Systems Thinking:Thinking in loops

5.4 Thinking in loops

One of the more difficult aspects of systems thinking is to stop thinking in terms of cause and effect and to begin to think in terms of loops and circles of causality. We can see how loop-thinking works in the example of the steersman. The simple, cause and effect view says, 'I set the position of the rudder on a heading (cause); this makes the boat head toward my destination (effect)'. As we have already seen, this activity is rather more complex than it

seems. As our boat makes progress through the water we monitor the gap between the current course and the desired course and make adjustments to bring the boat back on course. Larger changes to the position of the rudder will result in larger changes in the course of the boat. Such a navigational system has five variables:

• The desired course

• The current course

• The gap between the desired course and the current course

• The rudder position

• The magnitude of the change in direction.

These five variables are linked in a loop diagram in figure 5.4 (see Senge's (1990) The Fifth Discipline for further details). The arrows represent influences: the rudder position influences the direction of the boat, which affects the course. The change in the course of the boat affects my perception of the gap and as the gap changes then my hand changes the position of the rudder ...

Causality suggests that I am in control and that there is a one-way arrow of cause and effect through which I cause the course of the boat to change. But this is only half the story, that is, setting the rudder position determines the direction of the boat and the current course. It is also possible to see the other half of the story, which says that the current course of the boat affects the perception of the gap which influences the position of the rudder. In this version of the story the course of the boat is controlling the movement of my hand. The whole story is that there is intent to set a course in order to reach a specific destination (the purpose of the system) and that the system allows the direction of the boat to change course when a gap between the desired course and the current course is perceived. The pattern of organization of the system gives rise to the behaviour that can be observed, i.e., the emergent property of steering a boat on a course.

5.4.1 Feedback loops

All the elements of a system are connected, either directly or indirectly. Since feedback is a loop, any change in one element of the system will lead to changes in all the other elements; as these elements are changed so the effect of these changes will ripple out in turn to affect the original element. The influence, modified by its journey, returns to the originating element in a loop, making the discernment of cause and effect difficult – it depends on where one starts in the loop and from where one is observing.

Negative (or balancing) feedback

Figure 5.4 is an example of a feedback loop. In this instance it is an example of negative, or balancing, feedback. In balancing feedback, changes in one part of the system result in changes in the rest of the system that limit or oppose the initial change. We use the term balancing feedback to avoid the sense of something bad that can accompany the word 'negative'. Balancing feedback is neither good nor bad; it simply helps to maintain the stability of the system by reducing the gap between what is desired and what currently exists.

For example, the treadmill in a gym has a speed setting that is used to set the running speed in kilometres per hour (kph). This is a balancing loop, as indicated by the scales symbol (figure 5.5). Starting at the gap, the difference between what is desired and the current state of affairs, there is a difference between the speed I want to run at and the speed of the treadmill. The action taken is to increase or decrease the speed of the treadmill depending on whether the desired speed is less than the current speed or greater than the current speed. The balancing process will continue to work even if the desired speed is moving, as would be the case as the runner gets more and more tired.

The treadmill in a gym does not respond to requests to change the speed instantaneously. When the increase speed button is pressed the speed of the treadmill does not make an immediate and step change to the new setting; the speed increases slowly at first and then more rapidly the longer the increase speed button is held down. Because nothing seems to be happening at first, it is only natural to hold the button down, but that leads to the speed surging and increasing too rapidly and the target being overshot (figure 5.6). The more frantically one stabs at the speed control the longer it takes to get to the right speed – the actions produce instability and oscillation in the system.

With the treadmill the delay is small – a couple of seconds. Some delays are very short and the feedback immediate (try putting your hand on a hot car exhaust pipe). But in other cases the delay can be substantial. For example, after a hard session on the treadmill and the weights muscle pain and stiffness set in 48 hours later rather than the next day. Other delays are even longer, as workers in asbestos producing factories found out when lung cancer appeared many years after they had stopped working with asbestos.

Delays can lead us into adjusting too late and by too much, destabilising the system as it swings from one extreme to another. To deal with delay we have to either get more immediate feedback (perhaps the way the speed mechanism on the treadmill works can be changed) or take the delay into account when making adjustments. Over time, one learns to compensate for the delay of the speed control, but possibly only after having had a nasty shock in trying to keep up with a runaway treadmill.

Positive (or reinforcing) feedback

Reinforcing, or positive feedback, occurs when changes in the system feedback amplify the original change, leading to the system, moving it away from its steady state. A small change can be amplified into large consequences. This 'snowball' effect is illustrated in figure 5.7. With a good web site more site visits mean more satisfied users, which means more positive word of mouth (an important factor in driving up web site accesses). Reinforcing feedback can lead to exponential growth, but does not always do so. However, reinforcing feedback always amplifies change in the same direction, such as increasing the number of site visitors.

The web site visitor growth is a virtuous cycle, but it is also possible for it to be a vicious cycle. Imagine that those who visit the site hate the design and give poor word of mouth reports, leading over time to less site visits. The change is amplified in the same direction, but this time it is a downwards spiral.

5.4.2 Feedforward loops

In balancing and reinforcing feedback, cause and effect go in circles. Feedforward takes into account our ability to anticipate the future. This is the case in management when action is taken on the basis of forecasts or where disturbances are anticipated that cannot be controlled (figure 5.8). The process and the controller have been separated in figure 5.8 with a classical feedback loop whereby the controller monitors the outputs of the process and uses this information to set the values of the input variables.

An example of this is a sales manager responsible for the activities of a sales team. The manager makes adjustments to the current inputs based on the outputs from the sales processes. There will also be uncontrolled and possibly unexpected inputs to the process, such as a product failure at a customer site, which might require the department's activities to be rescheduled in the light of adverse publicity. The sales manager can also anticipate uncontrollable events, such as budget cuts to the sales department that arise from a drop in sales as result of a downturn in the economy. The anticipation of a downturn might lead the manager to take action in advance of the imagined budget cut, such as to reduce the number of sales staff. The cut in sales staff contributes to the sales target not being met and the likelihood of budget cuts – a self-fulfilling prophecy.

In short, when you expect to succeed you often do – nothing succeeds like success. However, the converse is also true – nothing fails like failure. The easiest way to have an IS development failure is to convince yourself in advance that it will never work and that the users will resist it. An example of the self-fulfilling prophecy arose in the UK in September 2000 when lorry drivers protesting against what they considered to be the high price of fuel blockaded fuel depots. The outcome of this action was that it was difficult to get fuel to petrol stations and as a consequence there were long queues at the stations and empty pumps. However, the petrol pumps emptied much more quickly than they need have done as a result of panic buying and motorists queuing just to squeeze a few more drops of fuel into their cars. The expectation that petrol would run out ensured that it did so more quickly than it need have done. Our fears (or hopes) for the future can lead to the very situation we would wish to avoid. A couple of days after the fuel crisis had been resolved, a radio announcer in South Wales commented (mistakenly) that the blockades might start again. Other radio stations picked up the story and

the next day the spectacle of queuing motorists was back again. Now, the rumour was wrong, but petrol stations do not hold enough fuel to meet the demands of panic buying and a very real shortage was manufactured from a false rumour (figure 5.9). What we and others believe and take action on does indeed shape our future.