otherwise we cannot use it having that conceptual framework on mind. In other words, the GDSS must be relevant to the conceptual background that informed its design.
Finally, the third approach for considering success brings GDSS to the people and situation to which it is applied and asking whether it is applicable to the situation. In the source paper, Eden (1992a: 215) suggested to “ask the client to explain, in his/her own language, what goes on when using a GDSS and compare with the conceptual framework of the designers.” I do not think it is possible to overstate the significance of this approach. If any model or artifact is used for something else than it was designed for, it may not be simply useless, it can be outright harmful or, at least, dangerous. I still remember, from my student years, a particular finite element modelling package which some architects used to perform some design calculations for a bridge. They did not check the underpinning conceptual framework (the tool was, at the time, designed for a particular type of mechanical engineering problems) and the bridge collapsed, killing a dozen people. While consequences of misapplying GDSS are rarely so severe (at least the consequences are not so directly linked), they are not necessarily less disappointing.
Overall, the three suggested approaches for considering the success of GDSS – through controlled experiments, by comparing GDSS to their conceptual underpinnings, and by asking the users – are as valid today as they were three decades ago. It may be possible to replace some controlled experiments with computerized simulations but that does not affect the logic of the three approaches. When I started to study these approaches, I did not know that I will connect it to the notions of consistency, relevance, and applicability, which I previously used in different contexts for knowledge validation (e.g., Velencei et al. 2016). The significance of this coincidence is that if we can observe similar patterns in different validation situations, we may trust it more, as this means that from different starting points, following different routes, we arrived at similar considerations of what works well.
GDSS: To Support or to Substitute?
Based on an earlier conference paper version of Ackermann and Eden (1994), the source paper distinguishes three categories of GDSS with respect to the role(s) computers play in the process:
[…] not everything that can be counted counts,and not everything that counts can be counted.William Bruce Cameron: Informal Sociology2
Computer-driven GDSS involves direct entry from members of the group, e.g., Group Systems; if there is a human facilitator involved in these systems, their role is primarily to help the participants feed the input into the system the right way, so that its algorithms can deliver the best performance.
Facilitator-driven computer-supported GDSS involves real-time computing that is integral to the activity of the group, e.g., Decision Conferencing (using HiView, now Hieview3), SODA (at the time using COPE, later Decision Explorer, and now strategyfinder), Metagame Analysis (using CONAN), and Strategic Choice (using STRAD).
Facilitator-driven GDSS with no computer support, such as SSM and Strategic Choice (as practiced by Hickling 1974), solely relies on the facilitator, there is no computer involved in the process in any way.
Only completely well-structured tasks that are solely based on factual data, without any need for judgment or opinion, can be fully computer-supported. Basically, we are not talking about much beyond what database transactions can cover.
Clearly the more central the computer is in determining process, the more problems there are in the system identifying when a group is ‘finished’ with the problem. Nevertheless several GDSS designers seemingly are determined to design facilitator-less (and chauffeur-less) systems. I cannot envisage such a system having a sophisticated enough ‘expert system’ qua facilitator embedded within it for this ever to be a sensible proposition – unless the group is working on highly structured tasks, and even then I am dubious. Eden (1992a: 203)
There has been an incredible progress in computer technology over the past 30 years, so it is reasonable to expect that GDSS has been significantly affected. Indeed, GDSS has been significantly affected, at least at a surface level. But there has been no significant impact at the deep, thinking level. Inputs from participants now must work from their own devices, these must include tablets and mobile phones, it is expected that everything works with more or less no setup and seamless connection (typically using TCP/IP protocol), and all interfaces must be intuitively usable. This is a long way from the keyboard input and wired networks. However, it does not change the basic premises that there is user input, a joint display of the emerging model, and some analytical capabilities provided by the computer. In some cases, specific technology can be of great interest, for instance, approaches such as the Metagame Analysis could substantially benefit from virtual reality (VR). Any approach that displays a model of high complexity can also make use of VR for displaying 3D versions of the model, get the users inside the models to move around and explore parts in more detail, but these are only minor improvements in usability, that can be, at least for now, prohibitively expensive. It is possible to make use of 3D modelling to display richer information and possibly increase interactivity as well but, as far as I know, such developments are in early stages for now. GDSS can also be used in online distributed settings, with or without VR, but as many experienced facilitators will emphasize, there is significant advantage in being in the same physical space, seeing one another’s facial expressions, body language, hearing the tone of voice, etc., rather than just receiving smileys and similar symbols. Once we can have a real-time shared VR over great physical distances, the situation could be significantly improved, but for now, this is closer to science fiction.
Based on the previous train of thought, today I would only distinguish between computer-driven and facilitator-driven GDSS. These are, however, not separate boxes anymore but rather ends of a continuum. On the one end, we find computer-driven GDSS that works with minimal human input and with no facilitator. In the extreme cases, these will be fully automated systems aimed at substituting rather than supporting the decision makers. On the other end, we find facilitator-driven GDSS that is fully focused on the human (personal, transpersonal, interpersonal, group, organizational, and possibly social) aspects. This does not mean that the computer does not play an important role in the GDSS process, only that its role is not focal. I like to describe the role of computers in these systems as a support for the facilitator. It makes the facilitator’s work easier in terms of getting the input from the participants, and it provides the facilitator with real-time analytics, but the facilitator decides how to make use of the output of the analysis. Similarly, the computer can provide further back-office data for subsequent reports and learning of the facilitator.
