How Does DramaFork Check for Mutually Contradictory Requirements During Planning?

The most common problem in planning is that each requirement looks reasonable on its own, but they undermine each other when put together. The check is not to fill out the creative sheet again, but to pull out five types of requirements and compare them pairwise: story goal, player identity, scale, form, and ending promise. The method is to first write them as judgeable short sentences, then ask pair by pair "can these two hold at the same time," write the ones that cannot into a conflict list, and finally use a contraction plan to change one of them rather than keeping all five.

Below, a fictional teaching example walks through the complete process. The world, characters, and numbers in the example are made up to illustrate the method, not measured data.

Suppose a project is called "Seven Days in Fog Harbor," with an initial plan written like this:

• Story goal: The player finds out the truth behind Fog Harbor's suspension of shipping within seven days, and decides whether to make it public.
• Player identity: A newly appointed port records clerk, with no law enforcement authority, who can only advance through records and dialogue.
• Scale: A single-scene short piece, estimated at about twenty minutes per playthrough.
• Form: An interactive film game, mainly dialogue choices and a small number of investigation nodes; the first draft also requires each chapter to span five cities.
• Ending promise: The player can clearly know what they changed, and see more than two endings.

These five sentences can all be judged true or false, which is far more useful than "suspenseful, immersive, with a sense of identification." The criterion is: if another author reads it, can they point out which piece of content violates it?

When comparing, there is no need to run through all five by five; first look at the three groups most likely to conflict: scale versus form, identity versus goal, and goal versus ending promise.

The first row is the typical contradiction this topic is meant to detect. "Single scene" limits the number of spaces, while "each chapter spans five cities" requires at least five locations per chapter; the two cannot be satisfied at the same time. This is not a question of whether the idea is good or bad, but a direct conflict in quantity.

The second row does not constitute a contradiction, but it changes how it must be written: the records clerk has no law enforcement authority, so "finding out" can only be completed through records, dialogue, and comparison, not through searches or arrests. This must be written into the constraints to avoid the protagonist suddenly breaking down doors during the script stage.

The third row is also not contradictory. The same definite truth can produce different endings depending on whether it is made public, kept secret, or handed to different parties. Multiple endings do not require multiple sets of mutually contradictory truths; what is needed is a difference between player decisions and visible consequences.

The conflict list is recommended to contain four columns: conflict number, involved requirements, conflict type, and which one must be changed.

C2 is only a risk record and cannot be disguised as a confirmed conflict. A single playthrough usually reaches only one ending, so the durations of two mutually exclusive routes cannot be directly added together. First list the dialogue and video length of each route, then judge whether twenty minutes can be retained; when it has not yet been estimated, there is no need to cut endings first.

C3 also does not directly derive "cannot make the news public" from "no law enforcement authority." The author needs to separately write this work's dissemination rules: for example, official announcements are signed by the director, and the records clerk can hand materials to the director or a reporter. At this point, what the player decides is to whom to hand them and whether to hand them over, not suddenly gaining the authority to sign announcements. Once the rules are written into planning, later dialogue has a basis.

For the hard conflict C1, give two contraction plans; C2 continues to be estimated, and C3 is judged after the rules are completed, not lumped together with C1.

Plan one: keep the single scene, cut locations. Change "each chapter spans five cities" to "the whole piece is concentrated in the port affairs hall," and the five cities become investigation subjects through five letters rather than being shown city by city. In this way, the spatial scale does not change and the form does not change; the cost is that it cannot rely on travel for scene changes and needs to advance through records and dialogue.

Plan two: keep five cities, change the scale. Change "single-scene short piece" to "multi-scene work," and re-estimate the duration of one complete route. The locations are retained, but the amount of material and transition work increases; both the twenty minutes and the number of endings must be re-evaluated, and the original volume cannot be promised before estimation.

Both plans must simultaneously check downstream: changing the plan affects script, storyboard, style, characters, nodes, video, preview, and export. The system marks related completed steps as "needs update" according to dependencies, and old data is retained; the creator then checks whether the expression of old materials still holds, and cannot treat the status marker as meaning the specific conflict has already been resolved.

After the changes, compare again with the same table to confirm there are no remaining conflicts:

1. Whether all five types of requirements have been written as judgeable short sentences.
2. Whether each pair of conflicts changed only one of them, rather than vaguely retaining both.
3. Whether the changed requirements have been synchronized to downstream descriptions such as storyboard, characters, and nodes.
4. Whether the ending promise is consistent with scale, form, and player identity.
5. Whether the conflict list leaves a one-sentence reason for "why it was changed this way."

If item 2 cannot be done, it means mutually exclusive requirements are still being retained at the same time, and the script stage will crash again.

This article was organized by DramaFork and explains according to the current project implementation: planning is an auditable and editable stage, and AI can assist in advancing it, but it is not claimed that it can automatically identify all conflicts. Prompt constraints do not guarantee correct output every time, so the conflict list and contraction decisions still require the creator's own judgment.

Next action: write each of the five types of requirements in your current plan as one judgeable sentence, first compare only the "scale × form" group, and mark the one that cannot hold at the same time.