Before the Dramatica software was released – before the theory was completed, we ran into a road block. We had completed the 3-D version of the model and were trying to plot the progression of various stories in each quad to see if we could find patterns of sequence that could predict how a story should unfold.

Though we could see circular patterns in some quads, Z patterns in others, and hairpin patterns in others, they varied from quad to quad and story to story, not only in placement but in the direction in which events progressed along the patterns through each quad.

In short, we were going nuts trying to understand something that felt like it made sense but appeared to be completely chaotic. This was a very depressing time – to get so far and be stuck for so long. And then, one weekend, I took my daughter to the Museum of Science and Industry at Exposition Park in Los Angeles.

There was an exhibit of hands-on science experiments for children. One of these was a row of 21 bar magnets on spindles. You would turn one magnet and the one next to it would turn in response, and so on, until eventually you could get all 21 turning in sync at once.

While we were using this, it struck me – the patterns in the model weren’t to be understood by plotting them on the chart. Rather, all patterns were really just circular around the quad, but the pressures that built tension in a story flipped and rotated the items in a quad, and that each quad affected the ones around it like the row of magnets did.

Armed with this epiphany, I returned to the model and began to write the algorithms the ultimately came to drive the story engine.