Paul Thomas Anderson does not write screenplays. He composes them. And One Battle After Another โ loosely inspired by Thomas Pynchon's Vineland โ is his most compositionally precise work since There Will Be Blood.
The numbers: 94% Rotten Tomatoes. 7.7 IMDb. $200 million worldwide on a $130-175M budget. 13 Oscar nominations. Golden Globe for Best Screenplay. Critics' Choice for Best Picture.
When we model this film's likely structural features against our correlation engine, it lights up like nothing else in 2025.
The Three Features That Define PTA
Question Density: The Pynchon Inheritance. Anderson's screenplay, drawing from Pynchon's paranoid fiction, is structurally interrogative. Every scene asks questions โ of the characters, of the audience, of the narrative itself. Where is Willa? What does Colonel Lockjaw want? Why did Bob Ferguson abandon the revolution? Is this a comedy or a tragedy? Our data shows question density at r = 0.122 with RT score and r = 0.108 with audience score. PTA's scripts are among the most question-dense in cinema. This one may be his most interrogative yet.
Sentiment Mean: Precisely Calibrated. One Battle After Another is nominally a comedy โ a dark one, punctuated by sudden violence and long stoner conversations. But the overall sentiment mean likely registers as moderately positive. PTA's signature is finding warmth inside chaos: Bob Ferguson loves his daughter despite being a disaster of a human being. The car chases are absurd rather than terrifying. Even the violence has a slapstick quality. Sentiment mean is our strongest predictor of critical score (r = 0.125 with critical, r = 0.144 with RT). PTA's comedic undertone buoys the entire metric.
Avg Dialogue Length: The Long Speech. PTA writes long dialogue blocks. His characters monologue, digress, circle back, lose their train of thought, and arrive somewhere unexpected. Average dialogue length has a modest positive correlation with IMDb rating (r = 0.020) โ but at the extremes PTA operates in, the effect amplifies. Long speeches signal trust in actors, trust in audiences, and trust in the material. All three are present here.
The Nonlinear Time Jump
The screenplay features a 16-year time jump from a 2009 prologue. This structural choice affects several of our features in interesting ways:
- Scene length variance: Elevated (the time jump creates two distinct rhythmic halves)
- Character intro rate: Elevated (characters appear at different ages, essentially as different people)
- Sentiment arc slope: Complex (the arc resets at the time jump, creating a double-helix structure rather than a single line)
Nonlinear scripts tend to score higher on critical metrics in our dataset, though we do not have a dedicated "linearity" feature. The proxy features โ scene length variance, character intro rate โ capture the effect indirectly.
The Leonardo DiCaprio Variable
DiCaprio's comedic turn as Bob Ferguson โ a stoned ex-revolutionary with a panic disorder โ represents a tone our engine would detect through sentiment variance: wild oscillations between anxiety, humor, tenderness, and rage within single scenes. High sentiment variance correlates positively with box office (r = 0.052). When DiCaprio commits to emotional range, audiences show up.
Thirteen Oscar nominations confirm what the structural features predicted: this screenplay operates at the highest tier across the widest range of our metrics. It is the closest thing to a structurally perfect film in 2025 โ the only competition being Sinners, which optimizes a different set of features.
Predicted tier: S-Tier. Actual: S-Tier. PTA did not need our engine. But the engine confirms his instinct.