Storytelling is a wonderful human faculty that challenges generative AI, particularly when it comes to long, original works. One thing missing from these discussions is a rigorous definition of good storytelling, a standard that determines whether one story is better than another. Here, our Publishability Index™ (PI) might be useful. It offers a comprehensive criterion for humans to judge a story's quality. In so doing, it specifies what makes a story not so good, good or great.

The PI is the cornerstone of Bardsy's efforts to help our members write effectively. It was developed as part of a PhD thesis to evaluate creative writing instruction. We use it to judge Bardsy's writing contests and in other formative and summative evaluations.

It also provides a more precise language to give more feedback and to discuss the relative merits of stories. Scientifically speaking, it's an intersubjective criterion designed to increase the reliability and precision of coders' evaluations. We have had good results concerning intercoder reliability thus far.

In looking at the PI, you'll see it organizes the conventional wisdom regarding good storytelling into 23 dimensions - covering seven familiar elements, like character and plot. Use the PI as a checklist to evaluate a story. Start at the top and move down, dimension by dimension. Read the criteria for each, first “good,” then “great,” reflecting on whether or not that criteria is present or absent.

To illustrate, the character backstory dimension calls for characters to have a "past that contributes to their identity" to be considered good. If, after reflection, the author determines that this statement applies, they'd check it. If not, they'd move down to the next dimension. If the author checks "good," they can consider the “great" criterion, defined as "relevant histories that spur individual development and action." If their character lacks such relevant history, they should seriously consider this omission when revising.

When finished, the author (or judge) can tally the goods and greats to identify a story's strengths and also find its weaknesses. There's a tab for notes, too

The PI is intended to be used iteratively, to identify and correct problems with each successive draft until none remain. The name comes from that outcome. Should a story receive 23 greats, we'd consider it to be publishable. More explanation can be found here: https://bardsy.com/forums/blog/the-publishability-indextm-an....

AI seems to have difficulty with this kind of holistic reflection, let alone being able to apply this kind of multi-part standard to creative writing. Our question is whether you would find this useful, either as an adjunct to your own writing or in AI development.

We welcome your thoughts,