Eden noted his suspicion that GDSS had been more successful with top managers than forms of DSS and EIS that did not involve a facilitation, as the facilitator’s role is crucial. My personal experience matches this suspicion. For many years, I have been involved in developing a knowledge-based expert system (KBS).3 We predominantly supported top executives using the software; sometimes working with groups of experts and at other times with individual decision makers – and we always provided facilitation. Doctus was well received by senior managers, and they often emphasized that it was all about the facilitation. So much so, that when we tried to sell the software, the response was that it would not make sense to buy it, as it did not work without us.
In every variant of GDSS, with the exception of the fully automated system, the most significant role of the computer is to display the evolving model, the so-called “transitional object”; this point is discussed in further detail below. Before getting to the dimensions of analysis, it is useful to take a quick look at how the conceptualizations of decision-making have evolved since 1992.
The Brave New World of Decisions
A former MBA student said many years ago: “you know, those decisions that you talk about in your lectures, do not exist in our organisations or, at least, they are extremely rare.” This sentence keeps haunting me, particularly if I look into what is taught in MBA programs about decisions.
Alice: Which way should I go?Cat: That depends on where you are going.Alice: I don’t know.Cat: Then it doesn’t matter which way you go.Lewis Carroll: Alice in Wonderland
The purpose of this section is twofold. On the one hand, I want to have a quick look at how the conceptualizations of decisions have evolved in the recent times. On the other hand, I want to show that, even though he could not have known about these newer conceptualizations, Eden’s comments suggest a very similar understanding. The most significant advancement in understanding decisions happened about the same time when the source paper was published. It was James March’s (1991) seminal “How Decisions Happen in Organizations” that made this leap forward. After that a few other important reconsiderations took place, such as the role of intuition was taken more center stage (Sinclair and Ashkanasy 2005; Kahneman and Klein 2009), rationality was even more forcefully questioned (Ariely 2009; Ariely and Trower 2019), and some underlying aspects of our choices have been explored as never before (Iyengar 2011).
At the time of source paper, the dominant views of decisions were brought together and divided by the concept of rationality (Simon 1947, 1955, 1956). On the one end, we have total rationality, with the assumption of fully measurable and calculable variables, complete information, including the knowledge of consequences of alternative actions, and computing power that allows for optimization. On the other end, we have bounded rationality, providing a counterpoint to balance the nonexistent world of totally rational decisions towards real-world decisions. In the real world, not everything is measurable and even what is, not necessarily on the same scale, information is incomplete, time and computing power are limited, decision alternatives may not be readily available, goals may contradict each other, and the consequences of different courses of action are rarely known. Between the two extremes, there are a variety of approaches to handling risks, uncertainty, ambiguity, and conflict. This is also what an MBA course would cover regarding decisions; occasionally we may come across heuristics and biases (Tversky and Kahneman 1974; Kahneman et al. 1982), decision traps (Hammond et al. 1998) and possibly superficial mentions of intuition and emotions (Simon 1987). All these models address the thinking of the individual decision maker, and they come a long way explaining how an isolated person would make decisions. However, decision makers do not exist in isolation and thus March argues that decisions in real organizations look very different.
March suggests that that decisions “happen” rather than “being made” and therefore the organizational processes that result in decisions “may be poorly comprehended by a conception of intentional, future-oriented choice” (March 1991: 97). As the first reconsideration of the rational choice approach, March suggests looking into rules, including organizational procedures, traditions, cultural norms, considerations of what is appropriate, obligations, duties, and the advices or actions of others. Decision makers often observe these, while ignoring their own, fully conscious preferences. Of course, there are always too many rules that apply in a particular situation and, just like the goals, they can be conflicting. Therefore, the logic that March suggests is to describe the situation using all the rules, then ask the Don Quixotean question of “who am I?”, and then match the two. So the logic of decision-making becomes “what is appropriate for me to do in this situation?”, which is an approach focused on the starting point rather than on the consequences of the choice (rational or otherwise). Things get even more interesting, if we consider that decision makers may be misbehaving (Thaler 2015; Baracskai and Dörfler 2017), i.e., not doing what they think is appropriate.
From this starting point, March arrives at a series of provocative observations – however, no one with experience in decisions in organizations can deny them. The first point is fairly trivial, i.e., that decisions do not happen in hierarchies but rather in constantly changing and partially overlapping networks of people and objects within and among organizations (cf Mintzberg and van der Heyden 1999). The second, less obvious, point is that orders tend to be temporal rather than consequential. The reason for this is that people are simultaneously involved in many different things. In addition, problems, decision alternatives, and decision makers are time dependent, both in arrival and life span. Furthermore, some people, activities, ideas, etc. are more attractive to a person than others, and this attractiveness also has a temporal aspect. Thus attention becomes a valuable currency – when people attend to some things, they do not attend to other things (cf Davenport and Beck 2001). As a consequence, problems, solutions, etc. get linked not because they are in means-ends relationship but because of their temporal proximity. The third point is about symbols and values in construction of meaning. March (1991: 110) observes that “[i]ndividuals fight for the right to participate in decision processes, but then do not exercise the right.” The reason, according to March, is that decisions are opportunities to demonstrate virtues or explain what is going on, to reconsider or reaffirm alliances and make new ones, to socialize, to educate newcomers, and have a good time being involved in the decision. (This reasoning parallels that of Eden and Ackermann 1998: 48–49; this will be discussed in more detail in the section on negotiation.)
