Page 338 - J. C. Turner - History and Science of Knots

328 History and Science of Knots

3) Generative motif grammar: There are common motifs in crochet lace
that can be specified by a generative grammar. The grammar provides a
mechanism for incorporating high-level design information into the model,
although this mechanism is not yet fully implemented.

These elements of the model give us powerful tools with which to char-
acterize aspects of crochet construction and design. The process of building
these tools has in turn provoked speculation about some of the underlying
assumptions of crochet lace composition, leading to our experimentation with
variations on the traditional forms.
Is CADD itself creative? Artistic creativity has been defined in many
ways: as the capacity to produce compositions, products, or ideas which are
essentially new or novel, and previously unknown to the producer; as an em-
bodiment of originality, which encompasses uncommonness, remoteness, and
cleverness; and as the ability to make guesses, attempt solutions and perfect
solutions, and finally communicate the results. At present CADD is creative
in the sense of knowing `the rules' and being able to manipulate them to pro-
duce a physically plausible pattern. These rules are sufficiently general so as
to permit a wide variety of patterns to be formed.
This generality is one of CADD's drawbacks: generative models, unfor-
tunately, tend to produce a copious flood of possible patterns, many of them
uninteresting or aesthetically unappealing. We are currently considering the
style parameters and meta-design principles necessary to permit the user to
drive the generation of new patterns in novel, attractive directions.
However, in its current form CADD does enable an effective creative part-
nering between human designer and machine. In its `semi-automatic' mode,
CADD examines the current state of a design and proposes incremental exten-
sions to it. The user selects one of these extensions, thereby guiding CADD
and winnowing out the less promising patterns. The design process is greatly
enriched, however, since CADD's set of suggestions frequently contains ele-
ments that the human has not considered.
The process of building the construction rules is also useful for stirring the
creative processes in the human collaborator. As we codified the parameters
defining a particular type of lace-the symmetric, flat, circular doily-we were,
in essence, exploring the limits of that form. Indeed, one of the early versions
of CADD contained an incorrect algorithm for of the method by which the
number of stitches in a round increases as the doily grows in size; interestingly,
this version re-created the `ruffled' doily popular in the 1950s. Another version
of CADD contained an incomplete geometric model, and produced designs
in which rows could overlap. Several of these `mistakes' proved intriguing,
and have motivated experimentation with novel forms of three- dimensional
lacework. By consciously examining the parameters of a style, the human