Archive for the ‘Control’ Category

Influencing a System by Finding its “Levers”

August 12, 2007

In the field of Complex Adaptive Systems (CAS) people talk about “finding levers” to influence a system [1].  This idea of a “lever” in a system has always puzzled me.  Now I understand!  What follows is an example.

Consider a road, say Route 3, which runs between New Hampshire and Boston.  The road is a “resource”.  Further, it is a limited resource – it cannot hold an unlimited number of cars.  During rush hour lots of people use the road, and travel can be quite slow.  What we have is a failure of individuals to cooperate on the use of this resource.

Route 3, the surrounding roads between New Hampshire and Boston, the cars, the buses, the trucks, and all the drivers collectively comprise a “complex system”. The system doesn’t work very well – during peak hours there’s lots of congestion  on Route 3, with little traffic on surrounding roads or during off-peak hours.

It would be useful to find a “lever” that could be applied to get people to cooperate better and thus have a smarter system.  Specifically, we want people to use other roads and stagger the times they drive on Route 3.  What lever that can be applied to this complex system?

Answer: charge a toll to use the road during peak hours. A toll will make the cost of using the resource obvious, and will force individuals to ask themselves, “Does the benefit outweigh the cost?”  Some people will answer “No” and use alternate routes or drive at non-peak hours.  Congestion is reduced on Route 3.  By introducing a toll we are influencing the system.  The toll is a lever.

A toll is a lever that helps eliminate traffic jams that result from pure congestion.  Other levers are needed to deal with road problems that are not due to pure congestion, problems such as accidents, construction, sun blindness, or a slow-moving truck in the right lane.

Can you give examples of other levers?

[1] We do not attempt to “control” the system; rather, we just “influence” the system.  Complex systems are typically not controllable.