Many things are happening at once; technologies are changing and poorly understood; alliances, preferences, and perceptions are changing; problems, solutions, opportunities, ideas, people, and outcomes are mixed together in ways that make their interpretation uncertain and their connections unclear; actions in one part of an organization appear to be only loosely coupled to actions in another; solutions seem to have only modest connection to problems; policies are not implemented; decision makers seem to wander in and out of decision arenas. (March 1991: 107–108)
The three points raised by March together induce a surprising picture: decision makers request great deal of information (outcome of analyses) for their decisions, but then they barely use them, instead, they are scanning the horizon for ideas, opportunities, and people. Outcomes of particular decisions are significantly affected by aspects with no apparent connection to those particular decisions, as they are happening around the same time, as others are interested or disinterested in them, and as they have a particular value or affect particular networks. March (1991: 110) goes so far to say that “decision processes are only partly – and often almost incidentally – concerned with making decisions.” Although this sounds shocking when brought together this way, anyone who has facilitated decision-making groups in organizations or participated in such groups can confirm that this picture resembles reality.
On a superficial look, it may seem that the first part, the rational choice approach is more linked to substantive rationality, while the other two are closer to procedural rationality (Simon 1976). However, on a closer examination, we can realize that this is not the case. Each of the three approaches described above have an aspect of substantial as well as procedural rationality. If we think about the decision outcomes, in any of the above three approaches, in terms of whether the outcome makes sense, whether it can be justified, explained, and communicated, we are within substantive rationality. If we look into the behavioral and social processes that led to the outcome, the realm of examination is procedural rationality. Eden is very firm in pointing out that a GDSS must feature both forms of rationality, particularly because both are necessary for achieving political feasibility. Of course, how we approach political feasibility will also affect, and be affected by, how we approach decisions in terms of the above three categories, when designing a GDSS.
It is remarkable that from the duality of substantive and procedural rationality and some observations, Eden paints a picture of GDSS that covers virtually every aspect of the two revisions of decision approaches suggested by March. This is well reflected in the SODA approach to Making Strategy that Eden has developed with various collaborators (Eden and Ackermann 1998, 2001, 2009; Ackermann and Eden 2011a). Typically such GDSS workshops start from the burning issues of the participants (see “Group Support Systems – Concepts to Practice” and “Procedural Justice in Group Decision Support”), there is attention paid to the dynamics, courses of action, values, and preferences are negotiated, the causal map is co-constructed, the aim is to achieve a consensus, power brokers are included, intuitions and expert opinions of participants are explicitly welcome, emotional commitment is not only fostered but also declared.
Having reviewed the approaches to considering the success of GDSS, having had a quick look into the role of the computers and having overviewed the recent changes in the conceptualizations of decision-making, the scene is now set for revisiting the dimensions of analysis used by Eden in the source paper.
Dimensions of Analysis
In the source paper, Eden conducted his analysis along four dimensions: political feasibility (with coordination and cooperation being featured particularly prominently), meeting productivity, negotiation, and creativity. The scope of the analysis was developments of GDSS at the time and what was then a likely future, including the past 30 years. And his declared purpose was the following:
There are more things in heaven and earth, Horatio,than are dreamt of in your philosophy.William Shakespeare: Hamlet
I stick approximately to the same dimensions of analysis. There are two differences:
The purpose of this discussion, at the level of conceptual and theoretical assertions, is intended to inform a further debate about the implications for the effective design of GDSSs. Thus the article is specifically seeking to relate decision making in a group to issues in the design of a group decision support system. (ibid.: 200)
Together with creativity, I also include intuition, for several reasons. The scholarly literature on intuition has significantly developed over the past two decades. Creativity involves intuition, and consulting experience repeatedly highlighted the immense value of intuition in GDSS.
I am more prominently emphasizing the notion of “transitional objects,” as today we have a much better understanding of their role in GDSS and any setting where interpersonal relationships matter. I primarily talk about this in the dimension of negotiations.
In what follows, I address these dimensions one by one, exploring the claims from the source paper and, where possible, providing updates.
Political Feasibility: Focusing on Implementation
Political feasibility is a very simple concept, which acquires extremely high complexity when it manifests in the real world. Other chapters in this handbook deal with political feasibility (“Group Support Systems - Concepts to Practice” and “Procedural Justice in Group Decision Support”) from a somewhat different perspective; the chapters are complementary rather than overlapping. The chapters use the same approach to political feasibility, namely that regardless how good (or sound, or rational, etc.) a decision is, if it is not politically feasible, it will be ineffective, that is, it will not bring about the intended change.
The notion of political feasibility is closely linked with the notion of implementation. This means that we need to get beyond Simon’s (1977) decision phases (intelligence, design, choice, implementation), as in reality all the participants think about the implementation from the very beginning, as soon as they start to formulate the problem (Eden 1987). It helps if the phases are refined using cycles (Eden 1987: 103), and these can be reasonably well observed in GDSS if they are allowed and supported. I tend to describe the phases as paradoxical, in the sense that each phase seems to contain all the other phases.
The manoeuvring of people along Machiavellian dimensions is relatively easy to identify, but it is, in my experience, much less common than the politics that results from the wish to define reality. This latter form of politics is the essence of human life, it derives from honest people believing they know what is best for the organization. (Eden 1992b: 803)
What we need to achieve is sufficient alignment between the players regarding the particular actions at hand. In other words, political feasibility will enable enactment (cf Weick 1979, 1995). This will have two aspects, as political feasibility links back to substantive and procedural rationality. For any decision alternative, or course of action, to be politically feasible, it must be both appropriate in itself as well as arrived at through appropriate processes. Therefore, the GDSS has to offer a method for influencing both the analysis of the outcomes as well as the attitudes of the participants (“Behavioural considerations in Group Support”). The term order (instead of solution or decision) is useful for increasing political feasibility, as it implies settlement, harmony, progression, and arrangement, which are well aligned with being satisfactory in terms of both substantive and procedural rationality. The notion of order is also more consistent with decisions not being only about creating (with a very inappropriate word generating) alternatives and choosing among them. “Decision making is influenced by the way in which issues are presented, the identification of their significance, their exploration as the group constructs a shared understanding of them, and the point at which a negotiated settlement is likely.” (Eden 1992a: 204). Thus negotiating this order is the most crucial element of political feasibility.
