A filter is a device which stands **between action and perception** or the reverse. A filter may amplify (increase) or attenuate (decrease) a signal. The choice of filters is a major factor in the observer's definition of a system and its purposes. Our sensory organs filter out much that happens in the world around us that is either too small or too big to see, to subtle to touch, too fast or too slow, too high or low in frequency to hear, and so on. Filters may be intangible as well as tangible. The choice of logical levels or of a mathematics to use in a given situation determines what can and cannot be expressed. Language itself can act as a filter; much twentieth century complexity is inexpressible in, say, Old German. We may use a green filter on the lens of our camera to reduce the 'distortion' caused by bright sunlight. We may wear rose colored glasses when we have made a commitment to an ideology of optimism or do not wish to see indications that all is not well.
When information passes between a system and an environment, the transducer acts as a filter and must be consciously designed so that it does not reduce variety by filtering out the wrong information. In our normal activities, we filter out most of the sensory data around us. Otherwise we would be too overloaded to concentrate on any one activity. Both human beings and their organizations vary the sensitivity of their filtering processes to fit immediate circumstances. If we are intensely engaged in a task, our selection processes are set to filter out 'irrelevant' data. When the task is finished or we are more at ease, we will become aware of many more sensory perceptions. Without conscious attention to the way in which they are selecting or filtering out information, an organization under crisis management or an individual who is overextended may **screen out messages vital to their longer term well-being or even survival**.
In management, filters are necessary to attenuate the variety of the flow of incoming information and to amplify the important signals. Statistical filtering packages may be designed to screen indices for signs of incipient instability or to send an alert to notify the manager when non-routine information arrives. A staff usually filters out most of the data about the organization before it reaches the chief executive who needs to make a decision. If the staff is a good one, it will filter out the unimportant information and highlight the important. The distinction between the information which should be filtered in and that which should be filtered out is crucial. In some circumstances, data may appropriately be averaged: in others 'averaging' may eliminate the essential information. The study of disasters provides a litany of examples of **filtering out exactly the information that should have been amplified**.
# SOURCE Filter comes from a Latin word meaning filled wool or felt which was used to strain liquid.
# EXAMPLES • the range of sound audible to the human ear • the selection of what to carry on the evening news • standardized achievement and attitude tests • the modeling process • the legal system • a decision by the marketplace
# NON-EXAMPLES • the range of curiosity of a small child • the message drowned out by noise . • information overload • a weighty management report
# PROBABLE ERROR • choosing the wrong filter for the purpose, • Not taking account of the effect that both education and belief both impose massive filtration on future inputs • Not recognizing how much the presence of a filter is affecting your assessment of a situation.
# SEE Variety; Observer; Feedback; Transducer; Channel capacity; Model