How Fault Was Made
By Rob Chipman
I gave the brief. Ian made the game.
That's the honest version of what happened today.
The brief was simple: build a Wordle-style daily puzzle that lives on this site. NYT quality bar. A marketing hook disguised as a game. Wordle and Murdle both started as website apps, and both went viral before anyone treated them like products. I wanted to try.
What I didn't expect was the process that followed.
Constraints before concepts
Before anyone pitched a game concept, I asked for constraints.
Cleo, the business analyst persona I work with, gave the market ones: daily format, 2-5 minute sessions, shareable result that doesn't spoil the answer, no account required, mobile web first. Proven by Wordle, Connections, and Murdle. Non-negotiable.
Poppy, the developer, gave the technical ones: no game engine, client-side only, deployable to Vercel in hours, no audio dependencies. Build it from DOM and CSS or don't build it.
I added one: it has to feel like it belongs in the NYT Games lineup.
That last constraint is the one that did the most work.
The first concept was wrong
Ian, the creative director, pitched "Cut" first: a paragraph with one wrong sentence. Not factually wrong. Tonally wrong. Find the sentence that doesn't belong.
I pushed back. If you read a lot, you'd find it in seconds. If you don't, you might never find it. That's not a game. That's a taste test.
The constraint I gave back: all players start at the same spot. The only thing that makes you better over time is tactics learned from playing. No prior knowledge advantage.
That one constraint killed Cut and found Fault.
How the mechanic survived
Fault is a logical contradiction, not a tonal one. Anyone can spot a contradiction. A sentence that says a character is alone and then has her hand something to someone, that's checkable by anyone, reader or not. The tactic you learn over time is how to read carefully and systematically. Not what you already know walking in.
From there, Ian pushed for three beats of one cohesive story instead of separate vignettes. If the three sections are connected, the fault can live anywhere across them: a detail planted in the first beat contradicted by a sentence in the third. That's only possible when it's one story.
I added the Monty Hall step: new information arrives after you pick your beat. Hold or switch? Ian saw immediately what that does. It creates a moment of doubt even when you're right. The best moment in Fault isn't finding the fault. It's the second when you're not sure anymore.
The sequencing step came last. Drag the three beats into narrative order before you start looking. Get the order wrong and the fault is harder to find, because you're reading the story in the wrong arrangement. Ian's idea. It makes the whole game a nesting doll: drag to arrange, click a beat to freeze the sequence and reveal sentences, click a sentence to advance. Each commitment unlocks the next layer.
Built in an afternoon
The game went from concept to running prototype in one session. Three rounds of iteration followed: the reveal wasn't clear enough, the sentence tap targets were too close together, the confirm buttons were friction I didn't need. I played it, gave feedback, the team fixed it.
Scoring landed at: 2 points for the correct sequence, 2 points for the correct beat, 4 points for the correct sentence, 2 points for the right hold-or-switch decision. Max 10. The sentence is worth the most because finding the exact line where the story breaks is the whole game. Everything else is the mystery in the way of getting there.
What I'm still thinking about
Ian made the creative calls. Not all of them. I killed Cut, I added Monty Hall, I said the sentence should be the money ball. But the nesting doll structure, the three-beat cohesive story, the decision to use logical contradiction instead of tonal judgment: those came from the process, not from me dictating them.
That's not how I expected today to go.
It's not "AI generated content." The game didn't write itself. Every decision went through a feedback loop: pitch, constraint, push back, refine. I was the editor. But I wasn't the creative director.
I don't know exactly what to call that. I just know the game is live, it's hard, and I'm going to play it tomorrow to find out if I do any better.
If you want the longer story of what TFL is becoming, what TFL stands for covers more ground.