In political treatments, however, the emphasis is less on designing a system of contracts between principals and agents, or partners, than it is on understanding a political process that allows decisions to happen without necessarily resolving conflicts among the parties. (March 1991: 103)
Any GDSS, in order to ensure an egalitarian participation and the free expression of ideas, takes away power from some people and gives them to others, as power and social skills would otherwise determine who is heard, how often, for how long, and with how much impact. Based on Kim and Mauborgne (1991), Ackermann and Eden (2010, 2011b) call this redistribution of power procedural justice (see “Procedural Justice in Group Decision Support” and “The Role of Justice in Negotiation”), and they emphasize that procedural justice is not about democracy but about good management. In the first approximation, this is important as it enhances information exchange. Everyone should be heard and listened to. However, participation is not solely about information exchange. As March explained, people often fight to get into a decision-making position but then they often do not exercise their role. We do not only need people to get a seat around the table, we need them to want to be part of the group processes, to participate the best they can. Creating the sense of procedural justice through temporarily redistributing power disturbs the social order. This is very useful in obtaining expert opinions (which is frequently the highest quality information available), as the experts are less likely to try to guess what their bosses want to hear. Of course, we need to be attentive that the disturbance of the social order remains temporary. Finally, as will be emphasized below, procedural justice is tightly linked with emotional commitment, which is of paramount importance. On the technical side, anonymity can help a lot with achieving procedural justice, and this is easily achieved when using computers – which, as said above, characterize all GDSS today.
Procedural justice helps everyone’s perspectives, opinions, intuitions getting “on the table,” but this is only the first step of achieving political feasibility. However, if all is on the table that anyone wanted to add, we are up to a good start. Assuming that the decision participants are sensible people who are good at what they do, it is plausible that each participant will comprehend what the other participants mean, their respective priorities, issues, etc. The facilitator’s role is as crucial in the subsequent stages, as it was in the initial stage in achieving political feasibility. Time is of essence, as even a politically feasible solution will be ineffective if it is achieved too late, therefore the facilitator also needs to make the group work productive.
Meeting Productivity: Time Is of Essence
The topic of meeting productivity sounds trivial. This was also my impression when I came across some literature on decision-making, many years ago, getting into very fine details about the size and shape of the room, about the furniture and the facilities within it, about the color, shape, and size of the post-its and pens (or whatever else they were using), making suggestions on how to gain a few minutes here, and a bit better performance there (see more details in Huxham 1990). While these things really do not make for a particularly interesting intellectual journey, they are of immense importance once we come to the practice of GDSS. Of course, the pen-and-paper issues are now superseded by their computerized counterparts, but the principles are pretty much the same. So meeting productivity is concerned with achieving a reasonably good outcome in a reasonably short time. Experienced decision makers often say that a relatively good decision right now is usually much better than a perfect decision later.
Increasing commitment and improving decisions both point towards increased group size. These help achieving improvements both along substantive as well as procedural rationality. However, running a GDSS process with more participants takes more time, and the time increase can be exponential with substantially higher numbers. And the available time is limited, decisions are usually urgent. Many in the GDSS arena, including Nunamaker and Eden (Nunamaker et al. 1988), argued that meeting productivity is an important aspect of good GDSS. In principle, the story is relatively simple here. Not so much when we get our feet on the ground in the real world, to design a GDSS and work with it. There has been significant progress achieved over the past three decades in GDSS design, both in designing the social process as well as the software that supports it, in the area of meeting productivity. This progress has been the result of a series of miniscule steps, tiny gains, but there is a very large number of them and the tiny gains add up to substantial gains. For outsiders, some of these will seem trivial, others unnecessary or even counterintuitive. For instance, in SODA, one recommendation for facilitators is to add links between concepts using a keyboard shortcut. In today’s mouse- and touchscreen-oriented world, this does not seem to make sense. It took me several sessions to realize how much faster it is. Perhaps only a fraction of a second in some cases, a few seconds in other cases, but a few hundred times these seconds and fractions can add up to many minutes. In addition, the process gets less fragmented, which can bring additional minutes. And there are dozens of such speed gains that, together, make a significant difference.
With Doctus (the software mentioned earlier), such speed gains did not appear to matter. We were paid by the hour, the hourly rates were limited, but the hours were not. Our clients did not seem to be in a rush. We did create several efficiency gaining shorthand solutions anyway, and the clients were not interested in using them, so we stopped. Then, quite suddenly, we realized that we were slow. We could not deliver a full GDSS (or DSS) process in 1–2 days, and we lost numerous opportunities, as speed became expected. It did not even matter that the quality that we produced was excellent.
