QEDHow things connect
Relationships in fiction are discovered, not declared, which is why NVS never asks you to label a bond up front.
How things connect
Tip
A cast list is a phone book. The story is what runs between the names.
Line up your characters and you have a roster. The story is in the between: she loves him, he serves the king, the king fears the prophecy, these two were both there the night it happened.
Readers do not fall for a character so much as for a relationship, and most of the craft is arranging those, pulling them tight, and paying them off.
You can watch any pair scene by scene in the Relationships rail.
The part software usually gets wrong
This is where writing parts ways with every business-software idea of data.
| In business software | In a story |
|---|---|
| A relationship is a fact | A relationship is a discovery |
| True from the moment it is entered | Becomes true partway through, sometimes by accident |
| One correct value | Allies in chapter two, enemies in chapter nine, both true |
You often do not know in chapter two that these two become lovers, or that the kindly mentor is the traitor. You find out by writing.
This means unlike a B2B SaaS software, the ontologies built for traditional software doesn't work for screenplay writing!
So NVS never asks you to label a bond up front.
There is no dropdown where you declare "ally" or "enemy". It records what each scene actually showed, keeps the line that showed it and when, and lets the nature of the bond emerge and shift.
Fuzzy is allowed, and often truer
You are free to leave a relationship undefined. "These two have a complicated, shifting thing" is often better than forcing one word onto it, and the app is content to hold that.
Blank is a real answer. It means undecided, and nothing in NVS requires you to know more than the story knows yet.
If you later want a clean read, you can ask for one and it will cluster what it has seen into a best guess. It will never tidy your story behind your back.
The declared and the observed stay apart
There are two kinds of truth here, and NVS deliberately never merges them.
What you declared lives on the character's page: this is who she is.
What the engine observed lives in the analysis: this is who she acted like.
Keeping them separate is the only reason coherence findings can tell you the two have diverged.