7 Decision-Making Mistakes Smart Founders Keep Making

The expensive decision mistakes aren't intelligence failures — they're architecture failures: running every choice at one speed, optimizing inside a choice set you never questioned, shopping for consensus, growing pros-and-cons lists instead of deciding, betting big enough to leave the game, discarding the body's data, and never once deciding how you decide. Smart founders make all seven, because intelligence is no defense against a process you never examined.

That last clause is the thesis of Paper II of the DaD Papers, which builds a full five-level taxonomy of decision failure with the math and the cures. What follows is the field-guide version: the seven patterns that show up over and over in capable people, each with its cure.

1. Running every decision at one speed

Some founders deliberate everything — reversible calls get committee treatment, weeks pass, feedback never arrives. Others sprint everything — including the one-way doors, where "we'll adjust later" isn't available. Both are the same mistake: no triage. The cure is one question asked before engaging with any decision's content: if this goes wrong, can I undo it cheaply? Reversible: decide today. Irreversible: slow down on purpose. Speed is not a personality trait; it's a per-decision setting.

2. Optimizing the choice instead of the choice set

A founder agonizes brilliantly over candidate A versus candidate B — never asking who decided those were the options. The most consequential failures happen a level up from the decision, at the level of perception: the option that never made the list because you lacked the distinction to see it. Before optimizing, interrogate the menu. Who drew this boundary? What would a third option look like? A perfect decision among poor options is a poor outcome, executed well.

3. Consensus shopping

Asking one person who holds distinctions you lack is borrowing perception — wise and cheap. Asking twelve people is something else: responsibility diffusion dressed as diligence. You get the average of twelve risk tolerances, none of whom carry your consequences, and the tell is unmistakable — you keep asking until someone says what you already wanted to hear, then call it validation. Cap your consultations at two or three people chosen for specific distinctions, and let the last vote be yours, cast alone.

4. The pros-and-cons list that grows until both columns are useless

The list has a fatal flaw: every entry weighs one. "Slightly longer commute" sits next to "wrong co-founder" as equals, and adding entries feels like progress while actually postponing the act the entries were supposed to serve. Meanwhile the decision ages, and latency bills by the day whether or not you eventually choose well. If you must make a list, force a ranking: which one pro and which one con actually dominate? The other eleven rows were procrastination with formatting.

5. Betting big enough to leave the game

The Kelly Criterion's deepest lesson isn't the formula — it's that overbetting destroys you even when your edge is real. Variance arrives before the long run does. Founders violate this in all three currencies: all the cash on one launch, all the calendar on one unproven initiative, all the reputation on one unverified claim. The cure is a pre-committed sizing cap, written into your decision framework on a calm day: no single bet, however delicious, may be sized so that losing it removes you from the table. Survival is the prerequisite for compounding, and compounding is the whole game.

6. Discarding the body's data

Damasio's Iowa Gambling Task subjects showed measurable bodily responses to bad options before their conscious minds caught up — and patients who couldn't feel became catastrophic deciders, analysis intact. Yet the standard founder posture is to treat every somatic signal as bias to be overridden: the contraction before signing, dismissed as nerves; the flatness about the "obviously correct" strategy, dismissed as fatigue. Some of it is noise. But discarding the channel wholesale means deciding with instruments missing. Log the body's read — expansion or contraction — next to the analysis, and let your journal tell you over time where it's calibrated. That's the working method in frameworks versus gut feel.

7. Never deciding how to decide

The root mistake underneath the other six. Most founders' decision processes were never chosen — they accreted from family, first bosses, industry culture. Triage rules, sizing caps, consultation limits, admissible signals: all of it inherited, none of it examined. The cure is the meta-decision — stepping up one level and choosing the process itself, which is the subject of what is a decision about decisions and, in full, of Paper II.

You can't out-think a process you've never looked at. The first decision worth making well is the decision about decisions.

What the seven have in common

Each one survives intelligence because each one lives upstream of intelligence — in the perception layer (mistakes 2 and 6), the process layer (1, 3, 4, 7), or the sizing layer (5). That's also the deeper claim of the trilogy: deciding well isn't a trait. It's an architecture — distinctions drawn, signals admitted, rules pre-committed — and architectures can be rebuilt by anyone willing to look at their own.

FAQ

What's the single most expensive decision mistake?

Running every decision at one speed. Deliberating reversible calls burns weeks that speed would have converted into feedback; sprinting irreversible calls converts recoverable situations into permanent ones. One triage question — can I undo this cheaply? — corrects both directions at once.

Why do pros-and-cons lists fail?

Because they weigh nothing. Every entry counts as one, so "slightly annoying commute" sits beside "wrong co-founder" as equals, and the list grows until both columns are equally useless. Worse, list-building feels like progress, so it postpones the actual act of deciding while consuming the energy that act needed.

Is asking for advice before deciding a mistake?

Asking one or two people who hold distinctions you lack is how you borrow perception. Asking twelve people is consensus shopping — you're not gathering information, you're diffusing responsibility, and you'll get the average of twelve risk tolerances instead of a decision. The tell: you keep asking until someone says what you wanted to hear.

How do I know if I'm making the "wrong question" mistake?

Check whether you're optimizing inside a choice set you never questioned. If the options are A or B, ask who drew that boundary and what a C would look like. Distinction precedes decision — a perfect choice among poor options is still a poor outcome, executed well.