What Aliens taught us about finding a Storyform

A test with the Aliens screenplay showed how much Narrova can pull from a script and where finding the right Storyform still takes human judgment.

What Aliens taught us about finding a Storyform

We’ve been testing a new way to bring a screenplay into Dramatica. Drop the script into a project, and Narrova can assemble a Narrative Context Package (NCP) with the screenplay, a Storyform, and passages that show how the script supports individual storypoints. In our test, Aliens took about fourteen minutes with the published Storyform available as a reference. The package included evidence for 75 components and imported into a fresh project with the analysis intact.

Our test involved giving Narrova, our narrative intelligence, the screenplay while withholding the published Storyform, so as not to skew its findings. Narrova found Ripley, Newt, their relationship, and the Domains for all four Throughlines (Domains are the sources of conflict within the narrative). The run summary reported matches on eight of sixteen Signposts, but the model picked a different Storyform than the official canonical one in our library. The output we captured put Steadfast next to Ripley’s Resolve and Time next to the Story Limit. These were clearly incorrect—Ripley has a Change Main Character Resolve and the Story Limit is Options, not time. That said, Narrova did correctly identify her as a Do-er and the Story Driver as Action. The gap between a correct answer and the initial finding starts to reveal itself in these two primary questions about the same character. Ripley taking the wheel of the armored carrier helps show how she approaches problems and is easy enough to identify as evidence for a Do-er Main Character. To figure out her Resolve, you have to follow what happens to her way of dealing with the threat over the course of the film.

Where Control starts to explain everything

Aliens gives you plenty of material for a reading about control. The Company tries to contain and exploit the creatures. The Marines arrive expecting their weapons and chain of command to keep the situation in hand. Both efforts go badly. The blind run selected Control as the Problem. Those scenes illustrate why Control can sound persuasive, and how Uncontrolled, often expressed in ordinary language as freedom or letting go, can seem like a natural counterpart.

Over and over again, we find that frontier models reach for that same pair. It’s a plausible reading of many individual scenes, which makes the tendency easy to miss—but it’s one that exists because of the bias inherent in the training runs used to develop these models. In Dramatica theory, the Problem is the element that keeps generating conflict. Ask what starts the trouble in Aliens, and the Company’s desire to profit from the creatures takes you further. Burke sends the colonists to investigate the coordinates with the aim to get live specimens past quarantine. This Efforts to contain the creatures serve something he wants. Pursuing it repeatedly puts people in danger.

It also coincides with Ripley’s personal Problem with Desire. Her fear and loss and desiring of being a mother holds her back from taking right action. And the aliens themselves are reacting because of a desire to protect their own young.

These are meaningful instances of conflict that reflect the overall thematic intent of the piece—something that Dramatica excels at, and something that frontier models sputter on left to their own devices.

The published Storyform for Aliens pairs Desire with Ability in both the Objective Story and Ripley’s personal Throughline with Ability being the Solution to both Problems. We see what Ability looks like when Ripley recognizes the danger of firing inside the atmosphere processor, takes over the evacuation, and later uses the power loader to fight the queen. The loader pays off a skill she demonstrated earlier. At the climax, that skill gives her a way to protect Newt and confront the source of her own nightmares.

Ripley’s resolve and the story’s limit

The blind run called Ripley Steadfast. One way to see the appeal of that reading is to follow her warnings. She understands the danger early, holds her ground when others dismiss her, and turns out to be right. The published Storyform, however, gives her a Change Resolve. To see that change, follow her personal relationship to the threat. Early on, the encounter is something she relives in nightmares. By the end, she can face the queen directly, get Newt out, and even imagine sleeping and dreaming again. Her judgment of the aliens holds up throughout. Her way of dealing with what happened to her changes.

One reading was done in isolation and superficial; another is deeper and more meaningful when compared against the entire narrative.

Our test run also chose Time. Aliens supplies a particularly convincing piece of evidence for that reading in the reactor countdown—yet the Limit of a Story is not just one Act; it’s the limit on the entire narrative itself. The published Storyform sees Options. Look across the whole story and watch the viable routes disappear. The first dropship crashes. The survivors’ defenses fail. Newt is taken. The survivors have to bring down the remaining dropship remotely to get off the planet. There’s a clock in the climax, and it puts pressure on a set of choices that has already narrowed dramatically.

That distinction is easy to miss when you’re inside a character’s experience—which is where these frontier models have found their training. When the story is running out of options, it can feel to the characters like they’re running out of time—and so a model, trained on author’s subjective insights and the blended storytelling of several narratives erroneously classifies the limit as one of Time. When a story has a fixed deadline, characters can feel like they’re running out of options as the deadline rules out one plan after another—that’s not Aliens. As writers, we spend a lot of time inside that immediate experience. To identify the Story Limit, a successful narrative analysis needs to step back and ask what actually runs out and forces the story to its conclusion.

And that’s where our fine-tuning of Narrova arrives to save the day.

What we’re evaluating in Narrova

A Storyform has to hold up from four perspectives. In Aliens, there’s the expedition and the survivors’ shared predicament, Ripley’s personal experience, the pressure Newt’s perspective puts on Ripley, and the developing bond between them. Each Throughline needs evidence of its own. When a model treats Ripley’s urgency as evidence that Time governs the entire story, it lets one perspective answer a question about the whole.

That’s where much of the work on Narrova happens. We can run a script with the published Storyform available to see how well Narrova finds evidence for it, then withhold the reference to see what the model selects on its own. We compare individual Storypoints and look for patterns across scripts. When a model favors Control, Steadfast, or Time, we want to know which scenes made that answer plausible and which scenes might help it reconsider. An explanation gives us something concrete to test. The scenes still have to earn the conclusion across all four Throughlines.

We can then feed these findings back into Narrova and continue the process until our narrative intelligence becomes a useful and objective collaborator—and one available nowhere else.

For now, us humans still need to review the proposed Storyform—after all, stories are instruction books for humans; who better to judge them? Even with the published reference in hand, we can see further errors in the interpretation of Newt’s early Memory Signpost and the exact wording of Burke’s commercial motives. The new workflow available in this NCP-to-Storyform approach gives us a package we can inspect, with the screenplay evidence close at hand. That makes it easier to see where the analysis holds up and gives us specific cases to bring back into the next round of evaluations.