Crash (2005) Review
4/10
A car accident at night. A district attorney’s wife, two carjackers, a locksmith, two police officers, a television director, a shop owner and a detective, whose lives will intersect across thirty-six hours in a city where nobody touches anybody.
The 4 is for a film with real craft in individual scenes and a governing structure that converts a subject into a mechanism. Two or three sequences in it are very good and the arrangement they sit inside undoes them.
It also won Best Picture in the year of Brokeback Mountain, Good Night and Good Luck, Munich and Capote, and it’s the least defended winner of the last forty years.
I find Crash exasperating, and the exasperation grows the longer it runs. The moments that work make it worse, because I can see a better film buried under the lesson plan. By the hour mark I’ve stopped watching people and started watching Haggis move pieces into position so each one can be flipped.
The Structure Is a Machine
The film introduces every character holding a position about race and places each in a situation that tests it, and each one finishes somewhere different.
The mechanism is uniform and that’s the problem. The officer who abuses a Black couple saves a Black man from a burning car. A Black housekeeper catches the woman who’s frightened of a Latino locksmith. The shop owner who threatens a locksmith’s daughter shoots her and she lives. Each pairing is an inversion and each inversion arrives on schedule.
The locksmith is the one who breaks the pattern by not being in it, and the film seems unaware of that. Every other character is assigned a lesson and he’s simply doing his job. That makes him the only person in the picture.
That’s a designed object instead of a story. A structure that reverses every position is administering a set of lessons, and the audience learns the pattern within about forty minutes. After that I stop caring about the film entirely, because I resent it handing me a worksheet and telling me it’s a portrait of a city. Once the pattern is legible the film has nothing left. Every later introduction is read as a setup, the audience predicts the inversion, and the remaining ninety minutes are confirmation. Short Cuts and Magnolia are the comparison and both are longer. What those films do that this one doesn’t is let several strands resolve into nothing. That’s what keeps the reader from learning the machine.
A pattern the reader detects becomes a machine. Once they know that every position will be reversed, no individual reversal can affect them. Vary the outcomes. Some positions should hold, some should worsen, and one or two should reverse, and the reader can’t predict which.
A uniform structure is the most common failure in multi-strand work. The strands have to differ in what happens to them or the form is doing nothing.
Ask what your reader knows after forty minutes. If they know the shape of every remaining scene, you’ve finished the book early.
Two Sequences That Work
The invisible cloak scene is good and is the one piece of the film everybody remembers for the right reason.
A locksmith gives his young daughter an imaginary cloak that stops bullets. Later she runs into gunfire believing it. The film stages the scene without cutting away and the outcome is withheld for several seconds that are very hard to sit through. The staging also puts the audience behind the wrong information. The film has established what’s in the shop owner’s gun, the audience hasn’t been told what he bought, and the withholding is the only honest suspense in the picture.
It works because the setup wasn’t a setup. The cloak scene exists early as a piece of parenting and the film doesn’t signal that it’ll return. Those few seconds are the only stretch of Crash I’d defend without reservation, and they’re a big part of why the rating isn’t lower.
The other sequence is Matt Dillon and Thandiwe Newton in the car fire, and it works for one reason. Neither of them says anything about what happened between them earlier, the film declines the speech it’s set up, and the scene is two people managing a practical emergency.
Newton has said the traffic stop was extremely difficult to shoot and that Dillon was careful with her throughout. That’s the kind of production detail that matters when a scene asks that much of a performer.
The Performances Are Better Than the Film
Matt Dillon was nominated and the performance is the most difficult thing in the picture. Officer Ryan is a racist, is caring for a father in pain, is competent at his job, and Dillon plays all three without ever suggesting that the third excuses the first. The hospital scene where he pleads for his father is the character’s whole construction in four minutes.
The HMO call is the sharpest version of it. He’s on the phone with an administrator who won’t help, is racist to her about it, and the film puts an entirely legitimate grievance and an entirely illegitimate response in the same ninety seconds.
Thandiwe Newton has the film’s worst experience and the best two scenes in it. The traffic stop is almost unwatchable and is played without any of the dignity that material usually receives, and the argument in the kitchen afterward is the only place in the film where somebody is angry about the right thing.
