Revising

On a recent flight, I rewatched one of my favorite films: Michael Mann’s Heat. It is firmly in my top five and a reminder that the perfect runtime might be three hours. The 1995 film is now regarded as one of the great crime sagas. It was also the first time Al Pacino and Robert De Niro appeared together on screen. Both were in The Godfather Part II, but their characters lived in different timelines and never shared a scene.

What many people don’t know is that Heat was not Mann’s first attempt at the story. Six years earlier, NBC aired his television film L.A. Takedown, in which detective Vincent Hanna chases Patrick McLaren and his crew. The names and faces were different, but much of Heat was already there: the armored-car robbery, the obsessive detective, the disciplined thief, and even the coffee shop meeting between them.

The story goes back further still. Mann wrote the first draft in 1979, drawing on the experiences of his friend Chuck Adamson, a Chicago detective who pursued a real thief named Neil McCauley in the early 1960s. Adamson once sat down for coffee with McCauley. Months later, he killed McCauley after a robbery. Their conversation became the emotional center of Mann’s script: two men who understood each other, even as they accepted that one might eventually kill the other.

In many ways, L.A. Takedown was the prototype for Heat. Mann cut a 180-page screenplay down to roughly 110 pages for television and shot it in only nineteen days. After the success of The Last of the Mohicans, he had the leverage to return to his Los Angeles crime saga with more time, more money, and two of the greatest actors of their generation in the lead roles.

The easy explanation is that Heat is the same movie made with a bigger budget. But money alone does not explain why it works so much better. By then, Mann knew which parts of L.A. Takedown held up and which had been lost when he shortened the script. The pursuit worked. The parallel lives worked. What was missing was the effect these men had on the people around them: Hanna’s collapsing marriage, Chris and Charlene’s volatile relationship, and Neil’s attempt to build a life with Eady.

With Heat, Mann had 170 minutes instead of 95. He used that time to bring back the people he had cut from the earlier version. The diner scene was already there. The difference was that we now knew more about the two men sitting at the table.

I think Mann’s decision to remake L.A. Takedown as Heat offers three lessons for founders and builders.

Lesson 1: Ship the compressed vision

L.A. Takedown was worth making. It was not the full version Mann had in mind, but it got the idea out of a drawer and into production. On the page, Mann could believe a scene worked. Once it was filmed, that scene had to work with real actors, locations, pacing, editing, and an audience.

That is what a first version gives you: something real to react to. In The art of prototyping, I wrote that a prototype is not a miniature version of the final product. It is a tool to answer a question. L.A. Takedown answered whether the chase between the cop and the thief worked. It could not fully test the relationships around them because many of those scenes had been cut to fit television.

Founders often reduce this lesson to “ship fast.” I agree with shipping fast, but after you ship, be honest about what you actually tested. Did the prototype test demand, the experience, or the business model? Or did it only prove that your team could build it? If you cannot answer that, you might mistake an incomplete test for a failed idea. You might also mistake a good demo for a working company.

You can go beyond v0, but you have to start with v0. Be clear about the questions that version answered and what it left unresolved.

Lesson 2: Hold onto the thesis; revise the specification

The central idea of Heat barely changed. Hanna and McCauley are mirror images, consumed by their work and isolated by the codes they live by. When the first version fell short, Mann kept that idea and reconsidered how to express it.

Founders often confuse the vision with the first version of the product. The vision is your belief about what should exist. The product is one attempt to make it real. Staying true to the vision does not mean rebuilding the same thing over and over. Sometimes we are simply reluctant to admit that our first approach was wrong.

Steve Jobs faced a similar decision at NeXT. Its computer was beautiful, expensive, and commercially unsuccessful. In 1993, NeXT stopped making hardware and became a software company. The company gave up the box but kept the operating system, which later became the foundation for Mac OS X after Apple acquired NeXT. The part worth saving was the software inside the black magnesium cube.

This distinction matters in AI, where every new model produces a wave of products chasing the same trend. It is easy to copy what other companies are building. Ask yourself whether you still believe the original problem is worth solving, even if your first solution was wrong.

Lesson 3: Name what failed and succeeded

The difference between iteration and repetition is knowing what happened. For me, the retrospective is the most important ritual in an agile sprint. Saying “it didn’t work” is not useful on its own. You need to know why. Did the product fail to reach enough people? Did customers ignore a feature you expected them to love? Did the core experience work only when a founder personally onboarded every customer?

Mann knew what survived L.A. Takedown: the structure, the conflict between Hanna and McCauley, and the coffee shop scene. He could also see what the shorter version had flattened: the relationships that gave the pursuit its weight. That told him what to keep and what needed more work.

Without that diagnosis, a second attempt can easily become the first one with more features, more money, or a fresh coat of paint. Keep what worked. Give the weaker parts another pass.

The lesson is not to wait around for the perfect moment. Mann did not spend sixteen years doing nothing. When he could not make the full film, he made a television version instead. That first production showed him what the story still needed. When he finally had the opportunity to return to it, he knew where to spend the extra time and money.

That is the difference between a second attempt and a second draft. The diner scene still works thirty years later because Mann did more than preserve the dialogue. He revised everything around it until we understood exactly who was sitting at the table.

For founders and builders, the original vision is the script. Refer back to it, but do not treat it as sacred. Revise the script as you build and ship.

Hyperlinks + notes

• L.A. Takedown is the rough draft for Michael Mann’s masterpiece Heat
• Double Trouble: How Michael Mann Used L.A. Takedown to Make Heat
• Michael Mann on the long odysseys of Heat and Ferrari
• What Michael Mann Changed—and Didn’t—for the anniversary edition of Heat
• The 1993 report that NeXT would stop producing computers