Your Constraint Is Your Best Idea

I had forty cards on my desk last Tuesday. Blank. Forty chances to write something worth reading later. I stared at the pile and made tea. Then I stared at the pile some more. Then I gave up and opened my phone.

If that sounds familiar, the problem wasn't laziness. The problem was that I had no constraint. A blank page is not a starting point — it's an invitation to stall. The Hermes agent team just shipped an open-source creativity skill that calls this out in its first line: 'Constraint plus direction is creativity. No constraint = no traction. No direction = no shape.' That sentence changed how I think about cards.

One Constraint Beats Zero Every Time

The first rule of the Hermes skill is also the rule of every good index card: something must be on it. Not everything — one thing. A question, a fragment, a single observation. The card is the constraint. The card is also the direction, because once the question is written, the answer has somewhere to land. Without the constraint of the card, you have a notebook full of half-thoughts. With it, you have a stack of decisions.

Refuse the First Three

The second rule says: refuse your first three ideas. They're slop. Generate, discard, regenerate. I do this with cards by writing the first three thoughts in pencil, then flipping them face-down before I write the fourth. The face-down cards are still mine, but they don't get to be the answer. Most of the time the fourth card is the one I keep. Sometimes the discarded cards come back better a week later. That's fine — they were never the first move.

One Method, Not Five

The third rule is 'one method per response unless asked.' Don't stack. This is the one that took me longest to learn with cards. I used to try to capture everything on one card: the question, the answer, the next question, the deadline, the feeling. The card would dissolve into mush. Now: one method, one card. If I want to ask, 'what would happen if we shipped this Friday?' that's one card. The Friday answer is another card. The risk is another card. The stack is the conversation.

Specificity Beats Abstraction

The fourth rule: name a mechanism. 'An app for notes' is slop. 'A 200-line tool that lets me paste a screenshot and get a one-sentence summary back' is direction. On a card, the equivalent is: don't write 'think about the new product.' Write 'what's the smallest thing we could ship next Tuesday that would change how Prem uses the app on Wednesday.' The mechanism is the same. The card is sharper because of it.

Weird Must Also Be Good

The fifth rule: frame-breaking is the goal, but weird with no real situation is its own failure. Every set of ideas must include one that is buildable now — non-obvious but grounded, with a real first step. I read this and immediately thought of the stack of cards on my desk where everything was a daydream and nothing had a Tuesday. The rule for cards: at least one card in every stack must answer the question 'what's the smallest next move I could actually do this afternoon.' If none of them does, the stack is a wishlist, not a plan.

Name the Method

The sixth rule says: name the method you used and who invented it. Attribution invokes the discipline. On cards I do this by writing the source in the corner of the card — a tiny citation. 'Prem, 2pm Tuesday.' 'The Hermes skill, rule 5.' 'The conversation with Sam last week.' The corner of the card becomes a tiny provenance record. Six months later, when I find the card in a stack, I know exactly where the thought came from. That's worth the extra three seconds.

When You Pick One, Build It

The seventh rule is the most important for someone with forty cards on a desk: when the user picks one, build it. Don't keep generating. The trap with cards is that the next card is always more interesting than the current one, and the one after that is more interesting than both. But the seventh rule applies. Pick the card that has the smallest next move. Build that. The other thirty-nine cards will still be there on Tuesday — they're not going anywhere, and neither are you.

The Hermes team wrote those rules for an AI agent. They work just as well for a person with a stack of cards and a Tuesday. Constraint plus direction. The card is the constraint. The question on the card is the direction. Everything else is just tea and a phone.