Increasing meeting productivity is not a simple matter, in spite of the simple tiny gains described before. These worked, as they increased speed at an elementary level, by speeding up individual steps. However, these steps do not exist in isolation, and their interconnectedness is nonlinear. In other words, as already said before, the GDSS process is complex. Therefore, we need to be very careful not to end up in the situation of Mintzberg’s (http://www.mintzberg.org/blog/orchestra) hypothetical MBA student, trying to increase the efficiency of the symphony orchestra. He suggested removing two of the four oboes and distributing their activity more evenly, as they had nothing to do for considerable periods of time, drastically reducing the number of violins, as they were often playing the same notes, rounding up the notes, removing the repetitions, and replacing the several hundred years old instrument of the first violin with a newer model. It is so absurd and funny, and devastating. However, very often, similar things are done in organizations in the name of productivity, only the absurdity is less obvious. I have noted previously that, as the decision situations are complex, GDSS design must be complex too. The same applies to increasing the meeting productivity. I find it very worrying when it is suggested that productivity can be increased by improved software design. I categorically say NO to this. Software design cannot improve productivity. Approaching the problem as complex as it is, we must design processes, with a complex systemic attitude, involving excellent facilitators with considerable experience, listening to their intuitions, listening to experienced participants. All software design can do is to support the redesigned processes – and this is very important. Although the best software will not be worth much without good processes, a poorly functioning software can destroy otherwise excellent processes. The reason for this is that GDSS is all about the human-social aspects. Therefore, it needs excellent facilitators conducting excellent processes, so that the participants can productively negotiate politically feasible new orders. Next, we look into some aspects of the nature of negotiation.
The Nature of Negotiation: The Role of the Transitional Object
Linking back to political feasibility, I have noted that it is a new order that is negotiated; this new order has two interrelated features. The first one is called socially negotiated order, emphasizing that the new order is socially constructed, linking to procedural rationality. The second one is called negotiated social order (cf to March’s comments on participation in decision-making and see Eden and Ackermann 1998: 48–49 for more details), emphasizing that the new order, through organizational change, impacts the social relationships, linking to substantive rationality. A politically feasible GDSS process will balance the two aspects of negotiating the new order (Eden 1992b; Eden and Ackermann 2010); the facilitator plays an important role in this process that can be understood primarily by exploring the behavioral aspects of it. The role of the facilitator is discussed in detail in other chapters; here I focus on the role of the transitional object.
Whichever GDSS approach we consider, there is always a model which is developed in the GDSS process. This model is a boundary object, as it is at the boundary of the different individual perspectives, and it is also what de Geus (1988) calls a transitional object, as it is constantly changing during the GDSS process. Although Eden and others keep these two “objects” as separable notions, henceforth, in this chapter, I consider the concept of transitional object to cover the notion of boundary object as well. So why are transitional objects important for GDSS?
First, transitional objects help to get the right distance from what is happening in the workshop. The “right distance” means not being so close, embedded in the process, that we get lost in the details, losing sight of the big picture, but we are also not so far that we cannot see the details anymore. The transitional object therefore helps “seeing the essence,” which means seeing both the detail and the big picture and swiftly switching between the two (Dörfler and Eden 2019). Furthermore, having the right distance means that discussing what is being said becomes easier, as it is not the other participant one is commenting on but the transitional object. At the same time, the transitional object displays the various perspectives, helping the participants not only the make sense of each other’s views but also to develop an appreciation of each other’s priorities. The participants can see their views represented in the transitional object, so even if what they thought of as a high priority issue is deprioritized during the GDSS process, they will have an appreciation of how this happened. Experience shows that the participants find it much more acceptable to have their views acknowledged and then deprioritized than simply denying them without consideration.
Second, the transitional object helps develop an emotional commitment and the sense of ownership. The participants’ sense of ownership does not come as a surprise; each of them was a creator of the transitional object. However, as the transitional object stands for something beyond itself, it represents the new order in-the-making, and they will also have a sense of ownership for that new order. Furthermore, as they co-created the transitional object and the new order it represents, it is a sense of shared ownership. The process of co-creation brings the participants closer together, and they develop commitment not only to the new order but also to each other. Beyond learning to appreciate each other’s views and priorities, they also learn to appreciate each other. Therefore, the developing commitment will have a strong emotional dimension.
But it is not only philosophy but also hard science that teaches us about the significance of emotions. Antonio Damasio (1995), working with a patient referred to as “poor Elliot,” whose emotions were disabled due to an injury, found that emotions are necessary for decisions. Elliot was able to rationally argue about various alternatives, to analyze pros and cons, to evaluate different aspects, but he could not decide. If we just think about any of the most significant decisions of our lives, such as what profession to choose, with whom to spend our lives, where to live, we take all these decisions on emotional basis. Perhaps less obviously, all decisions are like that. What we want to do is mostly emotional and how we go about it, has more to do with reason. In order to acknowledge the significance of emotions, in psychology as well as in management and organization studies, today scholars talk about “cold cognition” and “hot cognition,” where the former refers to the detached reason while the latter to an involved emotional stance (Healey and Hodgkinson 2017; Hodgkinson and Sadler-Smith 2017). In GDSS, emotions play a particularly important role (Martinovski 2010, 2015), especially in relation to the transitional object, as they enhance the GDSS process as well as the sense of ownership.
Reason is, and ought only to be the slave of the passions, and can never pretend to any other office than to serve and obey them. (Hume 1739: Book 2, Part III, Section 3)
Finally, the transitional object helps enhancing the quality of the new order. This has several components. As all the ideas are displayed in it, the transitional object also supports obtaining further ideas by prompting additional thinking. This is not limited to the volume of ideas, and there will also be further cohesion, as the new ideas will relate to what is already there. It will also become easier to spot if there is a hole in the big picture, so the hole can be filled and the big picture becomes more complete. With the transitional object, it is less likely that something is forgotten or not considered. The developing big picture will also help the participants change their minds. Nobody else can change one’s mind but oneself, but the transitional object enables the internal dialogue as a reference point. This dynamics of changing minds, with the assistance of an excellent facilitator, helps achieving consensus. The consensus, together with the sense of ownership and the emotional commitment, goes a long way in achieving political feasibility.
