4.2. Redesigning Your World of Showers
In World of Showers A, the plumbing in your five-star hotel is of the highest quality. It is well-designed so that adjustments to the tap setting correlate closely with movements of temperature, albeit with a time delay. In other words, your decisions lead to the outcome you intend (once you have mastered the impact of the delay). It is a manageable world of low dynamic complexity in which you have the freedom to act independently of others who may also be taking showers in the hotel at the same time.
In World of Showers B, the quality of plumbing in your two-star hotel is distinctly inferior. Comments in the guest register suggest you are not the only one who has had problems with the hotel's showers. Your friend has confirmed this view, recalling in particular her experiences on the day you both arrived: 'It was impossible to get the temperature right. The tap was totally useless. I almost ripped it off the wall.' As you peruse a litany of similar criticism from long-departed guests, you realise that the hotel's shower problems probably arise because different rooms unwittingly share the supply of hot water. Showers interfere with each other. Hence, if you need more hot water, someone else (invisible to you) receives less and vice-versa. It is a less manageable world of high dynamic complexity and interdependence. Cause and effect are not closely correlated and your actions inadvertently have an impact on others, which can worsen their performance relative to goal.
You have time on your hands, so you start wondering how this down-and-out hotel might improve its showers. Unfortunately, the faults of the hot water system can't be eliminated without upgrading all the pipes and tearing down walls and ceilings, but that's life. Competition for scarce resources is endemic and you can't rebuild society overnight. Why should a two-star hotel be different? So, what else could be done? You muse idly over this puzzle. It is another warm and sultry Spanish summer afternoon. Your thoughts drift ...
All of a sudden you find yourself back in your shower cubicle. You notice some new controls in the shower that were not there before. To locate these new controls, press on the button 'To Policy Levers' to reveal the screen shown in Figure 4.5. There are two slide bars that allow you to modify your shower world.
Figure 4.5. Policy levers for responsiveness and patience
The top slider alters the responsiveness of the plumbing by changing the pipeline delay (in seconds) between a turn of the tap and a change in water temperature. The shorter the delay, the more responsive is your shower and all other showers in the hotel.
The bottom slider gives you unimaginable power to influence personality – you can change the patience of any hidden shower-takers who may share your shower world! A patient shower-taker is someone who reacts only gradually to a temperature gap. Such a person is willing to tolerate some discomfort and therefore moves the tap only gradually when the water is too hot or too cold. The longer the time to adjust the tap (in seconds), the more patient the person. (You reflect momentarily on where you yourself lie on this scale of patience.)
Responsiveness and personality are set before you take your simulated shower and are then fixed for the duration of each run. They cannot be altered during a simulation. To reset the two policy levers to their default values, click on the 'U' button in each slider box. If you cannot see the button, the default value is already set. Select new values for your first policy design experiment. (Hint: To begin with, alter only one slide bar per experiment. That way you will have a clearer idea of the impact of each policy lever.)
To take another shower in your modified shower world, click on the button 'To Charts and Controls'. Note that if you press the 'Reset' button, in order to clear the graphs, then you will also restore the policy levers to their default values. If so, just revisit the policy levers and set them back to the values you intended. Then return to the charts and controls. To start the simulation, press the 'Run' button and experience 120 simulated seconds. Is your new shower world better? Has your final score improved? Do you feel more in control of your own destiny or more at the mercy of unseen system forces? In the light of your experience, redesign the system once again by selecting new values on the slide bars for responsiveness and personality. Play as many times as you like. Can you equal the best score you achieved in World A? Which design changes are most effective and best recreate the feel of World A? Why?
You hear a voice calling. You awake. You are in the hotel lobby with the guest book on your knee. The warmth of a Mediterranean summer still surrounds you. Gradually your mind clears. Those new controls were just a dream after all ... You don't have the power to change personalities or even plumbing, but your imaginary shower world left you a message. Hotel showers are a microcosm of typical organisations – lots of people, striving for a variety of different goals, fulfilling their needs and coping with dynamic complexity. You now appreciate that dynamic complexity can be managed. It is possible to shape an organisation (responsiveness, mix of personalities, incentives, goals) to empower people, so that they have the freedom to succeed as individuals while contributing to collective ambitions and strategy. Already you see your name on the General Manager's door.
4.2.1. Reflections on the World of Showers
Economists recognise the problem of 'invisible' dependency as an externality. Externalities occur whenever a person's decisions and actions alter the frame of reference for others (Schelling, 1980). Externalities pose two basic problems for effective management. The first problem is that inefficiencies systematically arise because decision makers cannot possibly take into full account all the costs and benefits of their interdependent actions (the assumption of bounded rationality explained later in Chapter 7). The second problem is that specific individuals and groups are often put in a position of defending suboptimal allocations that satisfy local interests, but are collectively worse than the target efficient allocation.
Feedback systems thinking deals with externalities by explicitly representing the hidden dependencies from which externalities arise. In a sense, they are no longer 'external'. Nevertheless, due to blind spots in their understanding of dynamic complexity, decision makers will behave as though they are facing externalities (Sterman, 1989).
