Design Goals

My research goal is to develop a dynamic “story” engine for games that requests meta-data about the game world from the game engine via a standardized callback API, and then creates sensible stories that fit into the game world. Ideally, the stories created by the engine would equal or even surpass purely static content, but simply augmenting static content would represent a significant advance of the state of the art.

A Turing Test for Dynamic Stories

The question is, can we create a story engine that creates stories that seem at least as hand-made to players as actual hand-made stories? Obviously, concocting a fair test for this is non-trivial, since we're either comparing ourselves to pre-existing games ("would you rate the stories you experienced as more/less original and clever than the stories in Baldur's Gate?") or to our own contrived "control" group ("do you like story A more/less than story B?" setting aside the fact we're responsible for both, and we want A to win).

The devil will be in the details. Can I/we operationalize such a comparison and document it in such a way that the comparison seems fair and compelling? (And, that's before tackling the harder job of faring well in the comparison!) I suspect that the methodology will be something like: compare hand-made stories to dynamic stories. Based on results, iterate, improving both kinds based on what was learned. This will tend to lead to increasingly demanding conditions A vs. B becomes A vs. B vs. A' vs. B'. In order to make our documentation of the process compelling, we will need to capture multimedia from actual gameplay (so when we compare a hand-made story to a dynamic story we can show what the player experience of each was actually like).

goals markdown 12/16/2014 14:58:39


Powered by FolderMark