Entropy's inevitable increase in a closed system may be a metaphor for
pessimism about the future of organized systems, but it is not a scientific
argument for the impossibility of increasing organization.  Artifacts such as
programs are not closed systems; in this they resemble natural self-organizing
systems such as life.  

Energy is applied from outside these systems to increase the organization of
the system.  The difficulty of software evolution is not even vaguely
thermodynamic, since it does not lie in the inability to apply resources to
increasing organization.  A great deal resources are expended on the
development and application of software; the problem is that existing software
products fail to live up to expectations.

The goal of Gwydion is minimize the cost of the larger system which uses the
software product as a tool.  We intend to do this partly by supporting an
evolutionary lifecycle model that eases the continuing development of relevant
functionality, and partly by trading hardware for manpower during early
development via high-level programming and dynamism.

Hypercode tools implement the evolutionary lifecycle by recording and recalling
the design record, and by supporting emergent organization of the program.

The real problem is brittleness:
    Evolved systems often seem to reach a point where they become impossible to
    modify because making any change causes severe ripple effects.

Causes of brittleness:
 -- Poor system architecture and modularization allow ripples to propagate
    widely.
 -- Poor organization and design documentation make it hard to maintainers to
    anticipate the implications of a change.
 -- An evolutionary programming style creates components with initially unused
    generality.  This generality is invaluable for later modification, but when
    a general mechanism is used in new ways, it is fairly common for problems
    in analysis, design and implementation to be discovered.

Causes of complexity:
 -- Gratuitous (pseudo) generality
 -- Creeping featurism
 -- Architectural decay

Emergent organization:
 -- Not free, but usually cheaper than throwing it all away and starting over
    again.  Not only does it save software costs, but it preserves what is 
    good about the existing system.
 -- Component-wise re-engineering
        specialization
        obsolete component deletion
 -- Architectural synthesis
      - develop "big pictures" with increasing explanatory and predictive 
        power.
      - re-engineer component interfaces to fit the overall organization.
 -- Facilitated by tools for architectural analysis and ripple-effect
    discovery.