What has been discussed about transitional objects so far in this section has been known for a while from the GDSS literature and from the related consultancy experience. So far I have brought together what has been fragmented in the literature, but I want to go a step further. When they were exploring what makes Communities of Practice (CoPs) work, Pyrko et al. (2017) introduced the concept of “thinking together” as the core process of CoPs. Thinking together is a transpersonal thinking process that can be conceptualized based on Polányi’s (1962) notion of indwelling. In CoPs, indwelling is interlocked on the real-life problems the CoP members care about, based on their shared knowledge tradition (see also Pyrko et al. 2019). Dörfler and Stierand (2018) have subsequently explored different modes of indwelling and argued that in different contexts, indwelling can be locked or interlocked on different things, for instance, personal tacit knowing in locked on the subject of study. Based on this, I propose that in a GDSS process, thinking together can happen by the indwelling being interlocked on the transitional object. The GDSS participants may care about some of the same problems but they will also care about some different ones, and usually they will not have a shared knowledge tradition. This raises obstacles to thinking together, and the transitional object can serve the purpose. It can act as an enabler of thinking together, although it will not make it happen on its own, the eagerness of the participants and an excellent facilitator will be essential. In GDSS, particularly if the participants think together, creativity and intuition plays an important role – this is what I explore below, as the last dimension of analysis.
Creativity and Intuition
Above I have noted that the transitional object enables creating new ideas in the GDSS process. The significance of these new ideas is that instead of fighting over old options, the participants create new ones. These new options make it easier to achieve consensus while supporting the development of emotional commitment and of the sense of shared ownership. The new options require creativity, which is typically defined as new and useful ideas (Amabile 1983, 1996). This is perhaps sufficient to justify the importance of creativity in the GDSS process.
The greatest benefit that GDSS can bring to the organization is realized through the political feasibility, supported by the consensus, emotional commitment, and sense of shared ownership. For this, it is of paramount significance that the participants create new options rather than fight over old ones. An excellent facilitator could achieve a conflict resolution through compromise based on old options; however, consensus can be achieved, at least more easily, on the basis of new options. Moreover, there is more to creativity in GDSS than just adding new options. The participants also synthesize options already displayed in the transitional object as well as the newly added ones, and this leads to further synergies. It is trivial that, when they come to the table, the participants bring their often conflicting goals and viewpoints. Synthesizing the old and new options helps getting the participants’ directions aligned, and this alignment is not forced upon them from the outside but emerges from the GDSS process.
While each GDSS process will have a different take on fostering the creativity of the group, there are three commonalities: it is anchored in the transitional object, it is fostered by the facilitator, and it does not resemble the brainstorming processes often advocated in the creativity management literature. According to Eden, the reason is that brainstorming works better in situations when the participants have expertise (preferably from the same discipline or problem domain) but no decision-making prerogative. Furthermore, processes like brainstorming may be harmful, as they may interfere with the problem-solving process that is at the heart of any GDSS (Eden 1992a: 210).
The social aspect of the problem-solving process is important here, and therefore, creativity in GDSS is highly linked with the relationship-building that takes place. The computerization does not help relationship-building, which is why Eden (1992a: 211) repeatedly argued that GDSS is usually too tedious and not enough fun, that it lacks humor. This becomes even more important, if we take into consideration that the underlying logic of creativity is in essence the same as the logic of jokes (Dörfler et al. 2010). A good facilitator will make humor part of the GDSS process, but this is done in nontrivial ways rather than following a recipe, as creativity is a complex systemic process (Stierand et al. 2014), just like everything else in GDSS. Creativity also makes the story of the GDSS process more interesting, and this story is important. The initial options brought to the table by the participants can be considered an antenarrative, while the resulting model could be considered the final story (cf Stierand et al. 2019).
Intuition is relevant to GDSS at least in two different ways. First, in relation to creativity, as creativity requires intuition, namely what we call “intuitive insight” (Dörfler and Ackermann 2012; Stierand and Dörfler 2016). The significance of intuitive insight is that in any GDSS, the creative ideas are not scrutinized on the basis of justification but rather based on whether they make sense in the context of the decision(s) at hand. Second, “intuitive judgments” (Dörfler and Ackermann 2012) brought to the table by the participants are useful for evaluating the options, the ways forward, and everything that has a value attached to it. It is important that intuitions are the intuitions of experts, as intuition works reliably at a high level of expertise (Kahneman and Klein 2009; Dörfler and Stierand 2017). In both cases, intuition is considered to be a form of tacit knowing (“From Soft and Tacit to Deliberate”), and I would go so far to argue that a GDSS can be only as good as much it makes use of the participants’ intuitions.
GDSS, Big Data, and Artificial Intelligence
The increased computerization of GDSS and the more general hype of big data and artificial intelligence (AI) makes it necessary to consider what they can bring to GDSS – but not to become uncritical enthusiasts. There are possible benefits, but these are limited; intuitively this is obvious from the complexity of GDSS and from the importance of the facilitator.
The prospect of machine interpretation is not only whimsical; it is absurd. Interpretation belongs solely to a living mind in exactly the same way that birth belongs solely to a living body. Disconnected from a mind, ‘interpretation’ becomes what ‘birth’ becomes when it does not refer to a body: a metaphor.Theodore Roszak: The Cult of Information
It is frequently asserted that, more often than not, focusing on big data leads to “big data – small insight.” Instead, Michael Pidd (2017) recommends “small data and big thinking.” At first sight, big data seems to be irrelevant to GDSS anyway, as even a comparatively large number of participants would still conveniently fit within the scope of “small data.” Furthermore, as thinking, and even better, thinking together is at the center of GDSS, we should and often do achieve “small data and big thinking.” However, this is not the complete picture. The GDSS participants do not come to the table empty-handed; they bring with them the analyses that they have conducted before or that they are familiar with. Eden suggested on multiple occasions (see, e.g., Ackermann and Eden 2011a) that analysis, at least good analysis, can and should inform the GDSS process. This means that big data, more precisely big data analytics (BDA), can serve as a useful input, and it can inform the GDSS process.
