|
||
| Planners
as Nonlinear and Complex Explorers
|
||
| Exploring Fitness Landscapes Using the N/K Model Yet, adaptation need not take place on a purely random landscape. To envision nonrandom fitness landscapes whose contours reflect the underlying nonlinear and complex dynamics among the components in a system or ecosystem, Kauffman has developed a N/K model of adaptation. In this N/K model, N = number of traits (such as bowed or straight legs, webbed or separate toes, long or short feet) and K = the number of inputs from other genes (which is a measure of the dependence of traits on one another, i.e., the nonlinear coupling or feedback among the traits). Kauffman adds this K parameter since the contribution of a single trait to adaptability may depends on other traits (e.g., the contribution of bowed legs to adaptive fitness may simultaneously involve whether the feet are long or short e.g., if N=3 and K=2, the genome has three genes each of which is effected by two others). Using this model, one can alter K as if twisting a control knob and observe what happens as the landscape deforms. As K increases, the more interconnected the traits or modifications are, so there are more conflicting constraints and, thereby, the landscape becomes more rugged with more local peaks. Unlike a landscape with one large mountain representing a very high value of adaptiveness, in this more rugged landscape, there are a large number of modest compromise solutions rather than a perfect one. In organizational planning, an analogy can be found in the Boston Consulting Group (BCG) portfolio analysis of products or business units. In the BCG portfolio grid, business units or products are grouped into four sectors which are really another way of talking about their adaptive value: stars; cash cows; dogs; and question marks. All four may represent compromise solutions, even stars and cash cows because it is undecidable from the grid alone whether the star or cash cow represents a high optimal peak or is trapped at a local peak. Most planners get stuck at that point, whereas the nonlinear fitness landscapes promises a way to envision the adaptive value of even currently highly productive products or business units. Adaptation becomes more difficult as K increases to its maximum value, N-1, where every gene affecting every other so the fitness landscape becomes completely random. In such a random fitness landscape, an adapting organism gets trapped at very low peaks, and the rate of improvement slows; thus, adaptation to highest peak becomes virtually impossible. This can be seen in biological as well as technological evolution since they are processes that attempt to optimize systems riddled with conflicting constraints (Kauffman and Macready, 1995). In such a situation, foolish adaptation, i.e., moving down a fitness slope, may be paradoxically advantageous since it frees up those modifications trapped on lower valued short peaks. (We will come back to this idea of foolish adaptation later to see how planners may be able to exploit it.) In a moderate degree of ruggedness, the highest peaks can be scaled from the greatest number of initial positions, so an adaptive walk is more likely to climb to a high peak than a low one (i.e., the basins of attraction for the high peak as attractors are larger than for lower peaks).
|
||
|
Next | Previous Copyright © 2001, Plexus Institute Permission |