How to Make Fewer, Better Decisions
The method is subtraction, not optimization: inventory what you actually decide in a week, sort every choice by reversibility, replace the recurring ones with policies, delegate everything below a written threshold, and spend the reclaimed attention on the few one-way doors that deserve it. Fewer decisions isn't a lifestyle preference — it's how decision quality is manufactured.
Most founders try to get better at deciding the way a tired lifter tries to get stronger — by adding volume. More inputs, more advisors, more dashboards. The DaD position, worked out in full in Paper II of the trilogy, runs the other way: your decision capacity is finite, most of it is being spent on choices that shouldn't reach you at all, and the highest-return move available is to shrink the queue.
Step 1: Inventory a week of decisions
For one week, keep an unglamorous list: every decision that crosses your desk, one line each. Pricing exceptions, meeting requests, copy approvals, whether to answer the partner's text now or tonight. Don't fix anything yet — just observe.
The list does two things. It shows you the true volume (almost always a multiple of what you'd have guessed), and it exposes the shape of the problem: the vast majority of entries will be small, recurring, and reversible. You are not suffering from hard decisions. You are suffering from undelegated easy ones.
Step 2: Sort everything by reversibility
Run the whole inventory through one question: if this goes wrong, can I undo it cheaply? This is Jeff Bezos's one-way door / two-way door triage. Two-way doors — reversible calls — deserve speed and tolerance for error. One-way doors — the irrevocable ones — deserve real deliberation.
Look at your week's list honestly and count the one-way doors. For most founders in most weeks, it's zero to two. Everything else was reversible, and every hour of deliberation spent on a reversible call was an hour taxed from the calls that weren't. If you notice you've been giving one-way-door treatment to two-way-door choices, you've found your leak. The deeper question of why you do that — what deciding-how-to-decide even means — is the subject of the meta-decision.
Step 3: Replace recurring decisions with policies
Any decision you've made more than three times is a policy waiting to be written. A policy is one decision that retires a whole class:
- "Podcast invites: two per month, my assistant books against written criteria."
- "Refund requests under $500: approved automatically, no escalation."
- "New tools: anyone can adopt anything under $100/month without asking."
Notice what a policy actually is: a distinction, drawn once, in writing. You are deciding where the boundary sits — under $500 versus over — so that no future instance requires fresh judgment. This is the practical face of Paper I's argument that distinctions are what perception and action are made of: a well-drawn boundary does the work of a hundred future deliberations.
Step 4: Delegate with thresholds, not vibes
Delegation fails when it's ambiguous — "use your judgment" — because the delegate can't tell which calls are theirs, so they escalate everything and you're back where you started, now with latency added. Delegation works when the threshold is written: below this line, decide and tell no one; between these lines, decide and log it; above this line, bring it to me with a recommendation.
The recommendation clause matters. It keeps you in the role of approving decisions rather than manufacturing them, which is a different and far cheaper cognitive act. This is also the design principle behind delegating to AI agents, not just people: a system given outcome, boundaries, and a definition of done can carry entire classes of decisions autonomously — which is much of what building with Optimus amounts to in practice.
Step 5: Decide two-way doors at speed — satisficing is the discipline
Herbert Simon's Nobel-recognized observation was that real decision-makers can't optimize; they satisfice — search until an option clears a defined bar, then stop. The failure mode isn't satisficing; it's satisficing with an undefined bar, which degenerates into endless search. So define the bar before you look: "good enough means X and Y; the first option that clears both, we take."
For reversible calls, pair the bar with a clock. Same-day is a good default. The point isn't recklessness — it's that on a two-way door, a fast wrong decision costs one correction, while a slow right one costs the weeks it took, and reality's feedback would have supplied the correction anyway. Velocity is how you buy feedback, and feedback is what judgment is made of.
Step 6: Keep a decision journal and audit quarterly
Once a decision resolves, your memory quietly rewrites the reasoning to match the outcome. The only defense is contemporaneous notes: the call, the reasoning, your confidence, and — the trilogy would insist on this — how your body read the choice at the time. Expansion or contraction. Damasio's research found the body scoring options before the conscious mind could explain its scores; your journal is where you find out whether yours is well calibrated, which is the whole question underneath frameworks versus gut feel.
Quarterly, read it back. You're not grading outcomes — good decisions lose sometimes. You're grading process: Were the reversible calls fast? Did policies hold? Which class of decision kept sneaking back onto your desk?
What this method refuses to do
It refuses to make you decide less about what matters. Every step above exists to protect deliberation for the handful of true one-way doors — the hire you can't unmake, the positioning you can't walk back, the partner you can't unwind. Fewer decisions is the means. Undivided attention at the moments of consequence is the end.
The queue shrinks so that, when a decision actually deserves you, all of you shows up.
FAQ
Doesn't making fewer decisions mean missing opportunities?
The opposite, usually. Opportunities are lost to slow decisions far more often than to absent ones. Cutting the volume of trivial and recurring choices is what creates the attention and speed to actually take the opportunities that matter. You're not deciding less about important things — you're deciding less about everything else.
What's the difference between a policy and a decision?
A decision resolves one instance; a policy resolves the whole class. "Yes to this podcast invite" is a decision. "I do two podcasts a month, booked by my assistant against written criteria" is a policy — one decision that replaces every future instance. Policies are how you buy back decision capacity without dropping responsibility.
How fast should I decide reversible decisions?
As a working rule: same day, often same hour. A reversible decision made quickly and wrongly costs you one correction. The same decision deliberated for two weeks costs you two weeks — and usually gets corrected anyway, because reality supplies information deliberation can't. Speed on two-way doors is how you buy feedback.
Does a decision journal actually help?
It's the only reliable way to audit your own process, because memory rewrites your reasoning after the outcome arrives. Record the call, the reasoning, your confidence, and how your body read the choice at the time. Review quarterly. Patterns you'd never admit in the moment become undeniable on paper.