Beyond Writing: Madman, Architect, Carpenter, Judge

Back in 1981, Betty Flowers wrote about four roles in the writing process: Madman, Architect, Carpenter, Judge. I first wrote about this framework a couple of years ago because it changed how I think about writing. But the more I've lived with it, the more I realize it has almost nothing to do with writing. It's about how we think through anything complex. Strategy, problem-solving, building teams, building systems. The roles are universal. The sequencing is everything.

If you haven't read the original post, the short version: Flowers identified four distinct modes of thinking that show up when we create something, and she argued they need to happen in order. Most of us blow past that and wonder why things feel stuck.

Here's what I've learned about each role since I first wrote about them, and why I think this framework matters well beyond putting words on a page.

The Madman

The Madman is the generative force. Unfiltered, associative, chaotic. This is where ideas come from, and the only rule is that there are no rules. Mind maps, voice memos, random scribbles on the back of whatever is nearby. You're not building yet. You're excavating.

For me, the Madman shows up on walks, in the car, in conversations where something unexpected connects to something I've been chewing on. I capture it however I can. A quick voice recording, a note scribbled on my reMarkable, sometimes just a sentence fired into a text thread with myself. The medium doesn't matter. What matters is that I don't filter it.

The anti-pattern: The Judge shows up early and kills the Madman. You start evaluating ideas before they've had room to breathe. "That's not practical." "Nobody will care about that." "That's already been said." Every one of those thoughts is the Judge wearing the Madman's clothes, and it shuts down the exact kind of divergent thinking that produces breakthroughs. If you find yourself editing while brainstorming, you've let the wrong role into the room.

How to protect this stage: Give yourself a time window where nothing gets deleted, crossed out, or critiqued. Brene Brown called this the SFD, the Shitty First Draft, and it's a good frame. The Madman needs permission to be bad. That's the only way he gets to be brilliant.

The Architect

After the Madman has emptied the tank, the Architect walks in and starts grouping. Not writing. Grouping. What ideas belong together? What's the actual argument hiding inside the chaos? Where's the thread that ties three random thoughts into something coherent?

I've learned the hard way that I need distance between these stages. A week, sometimes more. The Madman's energy is seductive. You want to ride that wave straight into structure. But if you do, you'll organize around your enthusiasm instead of your insight. Let the ideas cool. When you come back, you'll see patterns you couldn't see in the heat of the moment.

The anti-pattern: The Architect who never lets the Madman finish. This looks like starting to outline halfway through brainstorming. The moment you begin categorizing, you stop generating. You've shifted from "what could this be?" to "where does this go?" And those are fundamentally different questions. The Architect is convergent. The Madman is divergent. They can't coexist in the same moment.

How to protect this stage: Don't outline until you've declared the brainstorm done. Put a hard boundary between generating and organizing. The Architect's job is to find the skeleton. Not to build it yet. Just to see it.

The Carpenter

Now she shows up. The Carpenter is craft. Taking the structure the Architect laid out and making it hold weight. Is this paragraph earning its place? Does this sentence actually say what I think it says? Does the sequence make sense, or am I asking the reader to follow a path I haven't actually built?

This is where I lean on tools. Grammarly catches things my eyes skip over. Reading out loud catches the rest. The Carpenter is methodical, patient, and a little obsessive. She files off the rough edges, mends related thoughts, and asks the unglamorous question: does this work?

The anti-pattern: The Carpenter who starts building before the blueprint is done. If the Architect didn't finish, the Carpenter will spend hours polishing paragraphs that shouldn't exist. This is the most common time trap I see in myself and others. Refining the wrong thing because it feels productive. Editing bad structure doesn't make it good structure. It makes it polished bad structure.

How to protect this stage: Before you start refining, ask: is the structure settled? If the answer is "mostly," go back to the Architect. The Carpenter works best when the blueprint is locked.

The Judge

The Judge comes last for a reason. This is the critical eye. Evaluating the work against its purpose. Who is the audience? Will they get what I'm offering? Did I actually say the thing, or did I talk around it? Is the language right, not too simple, not too clever?

This is also when I bring in other people. A trusted reviewer who knows me well enough to say "this doesn't land" is worth more than any tool. The Judge is about honesty, not perfection. The goal isn't a flawless piece. It's a true one.

The anti-pattern: The Judge who never leaves. Some people live in this mode. They edit endlessly, second-guess every sentence, and never ship. The Judge is supposed to be the final pass, not the permanent resident. If you've been editing the same piece for three weeks, the Judge has overstayed. Hit publish.

How to protect this stage: Set a deadline. Give the Judge a defined window, one pass, maybe two, and then it's done. The Judge's opinion matters, but so does momentum.

Beyond Writing: Why Sequencing Is the Real Insight

Here's where this gets interesting for me. Once I started seeing these four roles clearly, I noticed them everywhere. Not just in writing.

Building a strategy follows the same sequence. You brainstorm possibilities (Madman), organize them into a coherent plan (Architect), refine the details and make them executable (Carpenter), then pressure-test the whole thing (Judge). Skip the Madman and your strategy is safe but uninspired. Let the Judge in too early and you'll never propose anything bold enough to matter.

Team dynamics follow the same pattern. The best brainstorming sessions protect the Madman phase. No critiquing, no "yeah but," just ideas on the board. The best project plans reflect the Architect's hand. The best execution is Carpenter work. And the best retrospectives are the Judge doing what the Judge does.

I've even seen it in building AI systems. The generative phase is wild. What agents could we build, what problems could they solve, what workflows could we reimagine? Then the architecture phase, how do these pieces fit together, what's the right structure, where does that data live? Then the refinement, making each component actually work, polishing the prompts, tightening the handoffs. And finally, the evaluation. Does this system actually deliver what it promised?

The framework isn't about writing. It's about how humans think through complexity. And the insight isn't the four roles. It's that they need to happen in sequence, with space between them, and without letting one role invade another's territory.

The next time you feel stuck on something, a decision, a project, a conversation you can't figure out how to have, ask yourself: which role should I be playing right now? And more importantly, which role is showing up uninvited?