Chapter 8: The Architecture of Complexity
complex systems --->
one made up of a large number of parts that interact in a non-simple
way
main points of this chapter:
- frequency with which complexity takes the form
of hierarchy
- relationship between the structure of a complex
system and the time required for it to emerge through evolutionary
processes
- compare clues in problem solving to stable intermediate
forms (e.g.: the defect safe, reuse in object-oriented systems)
- relationship between complex systems and their
descriptions ---> there is no conservation law of complexity
Tempus and Hora
- watches of a 1000 parts - interruptions by phone
calls
- Tempus: interruptions lead to restart from scratch
- Hora: subassemblies of ten (at each level) --->
111 subassemblies
- the evolution of complex forms from simple
elements depends critically on the numbers and distribution of
potential stable intermediate forms
Problem Solving as Natural Selection
- problem solving requires selective trial and
error
- cues signaling progress (partial results, stepping
stones) = stable intermediate forms
- safe with 10 dials, each with 100 possible settings
working: 10010 settings ---> average:
5 * 1011
defect: 10 * 50 = 500 trials
Nearly Decomposable Systems
- Definition: the interactions
among the subsystems are weak but not negligible
- hierarchical systems have the property of
near decomposability: intracomponent
linkages are generally stronger than intercomponent linkages
- examples: the search
in computer science for systems architectures which allow to build
nearly decomposable systems (further example: structure of learning
environments, e.g., "skiing")
- near decomposability and comprehensibility
(a "chicken and egg problem"):
are we able to understand the world because it is hierarchic --
or does it appear hierarchic because those aspects of it which
are not elude our understanding and observation?
Simple Descriptions of Complex Systems
if a complex structure is completely unredundant
(i.e. if no aspect of its structure can be inferred from any other)
then it is its own simplest description
forms of redundancy:
- hierarchic systems are composed of only a few
different kind of subsystems (restricted alphabet of elementary
terms)
- hierarchic systems are often nearly decomposable
- "recoding" ---> the redundancy that
is present but unobvious in the structure of a complex can often
be made patent
- the task of science is to make use of the world's
redundancy to describe that world simply ---> the descriptions
are part of the "Sciences of the Artificial"
State Descriptions versus Process Descriptions
a circle is the locus of all points equidistant from
a given point
examples: pictures, blueprints, diagrams, chemical
structural formulas
characterize the world as sensed
to construct a circle, rotate a compass with one
arm fixed until the other arm has returned to its starting point
examples: recipes, differential equations, equations
for chemical reactions
characterize the world as acted upon
- declarative ("what") versus procedural
("how") controversy in Artificial Intelligence
Ontogeny Recapitulates Phylogeny
- the individual organism in its development goes
through stages that resembles some of its ancestral forms
- one way to solve a complex problem is to reduce
it to a problem previously solved
- catalogs, case-based memories
- phylogeny: domain-oriented
design environments (e.g., computer network design)
- ontogeny: individual
design project (e.g., the computer network here at CU Boulder)