|
EPSRC Network on Evolvability in Biology & Software SystemsSoftware Evolution and Evolutionary Computation Symposium Abstracts
University of Hertfordshire, Hatfield, U.K.
|
ANDREW LORD
[joint presentation with
ILFRYN PRICE]
Facilities Management Graduate Centre, Sheffield Hallam University
Memetics, drawing on Dawkins (1976) suggestion of the meme as a unit
of cultural replication, has begun to support a body of scholastic
investigation into evolving organisations considered as the memetic
'phenotype' (Price, 1995; Gell-Mann, 1996; Price and Shaw, 1998;
Williams, 2000). If organisations evolve (Aldrich, 1999) evolutionary
pathways should be testable by comparing their 'memetic' characteristics
and precisely what aspects, if any, play the 'memetic' role is, in
principle analysable. In biology molecular phenetics, the comparison
of genetic similarity between species, shows an ability to reconstruct
cladograms, classifications based on descent from common ancestors. Our
concern is with a methodology for drawing comparable classifications
in organisational milieu.MENDEL tm was developed as a bespoke tool for
reconstructing hereditary and lineage in sociological entities from their
traits using algorithms adapted from numerical taxonomy. The software
has been implemented within the Microsoft Excel dialect of visual basic
for applications (VBA) that allows a conventional spreadsheet interface
to be populated with a matrix of taxa - character states. The featured
coefficient of similarity is the Hamming distance, which provides
an absolute metric of digital error and unweighted pair-group method
using arithmetic averages (UPGMA) as a clustering strategy of minimal
complexity (Sneath & Sokal, 1973). The clusters are plotted as a
phenogram, which may be reinterpreted from the cladistic perspective
as depicting organisational descent. MENDEL was initially tested on
genetic algorithm generated data and later successfully reconstructed
the emergence of Christian denominational families (Lord and Price, 2001).
A cluster of firms entering a new market during the 1990s have been
shown (Lord et al. in review) to, in the main, cluster in a manner
that reveals common parentage. Tests are currently underway to whether
definitional stances to the term obsolescence applied to buildings, in
published literature, reveals a phylogeny which has any correspondence
to institutional and professional descent. Further opportunities to test
the method, and develop applications for non-binary data are being sought.
A parallel development EDEN-ML initially served as a test harness
for MENDEL by evolving synthetic Binary Encoded Meme Strips for
reconstruction. Further developments of this module are planned. These
will include the ability to explore combinatorial meme space thereby
finding uninhabited peaks on a fitness landscape and to navigate an
optimum path between peaks.
The practical organisational application of these enhancements is
the provision of insight into future market direction which acts as
a basis for business strategy and structuring. As information systems
are inexorably linked to business needs then anticipation of change can
assist in establishing software requirements. Tracing and simulating
memetic evolution, using MENDEL and EDEN-ML, may therefore contribute
to more effective software systems development methodologies.
This is a joint presentation with:
Ilfryn Price,
Facilities Management Graduate Centre,
Sheffield Hallam University,
Unit 7 Science Park,
Sheffield S1 1WB, U.K.
Email: I.Price@shu.ac.uk
MENDEL(tm) = Memetically ENcoded Derivation of Evolutionary Lineage
EDEN-ML(tm)= Evolutionary Dynamic Emergent Node - Modeling Language