Fitness Landscape and other landscapes. LO27241

From: Gavin Ritz (
Date: 09/21/01

Replying to LO27222 --

Dear At and other Org learners.

I am very pleased you have discovered Stuart Kauffmans work, His NK models
are absolutely wonderful. Yesterday I had an email conversation with
Stuart about fitness landscapes (and baysian networks) and using it in our
business to map some of our complex production processes. For those who
are interested in his work. Origins of Order, and At Home in the universe.
Stuart's company is called BIOS, web site:


AM de Lange wrote:

> In a recent private E-mail discussion on the "cloning of stem cells", I
> came under the impression how an acquaintance to the idea of a "fitness
> landscape" can actually impair understanding the dynamics of free energy
> and entropy production.
> The notion of a fitness landscape was already introduced in 1930 by the
> geneticist Sewall Wright. Stuart Kaufman with co-workers incorporated it
> into his theory of Complex Adaptive Systems (CASs). They developed a set
> of complex models (called NK-models) to study algorithms for solving the
> problem how complex interconnected systems interact. The theoretical
> NK-model requires an "element dimension" with a number N of system
> elements as well as a "state dimension" with a number S of different
> states which each element can be in. Interconnections between elements are
> modelled by assigning each element to a "spillover set," consisting of K
> elements. A "fitness function" is then specified which represents each
> element's contribution to a system-wide variable. This fitness function
> has to have a limit (maximum or minimum) with respect to the configuration
> of the K elements in the spillover set.
> Many kinds of theoretical functions with limits and many kinds of
> theoretical algorithms for finding these limits are possible. An algorithm
> which Kaufmann et al found particularly attractive is the simple
> trial-and-error procedure known as the "adaptive walk." It is very
> efficient at finding the highest point on the fitness landscape in systems
> that have no interconnections or spillovers between elements (i.e., in
> systems with K = 1) so that an element's fitness contribution is a
> function only of its own state. But as the spillover effects increase (K >
> 1) so that the system becomes more complex, the algorithm performs
> progressively less efficient. The adaptive walk becomes increasingly
> trapped on what is called a "local fitness peak" (local optimum) from
> which there is no escape towards greater fitness.


Gavin Ritz <>

Learning-org -- Hosted by Rick Karash <> Public Dialog on Learning Organizations -- <>

"Learning-org" and the format of our message identifiers (LO1234, etc.) are trademarks of Richard Karash.