*Creating Effective Management Simulations: Rapidly, Responsibly, Relevantly DOI: http://dx.doi.org/10.5772/intechopen.106430*

testing. But beneath the interface exists the code that manages the other aspects of all software. For management simulations, this code performs a relatively limited number of actions. The decisions being made in each turn are received from user input–usually a straightforward Web form-like interface—and formulaic calculations are applied. Some calculations will be informed by previous decisions made at earlier turns; some calculations will have random influences introduced to the calculation. These random influences can be as simple as a random number generator function between a specified range being applied in each turn. Some vendors also claim real-world influences can be incorporated into their simulation such as historical Nasdaq data.

At the core of a simulation are the calculations that it performs on the user's decisions and subsequent input. The calculations are represented through a variety of algorithms that have been documented in a significant body of literature [13] that particularly covers supply, demand, marketing, and finance. However, the body of literature also reveals that for some calculations there is no definitive scholarly agreement about which algorithm correctly models the observed experience. Beyond this level of scholarly debate, experiments have been conducted with business students that purposefully inserted faulty algorithms into a simulation. The conclusion of this research was that players were unable to detect the error and the error's presence had no impact on the quality of their economic performance [14]. In many ways, this conclusion confirms the opaque-box nature of simulations as students—or instructors—cannot see which algorithms are being utilized or when. It is a real consideration as to whether algorithmic validity is simply assumed when institutional purchasing decisions are being made, or whether algorithmic validity is only one aspect of the many considerations within the experience of the overall game world irrespective of the user interface [14]. Since the mid-2000s the number of academic papers discussing algorithms for management simulations have declined significantly. Two potential speculations regarding this decline are possible: the algorithms are now business sensitive to those companies productizing management simulations, and more generously, the algorithms that are being used are well-documented through conventional literature within the specific area of knowledge. This later speculation could be regarded as true for many common functions that are applied in everyday business, such as the power function and the price elasticity of demand. But as an opaque-box what specific algorithms are being used are obscured from the academic and their learners. The original Beer Game is perhaps a rare example of a largely transparent simulation, but this is partly a result of its longevity and relative simplicity. It is primarily based on the Bullwhip effect and the goal is to try and manage available resources on the wholesale side to minimize the amplifying effect of small changes applied on the retail side.

A common linkage between the Beer Game and many of the more recent developments is their connection to system dynamics thinking. That is a systems-based view of organizational activity that examines long-term change to recognize the functions of a system in order to abstract regular models of observed behavior. A system dynamics approach lends itself to algorithmic validity precisely because of this abstracted nature. While the algorithms will not model precise day-to-day behaviors, the tendency will be to provide an accurate outcome based on the received inputs. Many of the algorithms are relatively "simple"—with the benchmark for "simple" being the ability to recreate the algorithm as a formula within a single cell in Excel. But no commercial management simulation enables an individual installation to "tweak" existing underlying formulas or to completely replace the preferred option with an alternative perspective. The 30,000 papers regarding the bullwhip effect on

Google Scholar offer the clear suggestion that there have been some refinements to the concept and its representation since the Beer Game was first introduced. Although some of these papers are also admittedly and effectively formal academic documentation of how to win the Beer Game. This specific body of work could be seen as a marker of how effective, debated, and engaged with that this particular simulation has been across its lifetime.

While software-based management simulation provides the kinesthetic learning and Bunsen burner effect for business students, there is one further example of a more practical experiment from the 1970s that goes beyond being the computer interface of a simulation to becoming the reality. It is notable that the project, directed by Stafford Beer and a cybernetician, came from a different intellectual tradition than that of system dynamics. This is a sufficiently different enough community of practice to have already seen Beer create an analog computer in the 1950s out of a pond with scientist, author, and artist, Gordon Pask [15]. Beer's published work in cybernetic theory as well as related topics and practical work with United Steel, SIGMA, and the International Publishing Company all brought Beer to the attention of the new Allende government in Chile [15]. By 1971, Beer was in Chile working with the government building project Cybersyn to coordinate the many newly nationalized industries [16]. Project Cybersyn was closely modeled on Beer's theorization of the viable system model, which present a specific model of how the parts of an organization relate to one another and to the external world. Beer applied these principles to create a centralized control room—echoing the title of his book that had first attracted the Allende's government's interest, "the Brain of the Firm." [17] The control room received daily reports from individual industries, including farms and mines, that reported on outputs and other key metrics. Within the control room decisions were then made through a strategic lens considering the current prevailing international and domestic political and economic conditions. Most importantly individual production decisions were made with fuller knowledge of national conditions, including upcoming peaks in demands or changes to government policy. The resultant decisions were then conveyed back out to the different elements of the economy controlled by the government. In some cases, this may have been an instruction to reduce production or even to call for a temporary cessation of certain activities. Cybersyn used the latest technology of the period with custom-designed chairs for the control room—that oddly echoed those of the Star Trek television series a decade earlier—and Teletext machines to enable immediate nationwide communications [16]. In fact, Cybersyn was itself four sub-projects; Cybernet, the national network of telex machines, Cyberstride, the software system that created alerts when a variable fell outside an acceptable limit, CHECO, an attempt to model the entire Chilean economy, and the OpsRoom, the control room at the center of the system and based on the concept of the war-room with seven chairs facing each other representing separate elements of the economy that were being reported [18]. Ultimately, Cybersyn is an unfinished and unproven experiment as the Allende government was soon overthrown in a military coup, and Beer was forced to flee Chile.

The Cybersyn approach is instructive for a rethinking of management simulations in the way that it (necessarily) separates inputs, decisions, actions, and outcomes. The management simulation puts the student into the control room where they can see their decisions produce consequences in terms of resultant short-term and longer-term outcomes. The unfinished and inaccurate CHECO sub-project [18] also reinforces how difficult it can be to accurately define multiple models of different aspects of an economic situation that must also necessarily interact with each other.

## *Creating Effective Management Simulations: Rapidly, Responsibly, Relevantly DOI: http://dx.doi.org/10.5772/intechopen.106430*

Sometimes, any decision is better than no decision, particularly in an environment that is permeated with incomplete data. The Cybersyn experience also raises a more mundane question for management simulations. This is a question of the default settings offered for each decision at each turn. Many simulations may force a student to make a decision each turn with a blank input box or a slider set to zero. Some simulations will use the student's previous inputs as the default for their next turn—even if other circumstances have subsequently changed. Far fewer simulations will offer a "median" default option that would provide a "steady state" outcome for the turn being played. This approach would mean that the most neutral response would neither disadvantage nor advantage a student unexpectedly, something that is possible in the other scenarios for setting default values. This perverse situation can be occasionally spotted in practice too, when a simulation that uses the input of other students to calculate the current environmental situation accepts extreme inputs from disengaged students (such as "0") and unwittingly rewards them as inaction was ultimately, and unexpectedly, the most suitable decision for that particular turn.

These examples all highlight different ways in which simple algorithmic validity is not the sole or even primary basis for a successful simulation. The way the game world is created and enables students to interact and compete with one another are equally crucial factors for a successful simulation. These are the gamification elements found inside a simulation. But even these considerations still ignore the way a simulation might actually be played as a game in the way that it might be popularly understood outside a university environment. The majority of simulations are text-based affairs with inputs based on variants of HTML based forms, for example, radio buttons, checkboxes, dropdown menus and free-text boxes. There is a general absence of a rendered, flowing or interactive game world.

These are the multiple areas in which all management simulations need to progress. But also important is the need to re-introduce the capability for academics to create simulations for their own specific needs rather than needing to resort to a productized and generalist simulation that best matches their requirements.
