Semiautomatic design generation systems are more familiar than the fully automated variety, but have become marginalised during the twentieth century. Their general pattern is the creation of a set of rules and limits that allowing a designer to rapidly convert a first decision into a final design. The rules generally require design decisions at each turn, but those decisions are constrained. Constraints allow each decision to be made rapidly and decisively. This allows the designer to quickly develop and critique designs based on initial "moves" or decisions.
"Limiting languages" are, in great part, the attraction of these systems. The challenge facing a designer normally is not the search for a suitable design, but the search for an optimal design among a huge set of possible designs. By providing the designer with a selection of "acceptable" (close to optimal) solutions at every turn, these languages provide a method of "safe play," whereby a decent solution can always be achieved. Another way of looking at it is that the language describes what decisions need to be made at each level. This focuses the designer on those areas requiring attention and ignores those likely to be fruitless.
Looseness in these systems allow the end-designer to replace, add or refine elements in the pattern store with less work than that required by fully automated generative systems. Within those systems the task of developing new elements often requires arcane knowledge of the depths of the system. This looseness, coupled with the promises of efficiency and attraction which limiting languages make, encourages many designers to play with limiting languages, especially proportional schemes, at some point in their career.
Document Name: tc.genorg.gen.ruler.demiborg.htm
Copyright (c) 1997 -- Mike Fletcher
Reproduction for other than personal use prohibited without express written permission from the author.