While BDA was relatively straightforward to deal with, this is not the case with AI; to a large extent, this is due to misunderstandings and misrepresentations of AI. In order to figure out what role AI can play in GDSS, it is useful to distinguish between GDSSs that are AI-based and that are not. In AI-based GDSS, the model that becomes the transitional object is created or supported by AI; these include, among others, knowledge-based expert systems (KBS) and artificial neural networks (ANN). I will not discuss these here, as the role of AI would be specific to the particular GDSS – but I would be concerned of any AI-based GDSS that puts more emphasis on AI than on the facilitation process. As a colleague of mine said, the less AI the expert system contains, the better it is. The reason is the misrepresentation surrounding AI. The data processing underlying AI is not akin to thinking, as only a very small part of thinking is data processing. The reinforcement learning, used by ANN, is not akin to learning, as only a very small part of learning is done through reinforcement (see the TEDx talk at https://youtu.be/KAXfo-cZ8oY for a more detailed account). In short, AI and humans are good at different things.
This difference is what we should be focusing on; we can make the best use of AI in the GDSS process if we use what AI is particularly good at and humans are not. As Thomas Davenport (2018: 44) says, AI is only “analytics on steroids.” There are many who suggest that AI should replace the facilitator. With more than 20 years of experience in using and developing AI, I believe that this is never going to happen, at least not in a beneficial way. AI will never replace an excellent facilitator. However, as AI can be excellent in analyzing the data generated by the participants real-time, and I am not primarily talking about what they put in the model, but about behavioral data, and this can be supplied in the form of real-time support to the facilitator, who can then decide what to do about it. The facilitator’s perception of the group and her/his intuitive judgment of the patterns suggested by AI are crucial for an excellent GDSS process.
Finally, it is important to ask what role AI can play in creativity, specifically in the context of GDSS. I have explored the topic of AI creativity in detail elsewhere (Dörfler forthcoming), here I just want to outline the conclusions. First, I believe that AI cannot be creative. As knowledge in AI is limited to explicit knowledge and creativity requires intuition, which is a form of tacit knowing, AI creativity cannot happen. Surely AI may produce something that satisfies the two criteria of creativity, something that is new and useful. However, there is a hidden third requirement: it has to be an idea. AI does not have ideas – but if AI suggest something that people find new and useful, they can transform it into an idea. Thus AI can support human creativity by providing some sort of preprocessing. There is, however, another, perhaps somewhat counterintuitive way how I believe AI can help human creativity: AI is not affected by “tunnel vision” or “group think,” and therefore, AI can help us think “outside the box” by showing patterns that we may not allow ourselves to see imposing unnecessary limitations.
I have to note that what I said about AI is a personal view – even if it is a personal view that is rooted in two decades of experience. Many AI experts would disagree with me. I allow the possibility that they may be right and I may be wrong about some details, but these do not affect what role AI can play in GDSS today or in the near future.
Concluding Remarks Through Personal Reflection
I have experience with two types of DSS/GDSS: for 20+ years I have been doing consultancy work with knowledge-based expert systems, leading related software development, using it in my research, and teaching about the subject. In addition, for the past few years, I have been involved in using causal mapping (specifically the SODA approach), although primarily in teaching and research. Both these experiences informed my argument in this chapter. Admittedly, this is still only a personal opinion, and it is as much based on beliefs and opinion as on facts. However, it is a well-informed personal opinion.
I have found that the context of GDSS has considerably changed over the past three decades, particularly, now we have a very different understanding of how decisions happen and the underlying computer technology has significantly evolved. Although these are both of great interest to studying GDSS, they seem to be less significant in terms of the paper that served as my starting point, as Colin Eden has approached it 30 years ago in a way that is consistent with the changed understanding of decisions and he only focused on computerized GDSS in the first place. It has also been reinforced that we need to design GDSS processes with a complex systemic mindset, and all computers can do is to support these processes – software design cannot substitute system design. The dimensions of analysis that Eden used in that paper can still be meaningfully maintained, as it is indispensable for a viable output of the GDSS process to be politically feasible, and to achieve this, we need to conduct productive meetings, enable negotiation, and foster creativity and intuition. While computers can help a lot with the GDSS process, at the core of it are the GDSS participants, with their intuitions, creative ideas, social relationships, agendas, arguments, and personalities.
The first of the two main changes that happened in the last 30 years is that despite all the computerization, the role of the facilitator is confirmed as exceptionally important, perhaps the most important ingredient of good GDSS. The reason that I see this as a change rather than remaining the same is that it is a completely different world in terms of computers, big data and AI – and I argue that a good facilitator cannot and never will be replaced by AI. The second change is that we have a much better understanding of transitional objects, their role, significance, and modus operandi in the DSS/GDSS context. I maintain that further technological development will primarily benefit GDSS through creating better transitional objects, at least, in the short term.
This notion of strategic does not necessarily refer to the overall corporate strategy but can also mean strategy at the level of an organisational unit or team, whatever we are supporting with the GDSS.
Maxim, technically a chiasmus, usually attributed to Albert Einstein (allegedly he once wrote it on his blackboard), but it seems that it was first brought together, at least in writing by Cameron (1963: 13).
It is the Doctus KBS (www.doctuskbs.com), started and owned by Zoltán Baracskai.
