negative feedback
"Feedback" means that a process has a flow that comes back to itself.
Feedback can occur in a single process; or in a sequence of processes. In a long, complicated sequence the feedback can sometime be hard to see. For example, this systems model contains two feedback loops:
Feedback is important because it tends to generate a variety of behaviors that affect the entire system. It can make systems grow, or it can make them shrink. It can cause explosions, or it can permit the system to be controlled.
Negative Feedback
"Negative" feedback describes a situation where the effect of the feedback causes the system to shrink. This constant contraction is usually stopped only if the size of the system reaches zero.
As an example, consider a lake, fed by a river, and located in a desert. The river is the only supply of water flowing into the lake, and the climate is very hot and dry.
Water leaves the lake by means of evaporation. As the water evaporates, the lake becomes smaller. Because it is smaller, and because the conditions of heat and dry air remain the same, the size of the lake shrinks even faster. Eventually, if the river cannot keep up, the lake will disappear.
(The minus sign in the middle of the loop is a common way of showing negative feedback.)
As with many systems, graphs provide a good way to compare the same system under 
different conditions.
Suppose, for example, that the river floods early in the the year and then dries up. This puts a large amount of water into the lake, but then the supply shuts off. The next graph shows what would happen. First the river fills the lake with 100,000 acre-feet of water. Then, as each month goes by, the level of the lake drops. At first it drops slowly, but as it shrinks it evaporates faster ... until at the end of five months it is entirely gone.
Now suppose that the river has periods of high flow, and that these come five months apart. The water will flow in, then slow down, then speed up, then slow down, and so on. If the evaporation is constant, then the level of the lake will go up and down with the river.
Notice that both of the graphs can be produced by the same diagram. The difference in the amounts of water is due to varying rates of flow in the river and not to the organization of the system itself.
This is a very important observation. Systems thinkers draw diagrams and construct systems models in order to describe the structure, or the basic unchanging aspects of a situation. Later on, they vary the values of the parts of the situation that do change, and they watch to see what happens. This kind of information is often very helpful when trying to solve a complicated problem.
A Grassy Field
Rain and sun fall on a field of grass. The water and energy that they provide encourages the grass to grow taller. As the grass grows, it 
uses nutrients from the soil. As these are used up, the grass grows more slowly. If the nutrients are entirely used up, the growth stops.
(This is a simplification. In reality there are other factors that affect the growth of plants.)
The graph of the output of this model shows that the grass starts slowly, then accelerates (positive feedback applies to the growth process ... but it is not shown in this diagram), then slows down and tails off to zero.
This is the first time I have used the term "output" of a model. There are three terms that need careful definition:
The model is what the system looks like, the output is what it does.
That, too, is a bit over-simplified ... but it will work for now.
My Lawn
My front yard is similar to a grassy field. But it has one additional feature ... periodically, a 
human comes along and mows it. Because of this, even though there are plenty of nutrients, the grass never goes above a certain height.
In the diagram, look at the diamond-shaped box. 
This is a decision ... a new kind of element. Decisions are usually yes/no questions. In this case I ask if the grass is more than four inches tall. If not, then I let it grow. If yes, then it gets mowed back to two inches tall.
Notice that there is no minus sign on the mowing loop. This is not a negative feedback loop. Negative feedback produces a constant reduction ... negative growth ... in the system. In this case the system is not constantly shrinking; it is undergoing a forced reduction to a particular size. The graph illustrates this behavior. The grass never gets lower than two inches high, but every time it gets above four inches, it gets cut.
In the grassy field example, I focused on the nutrients. I did this because there was nothing else inhibiting the growth of the grass. In the lawn example, while the nutrients still matter - without them the grass would not grow - it is the mowing process that determines the behavior of the system.
It is up to the system designer to choose which aspects of the situation are to be included in the model. Usually, the choice depends on what the designer is trying learn about the situation ... that it, the choice often depends on the problem that the designer is trying to solve.
A Forest
Rain supplies water and sunlight supplies energy to a forest of trees, each of which consumes nutrients in the soil as it grows taller. Tree reproduce slowly, but over time the forest grows thicker.
The forest is home to a species of honey bee. The bees are essential for their ability to pollinate plants that are grown in the surrounding farms.
Once a year, a team of lumberjacks comes to the forest to cut down and remove a number of trees. The lumberjacks want to be sure that they do not cut down too many trees and harm the habitat of the honey bees.
This situation is similar to my lawn. As an exercise, change the lawn diagram to show how the lumberjacks might deal with their problem. What is their problem, exactly? What would the graph of the system's behavior look like? (You will benefit from working this out on your own, but you can look here for a solution.)
Remember This
When a system is shrinking, look for negative feedback. Models and diagrams show the structure of a system ... changing the variable aspects of the system can then produce interesting differences in the results.
Home - Diagramming Topics - Next: Filters and Sorters