The Feedback Trap

March 17, 2026. A product manager schedules her fifth feedback session this month. She has a Notion doc tracking every piece of input she has received, color-coded by theme. Her 360 review scores are improving. Her skip-level says she is "highly coachable." Her product, which she has been iterating based on all this feedback, has steadily declining retention. Nobody has connected these two facts. The feedback loop is working exactly as designed.

Most feedback doesn't improve performance. It improves performance on the metrics used to measure performance, which are proxies for what you actually care about. The gap between those two things is where the majority of feedback culture's value goes to die.

We treat feedback the way we treat vegetables: obviously good, more is better, the failure to seek it is a character flaw. We have built entire industries on this belief — 360-degree reviews, NPS scores, real-time coaching dashboards, A/B testing everything. The premise is simple: tell people what they're doing wrong and they'll do it right. The premise is largely false.

Goodhart's Law, Applied to You

The economist Charles Goodhart observed that "when a measure becomes a target, it ceases to be a good measure." This was originally about macroeconomic policy. It describes exactly what happens when you build a feedback system around human performance.

You receive feedback that your presentations run long. You shorten them. Your feedback scores improve. But the presentations were long because the ideas were genuinely complex, and shortening them required simplifying them, and now you're confidently presenting simple versions of complicated things to audiences that needed the complicated version. The feedback loop worked. The presentations got worse.

You receive feedback that your code reviews are too critical. You soften your language. Feedback improves. But the criticism was warranted, and now your team ships bugs you would have caught. You receive feedback that you're "hard to reach." You become more available. But availability was the wrong thing to optimize — what mattered was your judgment about which interruptions were worth taking, and you've traded that judgment for accessibility.

In each case, the feedback loop has worked precisely as designed and produced precisely the wrong outcome. You optimized for the measure and corrupted the thing being measured. The more responsive you were to feedback, the worse you performed at the underlying goal.

The Validation-Seeking Circuit

The second problem is structural: feedback systems don't distribute feedback evenly. They amplify the feedback people most want to hear.

Research on feedback-seeking behavior consistently shows that people seek feedback from sources likely to be positive far more than from sources likely to be critical. We ask our friends to review our writing. We present work to colleagues who share our assumptions. We design our feedback surveys to make it easy to express satisfaction and hard to express dissatisfaction — because the survey is, itself, subject to the same incentive dynamics as everything else.

This isn't malice. It's the rational behavior of someone who wants to feel like they're getting feedback while minimizing the discomfort of actually receiving it. Feedback-seeking becomes a performance of openness rather than an actual mechanism for learning. The result is a heavily biased sample — weighted toward validation — treated as representative evidence of growth.

The person who most visibly seeks feedback is often the person whose model of themselves is most insulated from challenge. They have built an elaborate feedback architecture that produces a continuous stream of mildly positive information. This feels like learning. It is the opposite.

Frequent Feedback Destroys Intuition

The third problem is the most counterintuitive: frequent feedback degrades the development of expertise.

Expertise requires building intuition — the capacity to make good judgments without being able to fully articulate why. Intuition is built through a specific process: you make a judgment, you wait for the outcome, you update your internal model. This requires tolerating extended uncertainty. You have to sit with not knowing whether you were right, for long enough that the answer eventually arrives and you can connect it back to the decision.

Frequent feedback short-circuits this process. Instead of developing autonomous judgment, you develop sensitivity to the feedback signal. You learn to produce outputs that generate good feedback, which is a related but distinct skill from producing outputs that achieve good outcomes. The pianist who receives granular feedback on every measure never develops the capacity to hear the piece whole. The manager who receives weekly coaching never develops the independent read of a room.

This pattern appears consistently across domains. Golfers given frequent, granular feedback on swing mechanics plateau below golfers given sparse outcome feedback — did the ball go where you intended? In medicine, surgery residents trained with high-feedback simulation perform worse on rare, high-stakes procedures than those who trained with fewer but more consequential exposures. In writing, authors who workshop every chapter often produce books that are locally polished and globally incoherent.

The mechanism is the same each time: frequent external feedback replaces the development of an internal standard. You become calibrated to the feedback, not to the outcome. When the feedback is removed — when you're alone with the problem — there's nothing there.

What Actually Works

The research on expertise development is consistent: the best feedback is sparse, delayed, and outcome-focused.

Sparse means less than you think. A volume of feedback you cannot integrate in a reasonable period is noise. Most feedback cultures provide far more than anyone can actually process, which means the marginal feedback is not improving performance — it's just creating a sensation of being worked on.

Delayed means waiting for outcomes, not opinions. Immediate feedback is addictive and often useless. Knowing right away whether your estimate was right doesn't help you understand your model — you need to know whether the reasoning that produced the estimate was sound, which you can only assess against what actually happened. Delayed feedback is uncomfortable. That discomfort is not a bug. It's the mechanism by which you develop judgment rather than reflexes.

Outcome-focused means asking whether the thing achieved what it was supposed to achieve, not whether specific behaviors matched a template of correct behaviors. Behavior feedback trains compliance. Outcome feedback trains judgment. The two are different skills and produce different people.

The Practical Takeaways

Treat feedback as a drug with a recommended dose. The right amount depends on your skill level (novices need more, experts need less), the domain (fast-moving environments benefit from faster feedback), and what you're trying to develop (mechanics versus judgment require different ratios).

Be suspicious of feedback that arrives continuously. If you're receiving a constant stream of feedback, ask what it's optimizing for. The feedback cycle itself — the act of giving and receiving — generates social rewards that can become self-sustaining independent of any actual improvement in the thing being practiced.

Build deliberate feedback holidays. Periods where you work without external feedback — making judgments and watching what happens — are not absence of learning. They are how intuition forms. The uncomfortable feeling of operating without external validation is not a signal that something is wrong. It is the sensation of developing internal standards.

Ask different questions. Instead of "what did I do wrong?" ask "what would a better outcome have looked like, and what would have produced it?" The first question produces behavior feedback. The second produces outcome feedback. One trains you to follow instructions; the other trains you to make better decisions.

The deepest problem with feedback culture is not that feedback is bad. It's that feedback has been framed as a virtue rather than a tool. When seeking feedback is what good people do, the incentive structure stops being about whether the feedback is improving anything and becomes about being the kind of person who seeks feedback. That's a different game entirely, and you can win it while getting steadily worse at everything that matters.