Death Machine: The Corporate Killer-Robot Cult Item

The characters in Death Machine (1994) are called Jack Dante, Hayden Cale, Scott Ridley, Sam Raimi, John Carpenter, Yutani and Weyland. This is not a subtle film. It is a debut by a British effects technician who had spent his twenties building other people’s monsters, and it announces its influences in the cast list the way a first album announces them in the liner notes. The remarkable thing is that underneath the graffiti there is a real picture, and inside the real picture there is a Brad Dourif performance that has no business being this good.

Dourif, unsupervised

Jack Dante is a weapons designer at Chaank Armaments, a corporation with a glass-and-steel headquarters and a research division that has been left alone for too long. Dourif plays him as a man who has been given a workshop and no adult supervision since roughly the age of nine: a genius, a hobbyist, and a collector of grievances, who has fallen in love with the thing he has built in the way a lonely child falls in love with a dog.

What makes it work is that Dourif refuses to play the part as sinister. Dante is delighted. He is thrilled to show you his machine, thrilled to explain it, wounded when you fail to be impressed, and it is the woundedness that curdles into the plot. The film’s new chief executive, Hayden Cale (Ely Pouget), makes the mistake of treating him as an employee with a performance problem rather than as an unexploded device with a grudge, and Death Machine is essentially a workplace-conflict film in which one party has a hunter-killer robot in the basement.

Dourif was, by 1994, one of the most reliable people in genre cinema and one of the least well used — a character actor with a lead actor’s engine, generally hired for four scenes of weirdness. Norrington hands him the film. It is the most screen time and the widest emotional range he had been given in years, and he spends it building a person: fidgety, funny, wretched, and absolutely certain that he is the wronged party.

The names are the thesis

The in-joke naming looks like sniggering and functions as a statement of lineage. Weyland and Yutani are lifted wholesale from the corporation in Aliens. Scott Ridley and Sam Raimi and John Carpenter are the directors whose films this one is digesting in public. Norrington is telling you, before a frame of monster appears, that he knows exactly which shelf he belongs on and intends to earn his place there rather than pretend he arrived from nowhere.

That honesty buys him a lot of goodwill, and he mostly repays it. The corporate satire is thin, and it knows it is thin. The film’s actual project is to get you into a dark building with a practical creature and a small group of people who are about to be reduced, and everything before that is scaffolding erected at speed.

Why the Warbeast lands

The Hardman project — the thing in the basement — is a practical build, and 1994 was very nearly the last moment when a low-budget picture could get away with that. It moves like a puppet because it is one, and Norrington shoots it the way you shoot a puppet if you understand puppets: in fragments, in low light, from below, with the camera moving so that the eye never gets a clean architectural read on the whole object. You assemble the machine in your head out of a claw, a jaw, a shadow and a sound, and the version you assemble is better than any version a budget could have delivered.

The sound is doing more than the models. The Warbeast announces itself through concrete — a hydraulic exhalation, a scrape, the particular clatter of something heavy with too many joints — and the film gets its best sustained tension out of long stretches in which the machine is audible and invisible. This is the oldest trick in creature cinema and it is old because it works, and a debut director who had spent years inside the workshops knew precisely how little of his own kit to show.

The building helps. An office block after hours is a genuinely good horror set: strip lights, identical corridors, fire doors, a car park, a lobby the size of a cathedral. Norrington uses the corporate architecture as a maze and lets the emptiness do the atmospheric work, which is a solution born of poverty and is better than the solution money would have bought.

Made at the exact wrong moment

Death Machine arrived a year after Jurassic Park, and that timing explains its commercial fate more completely than any judgement of its quality. In 1993 a studio picture demonstrated that a computer could render a large animal walking around in daylight and hold a wide shot on it. Every practical creature film made in the following few years was, fairly or not, read as the old way of doing things — and the audience that had spent a decade admiring a well-articulated puppet suddenly had a benchmark that no puppet could meet.

Watch it now and the calculus inverts. The Warbeast has weight, because it has mass; it displaces air; the actors are reacting to an object that is physically in the room with them and casting a real shadow on a real wall. A great deal of mid-nineties digital work from films with fifty times this budget has aged into unwatchability, and Norrington’s junk-shop hardware has not aged at all, because there is nothing in it to date. It was built, lit and photographed, and those three operations do not go out of fashion.

There is a decent argument that the whole British low-budget genre tradition — this film, Hardware, the creature shops that fed both — was an ecosystem killed by a technology rather than by an audience. The skills went into other industries. The films stopped being made. What survives is a short shelf of pictures where you can see somebody’s hands.

What happened next

Death Machine did the thing a good calling-card film is supposed to do. Norrington went from this to Blade (1998), a studio picture that reorganised the commercial prospects of comic-book adaptation several years before anyone was paying attention, and then to The League of Extraordinary Gentlemen (2003), after which he largely stopped. Two decades on, the debut is the one people go back to, because it is the one where you can see a specific sensibility rather than a specific budget: the fascination with hardware as flesh, the affection for the technician over the executive, the refusal to make the monster’s maker contemptible.

The case against

The film is fifteen minutes too long and the corporate-conspiracy strand is filler. A trio of activists break into the building to expose Chaank’s dealings, and they are present because the third act needs bodies; their scenes are a different, worse film, and every cut to them costs momentum that Dourif has just generated. Pouget’s Cale is written as a competence rather than a character, and the screenplay keeps giving her exposition when it should be giving her a position. The dialogue outside Dante’s workshop is serviceable at best and occasionally reaches for a quip it cannot land.

There is also a structural cowardice. The film is genuinely interested in the sympathy between a maker and his machine, and then converts that interest into a chase in the last half hour, because a chase is what the shelf demanded.

The real ancestor

The obvious answer is Aliens, and the names make it hard to argue. The truer answer is Hardware (1990), Richard Stanley’s shoestring British nightmare about a military drone reassembling itself in a flat. The two films share a nationality, a poverty, an aesthetic — rust, sparks, industrial noise, a machine built from junk — and a conviction that the interesting question about an autonomous weapon is who was lonely enough to make it. Hardware is the more feverish picture. This one has the better performance.

Sideways, Runaway (1984) had already worked the same seam from the respectable end: Michael Crichton’s version of the killer-machine film, with a police procedural bolted on and a designer villain who is the most interesting thing in it. And Screamers (1995) arrived a year after this one to ask the follow-up question that Death Machine declines: what happens when the machine starts building the next one itself.

It turns up in reasonable transfers and runs a shade under two hours. Go for Dourif, stay for the sound design, forgive the activists.

Spoilers below

The best decision in the screenplay is that Dante never becomes a villain in the conventional sense. He is fired, and the firing is the inciting incident, and everything that follows is a tantrum with industrial consequences. Dourif plays the escalation as heartbreak — the machine is the only relationship he has, and he sets it loose in the building because the building rejected him. The film understands that a man who builds a killing device in a basement is not executing a plan. He is furnishing a mood.

The Warbeast’s hunting logic — it tracks fear, adrenaline, the chemistry of panic — is the film’s cleverest single mechanic, and it produces its one great sequence: characters trying to stay calm in the presence of something that detects calm’s absence. Every horror film about hiding is really about breathing, and this one makes the subtext into a specification.

The finale gives Cale the fight, which is the correct choice and the least surprising one. The activists are spent as fodder along the way, exactly as the structure promised. Dante’s end is where Norrington’s affection shows: he is killed by his own device, and the film stages it as a reunion rather than a comeuppance, a man at last getting the undivided attention of the only thing he ever loved. It is a genuinely strange note to end an action film on, and it is why this one is still worth arguing about.