© 1990 by British Computer Society
| ||||||||||||||||||||||||||||||||||||||||||||||||||
Simplicity and Power – Some Unifying Ideas in Computing
School of Electronic Engineering Science, University of Wales, Dean Street, Bangor, Gwynedd, LL57 1UT, UK
The article develops the conjecture that the organisation and use of any kind of formal systems, knowledge structure or computing system may usefully be seen in terms of the management of yredundancy in information.
Every formal system, knowledge structure or computing system has two key dimensions - simplicity (complexity or size) and expressive or descriptive power - and there is a trade-off between them. The balance between simplicity and power corresponds to a balance between OR relations and AND relations in the system.
The efficiency of the system (the ratio of power to size) depends on the extraction of redundancy from the system. Key mechanisms for the extraction of redundancy are pattern matching and search (by hill climbing or equivalent mechanisms) for the greatest possible unification of patterns.
These principles are the basis of a proposed new language and associated computing machine, called SP, which combines simplicity with high expressive power. SP is a Prolog-like pattern-matching system well suited to high levels of parallelism in processing.
In SP, the boundary between ‘knowledge engineering’ and other kinds of information engineering breaks down. In SP there is the potential for full integration of artificial intelligence, software engineering and other aspects of computing - with Shannon-Weaver information theory as a unifying framework. SP also offers a bridge between ‘connectionist’ and ‘symbolic’ views of computing.
Received May 1987. revised September 1988.
* School of Electronic Engineering Science, University of Wales, Dean Street, Bangor, Gwynedd, LL57 1UT