Are Characters Fabricated or Excavated
Forrest Gump Isn't Real and Everybody Cries Anyway
Eric Roth says it flatly about the characters he's most famous for: I don't think they're real, the people.1 And yet audiences weep at Forrest Gump, return to him for thirty years, and show him to their eleven-year-olds. He's not real. He's also not a cartoon. Roth's phrase for the working zone is the other side of the moon — these characters live their own lives over there, they feel real to that extent, but it's all fiction.2 Fabrication, done well, on purpose.
David Whyte works the opposite way. His writing comes up through what he calls the bodily membrane — the material extruded from lived experience at the edge between what you know and what you don't yet.3 Characters, on that account, aren't built. They're found, in you, by attending long enough that they surface.
Two working writers, two incompatible accounts of where a person on the page comes from. The vault files this as an open collision rather than resolving it.4
Why "Both Are Valid" Isn't the Answer
The comfortable resolution — some writers invent, some excavate, both fine — misses that these aren't temperaments. They're claims about what makes a character land.
Roth's claim: characters land because of internal consistency. The exaggerations are surface; the interior has to be honest. Forrest never existed, but the way he loves Jenny is psychologically true, and the way he relates to his mother is psychologically true.5 Consistency is the load-bearing element and it has nothing to do with the character resembling the author.
Whyte's claim: characters land because of existential continuity with the writer's own life. They can't be assembled from outside because the thing that makes them real is that they came through a person.
Both writers are explaining the same phenomenon — why some characters bond to readers and some don't — and they locate the cause in different places. That's a real disagreement about mechanism, not a difference in style.
What Roth Says Actually Prevents Collapse
Roth is more specific than "be consistent," and the specifics are the useful part. Three markers separate fanciful from cartoon:6
Psychological truth under biographical falsehood. The surface can be exaggerated as far as you like. The interior can't be.
Voice that holds everywhere. Anyway, like I was saying, shrimp is the fruit of the sea. Bubba sounds like Bubba in every line. When a fanciful character's voice shifts to serve a scene's comic logic, they collapse into cartoon.
One thing the writer takes completely seriously. Every fanciful character has an element the writer refuses to play with — Forrest's simple wisdom, Benjamin Button's tugboat. And here's the mechanism worth extracting: the seriousness of the one element earns the writer the right to play everywhere else.
That inverts the standard advice. Told your characters feel cartoonish, the instinct is to add realism across the board. Roth's answer is the opposite — realism diffused across all elements produces wooden characters; seriousness concentrated in one element licenses stylization everywhere else. Forrest's love for Jenny is taken seriously. The rest can be slapstick.7
The Diagnostics Both Modes Share
The failure signs Roth lists don't actually depend on which mode you're in, which is the first hint the disagreement is narrower than it looks.8
Amusement-at rather than interest-in. Amusement means the audience is outside looking in; interest means inside looking out. Mixing the two within one character is fatal — a character the reader settled into being inside, suddenly played for laughs, loses them permanently.
The character has no contradictions. Cartoons are consistent in the wrong way: every behavior derives from one trait. A villain who only does villain-things is a cartoon; a villain genuinely tender to one specific person, in a way that isn't a redemption arc, is alive.
The character speaks like the writer. Roth heard Forrest's voice before he knew the plot, and the voice wasn't his. When you can't tell which line is the character's and which is yours, the character has collapsed into you — which is a form of cartoon even though it feels like authenticity.
That third one is the interesting one, because it's a failure available to the excavation mode specifically. If your method is extruding characters from your own lived material, the collapse-into-the-writer failure is the one waiting for you. And the fabrication mode has its own mirror: the vault flags that fanciful can become an excuse for not doing the psychological work at all — oh, they're not meant to be real disguising a character who was never fully imagined.9
You can swap two characters' dialogue without breaking the scene. Trade lines between two characters and if nobody notices, neither is alive. Both are puppets in the same hand.
What Actually Decides It
Set the two modes against each other and the vault's own reading is that the mode isn't decisive. Documentary fiction lands. Memoir-as-fiction lands. What's decisive is the commitment — internal consistency in Roth's case, lived truth in Whyte's — and both are ways of refusing to leave the character underspecified.10
There's supporting evidence from outside craft entirely. Parasocial bonding — the one-way attachment audiences form to figures they'll never meet — requires a character consistent enough to model. Cartoons can't be bonded to because their behavior is dictated by external requirements (plot needs, joke beats) rather than internal logic.11 That's the actual threshold, and it's indifferent to whether the character was invented or excavated. A fanciful character with airtight interior logic outperforms a documentary character with shaky interior logic on every measure of engagement.12
So the operative question isn't did I make this person up or find them in myself. It's is this person's interior specified enough that a reader can run them forward — predict what they'd do in a room you haven't written yet.
The Question Neither Writer Asks
The vault surfaces one thing neither Roth nor Whyte addresses, and it's worth leaving open rather than answering: is the fictional character inherently more or less likely to land as real than the autobiographical one?13 The vault says evidence accumulates in both directions, which is an honest non-answer.
There's a sharper version underneath it. Roth's mode requires the writer to not be the character — the whole point of the other side of the moon is that they live over there, independently. Whyte's requires the writer to be continuous with the material. Which means the two modes have opposite relationships to the one failure Roth names as hardest to see: the character speaks like the writer. In fabrication that's a collapse. In excavation it might be the point. Nobody in this vault has worked out whether that's a real distinction or whether excavated characters still have to achieve separation from their author to become bondable — and if they do, the two modes converge completely at the finish line and only differ in where they start.
Confidence
Roth's fanciful-versus-cartoon threshold, the three markers, and the four failure diagnostics are [VERIFIED] against the vault's page, sourced to a named practitioner interview and [PARAPHRASED] with line references at origin.168 The Roth-versus-Whyte collision is documented in the vault as an open collision stub, not constructed here.4 The parasocial-eligibility and narrative-transportation handshakes are the vault's own cross-domain analysis.11 The closing observation — that the two modes have opposite relationships to the speaks-like-the-writer failure, which may mean they converge at the finish line — is this report's.
Threads: → Are Characters Fabricated or Excavated — Thread Map