Terrence Howard is the other half of that scene and his is the more difficult performance.
He’s compliant, he’s humiliated, and the argument in the kitchen afterward is about what compliance cost. Then the film drops it. That choice is the worst in Crash, because it’s the one conversation the movie had no business walking away from.
Michael Peña’s locksmith is the film’s one character who holds a consistent position throughout and isn’t corrected. He’s a competent man doing his job around people who assume things about him, and the film lets him simply be right.
Don Cheadle has the film’s framing narration and the least rewarding position. Graham is a detective with a missing brother and an institutional compromise to make, and the character is asked to deliver the film’s thesis in the opening voiceover about people needing to crash into each other.
One character who’s not corrected shows what the others cost. Peña holds a position, is right, and the film doesn’t require him to learn anything. If every character in your work changes, none of the changes mean anything.
Withhold the speech you’ve set up. The car fire scene works because neither of them refers to the earlier encounter, and a lesser version has one of them say something about it.
Let a character be repellent and sympathetic without the sympathy excusing the repellence. Dillon plays both and the film never asks the audience to trade one against the other.
Why It Won
The explanation is documented and has several parts, none of which is mysterious.
It was a Los Angeles film about Los Angeles with a large cast of working local actors, screened aggressively on DVD to a membership based in the city, in a year when several voters said publicly that they hadn’t watched Brokeback Mountain. A film about a voting body’s own city, distributed to their homes, against a film a portion of them declined to see.
It also opened in May 2005 and campaigned for ten months. That was unusual then. A spring release that stays in the conversation until February has been kept there deliberately and expensively. The other element is the subject’s flattery. Crash tells an audience that prejudice is everywhere, that everybody has it, and that everybody can be surprised out of it. That’s a proposition nobody in a room has to resist. Brokeback Mountain tells an audience that two men lost twenty years and nobody gets them back.
That distinction explains the win better than anything else. A film that tells an audience everybody is a little bit guilty asks nothing of them, and a film that tells them a specific harm was done and not repaired asks a great deal. The voters picked the comfortable one, and I find that more depressing than anything in the film itself.
Haggis himself said several years later that the film shouldn’t have won. That’s an unusual thing for a winner to say and is the most honest sentence anybody involved has produced about it. He’s also said he thinks it’s a good film and that Brokeback Mountain was better. That’s a more precise position than the one usually attributed to him and is worth stating accurately.
The Screenplay Award
This page lists Crash twice, once for Best Picture and once for Best Original Screenplay, and the second entry needs stating carefully. Haggis won Original Screenplay for this in 2006. He’d won it the previous year as well, for Million Dollar Baby, making him the only person to write consecutive Best Picture winners.
Eternal Sunshine of the Spotless Mind won Original Screenplay in 2005, the year between. That means the page’s framing about the timing is slightly off and the underlying point stands. The screenplay award is the more objectionable of the two, because the structure is the thing being rewarded and the structure is the film’s central fault.
That’s the sharpest version of the complaint. Best Picture is a judgment about a film and Original Screenplay is a judgment about exactly the mechanism this review has spent five sections describing.
A screenplay award for a film whose structure is its weakness is the clearest possible statement about what a voting body values. Craft in execution is visible and structural soundness isn’t. Readers detect structure without naming it and juries reward what they can point at.
The Best Picture presentation is also part of the record. Jack Nicholson read the envelope, visibly reacted, and said the title in a tone several people in the room commented on afterward. Writing consecutive Best Picture winners is a fact about access and timing instead of about quality, and Haggis has effectively said so.
When a creator disowns a result, the record should carry that.
It’s the most useful thing anybody involved can contribute. The film also cost around six and a half million dollars and took about ninety-eight worldwide. That’s the sort of ratio that generates a campaign narrative on its own. A small film outperforming its budget by fifteen times arrives at a ceremony with momentum.
Should you watch Crash?
Watch the two scenes and skip the rest. The invisible cloak pays off without having been telegraphed, the car fire refuses the speech the film had set up, and Matt Dillon does something difficult. Around them sits a structure that reverses every position on schedule until the audience is forty minutes ahead of it. That’s why it’s a 4. The craft is real, it’s maddening to watch a machine grind it down, and a Best Picture trophy on top only makes it worse.
