Palace/Observatory

The Agency Gap

agencyleveragestrategyexecutionhigh-performance

The Agency Gap

Among people who have passed a high capability bar, the primary differentiator is not ability — it's the decision to act without being told. Most people have more permission than they use. The gap between what they could do and what they actually do is the agency gap.

This is not primarily a motivation problem or a courage problem. It is a strategic default problem: humans are not automatically strategic, and the environments that select for high-ability people systematically reward following instructions over finding the right thing to do.

Three Mechanisms That Create the Gap

1. Non-strategic defaults

Most human behavior is not optimized for goals even when the person cares about the goal. humans-are-not-automatically-strategic describes the verbal-motivational gap: most people can state their goals clearly and still spend most of their time on actions that don't move toward them. The 8 strategic heuristics Anna Salamon identifies — ask what the goal is, define success explicitly, track progress, test conjectures, focus on high-leverage actions — are not automatic. They require deliberate application every time.

The implication: someone who happens to apply these heuristics even occasionally will dramatically outperform the average, not because they're smarter but because most people's defaults are non-strategic.

2. The moat of low status

Even when people identify high-leverage actions, they don't take them because doing so requires crossing moat-of-low-status — the period of visible incompetence when learning or entering a new domain. Asking obvious questions. Being wrong publicly. Trying things that don't work.

The moat is what keeps most high-value skills and actions underexploited. It's not that the edges are hidden — they're just uncomfortable to pursue. The people who exploit them are not smarter; they're more willing to be visibly incompetent for a period. This is why relentlessly-resourceful people are rare: resourcefulness requires being willing to try things that might not work, in front of people who might notice.

3. Waiting for permission

The third mechanism is structural. Most institutions — schools, large companies, credentialed fields — reward following instructions. The selection process that produces high-capability people is largely a selection process for compliance: do the assigned work well. This trains the expectation that someone will assign the high-leverage work.

impact-agency-and-taste names this directly: "You have permission to exercise both [agency and taste] — most people wait for someone to assign them high-leverage work rather than finding it themselves." The permission already exists. The waiting is a habit formed by a decade of environments where waiting was correct.

Why the Mechanisms Reinforce Each Other

The three mechanisms compound:

  • Non-strategic defaults → you don't find the high-leverage work in the first place
  • The moat of low status → even if you find it, you avoid the visible incompetence of pursuing it
  • Permission-waiting → even if you'd cross the moat, you wait for someone to tell you to

Each mechanism provides a reason to not act. Together they create a stable equilibrium where capable people do assigned work well and wait.

The equilibrium breaks when someone decides to apply strategic heuristics before being asked, cross the moat before being told it's safe, and act on high-leverage work before receiving an assignment. This is what "high-agency" actually means — not a personality trait but a specific set of decisions made repeatedly.

What Closes the Gap

Strategic heuristics applied explicitly. Before starting work, ask: what is the actual goal? What would success look like? What actions have the highest leverage toward it? These questions are not automatic — they must be asked on purpose. The alternative is defaulting to the most visible or most recently assigned task.

Crossing the moat on purpose. The skill is distinguishing between "this is uncomfortable because it requires public incompetence" and "this is uncomfortable because it's actually wrong." The moat is discomfort that signals learning. Thriving in the moat — not just tolerating it — means the scrutiny stops distorting your decisions.

Treating permission as granted until explicitly denied. The default assumption should be: if this work is high-leverage and nobody else is doing it, I can do it. Waiting for explicit permission is a choice, not a requirement. The practical test: work backwards from the root goal rather than the assigned task. You often have more latitude to reframe the work than your position suggests.

Developing taste to find the work. The agency gap also has an information problem: even with permission and willingness, you need to know which actions are high-leverage. taste-as-predictive-models is the capacity to find these — the question "what does it seem like everyone is mysteriously bad at?" is a pointing device for where your taste is strongest and where the gap is largest.

The Diagnostic

The agency gap shows up as:

  • Spending time on tasks that feel productive but don't move the real goal
  • Waiting for feedback before trying something new
  • Knowing the high-leverage thing but not doing it because "it's not my job" or "nobody asked"
  • Avoiding domains where you'd have to be visibly incompetent first
  • Defining your role by what you're assigned rather than what the situation needs

The inverse: someone closing the agency gap will look, from the outside, like they're doing more than their job. They are — because their job as they've defined it includes finding the right work, not just doing the assigned work.

Connections

Sources