An Evaluation of the Impact of Constraints on the Perceived Creativity of Narrative Generating Software

This work investigates the impact of constraints on the

perceived creativity of the output of narrative generating

systems, in order to understand what level of constraint

application results in the most creative output.

To achieve this, software is written that generates short

stories, using adjustable levels of constraint meant to

reflect those utilised by other narrative generating systems.

These systems are presented at different positions

along a spectrum, which we posit arises from the application

of constraint. The creativity of the output is then

assessed by human evaluators. The results are promising

and show a clear variation of response based on the

level of constraint imposed on the narrative generation

process. The results show a sweet spot for maximal creativity

closer to the less constrained end of the spectrum,

which demonstrates the potential for more creative software

by the relaxing of constraints.