- Ackermann F, Eden C (2010) The role of group decision support systems: negotiating safe energy. In: Kilgour DM, Eden C (eds) Handbook of group decision and negotiation. Advances in group decision and negotiation. Springer, New York, pp 285–299. https://doi.org/10.1007/978-90-481-9097-3CrossRefGoogle Scholar
- Ackermann F, Eden C (2011a) Making strategy: mapping out strategic success. SAGE, LondonGoogle Scholar
- Amabile TM (1983) The social theory of creativity. Springer, New YorkGoogle Scholar
- Amabile TM (1996) Creativity in context: update to the social psychology of creativity. Westview Press, BoulderGoogle Scholar
- Ariely D (2009) Predictably irrational: the hidden forces that shape our decisions. HarperCollins Publishers, LondonGoogle Scholar
- Ariely D, Trower MR (2019) Amazing decisions: the illustrated guide to improving business deals and family meals. Hill & Wang, New YorkGoogle Scholar
- Cameron WB (1963) Informal sociology: a casual introduction to sociological thinking. Random House, New YorkGoogle Scholar
- Checkland PB (1999) Systems thinking, systems practice. Wiley, ChichesterGoogle Scholar
- Checkland PB, Scholes J (1999) Soft systems methodology in action. Wiley, ChichesterGoogle Scholar
- Damasio AR (1995) Descartes’ error: emotion, reason, and the human brain. Avon Books, New YorkGoogle Scholar
- de Geus AP (1988) Planning as learning. Harv Bus Rev 66(2):70–74Google Scholar
- Dennett DC (1995) Darwin’s dangerous idea: evolution and the meanings of life. Penguin Books, LondonGoogle Scholar
- Dörfler V (forthcoming) Artificial intelligence. In: Runco MA, Pritzker SR (eds) Encyclopedia of creativity, 3rd edn. ElsevierGoogle Scholar
- Dörfler V, Stierand M (2018) Understanding indwelling through studying intuitions of nobel laureates and top chefs. Paper presented at the AoM 2018: 78th annual meeting of the academy of management, ChicagoGoogle Scholar
- Dörfler V, Baracskai Z, Velencei J (2010) Understanding creativity. Trans Adv Res 6(2):18–26Google Scholar
- Eden C (1987) Problem solving or problem finishing? In: Jackson MC, Keys P (eds) New directions in management science. Gower, Aldershot, pp 97–107Google Scholar
- Eden C (1990) The unfolding nature of group decision support: two dimensions of skill. In: Eden C, Radford KJ (eds) Tackling strategic problems: the role of group decision support. SAGE, London, pp 48–52Google Scholar
- Eden C (1992b) Strategy development as a social process. J Manag Stud 29(6):799–812. https://doi.org/10.1111/j.1467-6486.1992.tb00690.xCrossRefGoogle Scholar
- Eden C, Ackermann F (1998) Making strategy: the journey of strategic management. SAGE, LondonGoogle Scholar
- Eden C, Ackermann F (2001) SODA: the principles. In: Rosenhead J, Mingers J (eds) Rational analysis for a problematic world revisited: problem structuring methods for complexity, uncertainty and conflict, 2nd edn. Wiley, Chichester, pp 21–41Google Scholar
- Eden C, Ackermann F (2009) SODA – journey making and mapping in practice. In: Rosenhead J, Mingers J (eds) Rational analysis for a problematic world revisited, 2nd edn. Wiley, Chichester, pp 43–60Google Scholar
- Eden C, Ackermann F (2010) Decision making in groups: theory and practice. In: Nutt PC, Wilson D (eds) Handbook of decision making. Wiley-Blackwell, ChichesterGoogle Scholar
- Handy C (1991) The age of unreason: new thinking for a new world, 2nd edn. Arrow Books, LondonGoogle Scholar
- Hickling A (1974) Managing decisions: the strategic choice approach. Mantec Publications, RugbyGoogle Scholar
- Hume D (1739) A treatise of human nature. Oxfor University Press, Oxford, UKGoogle Scholar
- Huxham C (1990) On trivialities in process. In: Eden C, Radford KJ (eds) Tackling strategic problems: the role of group decision support. SAGE, London, pp 162–168Google Scholar
- Iyengar SS (2011) The art of choosing: the decisions we make everyday – what they say about us and how we can improve them. Little, Brown Book Group, LondonGoogle Scholar
- Kvavilashvili L, Ellis J (2004) Ecological validity and twenty years of real-life/laboratory controversy in memory research: a critical (and historical) review. Hist Philos Psychol 6(1):59–80Google Scholar
- Polányi M (1962) Personal knowledge: towards a post-critical philosophy. Routledge, LondonGoogle Scholar
- Simon HA (1947) Administrative behavior: a study of decision-making processes in administrative organization, 1st edn. Macmillan, New YorkGoogle Scholar
- Simon HA (1977) The new science of management decision, 3rd edn. Prentice-Hall, Englewood CliffsGoogle Scholar
- Simon HA (1983) Reason in human affairs. Stanford University Press, StanfordGoogle Scholar
- Thaler RH (2015) Misbehaving: the making of behavioral economics. W.W. Norton, New YorkGoogle Scholar
- Velencei J, Baracskai Z, Dörfler V, Stierand M (2016) Supporting the competent practitioner: transdisciplinary coaching with a knowledge-based expert system. Int J Manag Sci Bus Adm 2(12):20–27. https://doi.org/10.18775/ijmsba.1849-5664-5419.2014.212.1002CrossRefGoogle Scholar
- Weick KE (1979) The social psychology of organizing, 2nd edn. McGraw-Hill, New YorkGoogle Scholar
- Weick KE (1995) Sensemaking in organizations. SAGE, Thousand OaksGoogle Scholar