Figure 4.6. Balancing loop with delay in World of Showers A
The World of Showers shows that you don't need much dynamic complexity for inefficiencies to crop up in the management of shared resources. Even in World of Showers A, most players have difficulty producing a smooth yet fast convergence to the desired temperature. Yet a fluctuating temperature is clearly inefficient – it causes more discomfort than is strictly necessary.
The dynamic structure that baffles players in World of Showers A is the so-called 'balancing loop with delay', introduced in Chapter 2 and reproduced in Figure 4.6. Typical players acting toward the goal of desired temperature adjust the flow of hot water in response to delayed feedback. If they are not conscious of the delay, or don't know how to take proper account of it (as is usually the case), they end up taking more corrective action than needed, causing the temperature to overshoot. Further corrective action may cause overshoot in the other direction leading to a cyclical pattern of temperature adjustment.
World of Showers B couples two balancing loops to produce a structure with considerably more dynamic complexity, as shown in Figure 4.7. Here the flow of hot water in either shower depends not only on the temperature gap experienced by the shower-taker but also on the flow of hot water in the other hidden shower. Besides the baffling effect of the time delay, there is the added effect of mutual dependence and invisible sharing of a common resource.
4.2.2. Metaphorical Shower Worlds in GlaxoSmithKline, IBM and Harley-Davidson
Interdependent structures of this kind repeatedly crop up in functional and divisional organisations. Imagine, for example, that two divisions of a firm arrange to market their products through a shared salesforce. This pooled arrangement is common in organisations like GlaxoSmithKline (GSK) or IBM that employ professional sales people to explain complex products and services to customers and persuade them to buy. Suppose that two drug-producing divisions of GSK (A and B) are pursuing ambitious sales targets. To achieve its target, division A must capture a larger proportion of salesforce time, thereby denying division B. As a result, division B's sales fall well below target prompting efforts to win back the salesforce. Salesforce time switches back and forth between divisions, leading to self-induced cyclical instability in sales and, in addition, costly manufacturing.
Figure 4.7. Interacting balancing loops in World of Showers B
A second example of interdependence occurs when product lines of a manufacturing firm share capacity, as shown in Figure 4.8. The diagram is based on a strategic modelling project with the Harley-Davidson motorcycle company (Morecroft, 1983). Motorcycle production and parts production both use the same machine tools to make components. The project team identified a performance puzzle surrounding the parts business. The company was losing market share in its highly profitable parts business even though competitors were just small job shops making look-alike Harley parts. A model to address this issue included the supply chain for both parts and motorcycles, extending from the factory to dealers and customers. The diagram shows the upstream factory end of these two supply chains. The explanation for loss of parts market share lies in the dynamic complexity of the interlocking supply chains.[] Bikes and parts share capacity (just as shower-takers share hot water) and the operating policies that control capacity expansion and capacity allocation are inadvertently biased in favour of motorcycles (the big, glamorous and politically powerful part of the company – even though return on sales of parts is higher than bikes). Moreover, demand in the two supply chains is amplified and correlated (demand for bikes and parts rises quickly in the springtime).[] The result is that the parts business suffers periodic capacity shortages and is unable to compete with job shops who can offer short and reliable lead times – a key competitive variable in any parts business.
[] There are many supply chain models to be found in the system dynamics literature, starting with the original production distribution system model in Chapters 2 and Chapter 15 of Industrial Dynamics (Forrester, 1961) and continuing in a recent special issue of the System Dynamics Review devoted to the dynamics of supply chains and networks (Akkermans & Dellaert, 2005). The Harley-Davidson model represents the bikes and service parts businesses as two supply chains that share capacity.
[] Incidentally, any individual supply chain contains vertically integrated balancing loops, each with a delay. This stacked structure leads to demand amplification and the bull-whip effect as characteristic dynamics of supply chains. In principle, the same effect should be produced by 'stacking' two or more shower systems one on top of the other so that turning the tap in shower 1 (to control temperature) depletes a tank of hot water observable in shower 2 and under shower 2's control. Turning the tap in shower 2 (to control temperature) then depletes a tank of hot water in shower 3 and under shower 3's control, and so on. This 'cascaded' World of Showers C would likely be even more difficult to manage than World B. A variation on this structure is for the outflow of water from shower A to be the inflow of water to shower B and so on – a kind of thrifty World of Showers C. The reader is invited to develop and test these novel Shower Worlds.
Figure 4.8. Managing product lines that share capacity
A useful management principle accompanies the troublesome feedback structure behind the business examples above, and echoes the lessons of the shower simulator. In a slow-to-respond system with shared resources, aggressive corrective action to achieve local divisional or departmental goals leads to instability and underperformance in the organisation as a whole. For better results, either be patient or reduce interdependence and make the system more responsive. In the case of Harley-Davidson, the solution was to reduce interdependence between the motorcycle and parts businesses by investing in a large finished inventory of motorcycles. With plenty of motorcycles in the supply chain at the start of the spring selling season there was less internal competition for scarce capacity. As a result, the parts business was much more responsive to dealer demand and able to offer lead times equal to or better than specialist job shops.