← Research
Research

Kind and wicked learning environments

Why twenty years in a job does not always produce twenty years of judgement.

Last reviewed: 26 September 2026 · Next review due: 26 September 2027

The value of experience depends on the structure of the feedback it returns, not on the amount of it. Hogarth's distinction is the most useful single idea available for understanding why judgement is hard to train, and why removing the junior work removes more than the output.

Questions this page answersAll 996 questions this research covers

A kind learning environment returns feedback that is quick, accurate and drawn from a complete sample, so experience teaches. A wicked one returns feedback that is delayed, partial, or censored by the learner's own decisions, so experience teaches the wrong lesson and teaches it confidently. The distinction is Robin Hogarth's, and it explains more about judgement than any amount of discussion about talent.

The answer, in one line

One where the feedback a person gets from their own decisions is delayed, incomplete, biased by those decisions, or drawn from a sample their decisions censored.

Share as a card

Definition#

Kind and wicked learning environments: Hogarth's distinction between settings whose feedback structure allows experience to build accurate judgement, and settings whose feedback structure does not. The property belongs to the environment, not to the person in it.

Share this definition as a card

What makes an environment wicked#

Four things, and a setting needs only one of them.

Delay. If the outcome arrives months or years after the decision, the link between the two is hard to hold, and by the time it arrives a dozen other things have happened that could also explain it.

Noise. If good decisions frequently produce bad outcomes and bad decisions frequently produce good ones, then outcomes are a poor teacher even when they arrive promptly. This is the ordinary condition of decisions made under uncertainty.

Censoring. The sharpest of the four. If you only see what happened to the candidates you hired, the deals you took and the strategies you chose, then the sample you learn from was selected by your own judgement. A recruiter who believes they can spot talent will be confirmed by every good hire and will never meet the good people they rejected.

Reactivity. Where your own action changes the outcome, success and failure both become ambiguous. A manager who intervenes in a project that then succeeds cannot tell whether the intervention helped.

Why this matters more than it sounds#

The usual story about expertise is that it comes from hours. Hogarth's account says the hours only pay in settings where the feedback is honest, and that in the other settings the same hours produce something that looks like expertise from outside and from inside: fluency, speed, confidence, a stock of remembered cases. What they do not produce is accuracy, and the person cannot tell the difference from within, because the signal that would tell them is the signal the environment fails to return.

This is the same territory as clinical versus actuarial judgement, where simple rules hold their own against experts. The environments where rules do best are the wicked ones, and that is not a coincidence. The same mechanism is why noise audits tend to surprise the organisations that run them.

What this says about AI and the removal of junior work#

Most of a professional job is wicked. What is generally kind is the small, fast, correctable work at the bottom: the draft that comes back marked, the calculation that is checked, the note that gets rewritten in front of you. Those tasks return feedback quickly, accurately and completely, which is what made them teach anything.

Those are also the tasks being handed over first. The argument at the missing rungs is usually put in terms of how many junior roles exist. The feedback account says something narrower and harder to fix: even where the role survives, the part of it that was kind may not. A person can hold a job for a decade, accumulate experience the whole time, and end the decade with the confidence of a decade and the judgement of a year, because nothing they did returned a signal they could learn from.

It also gives a practical test, the fourth question of the use-or-keep test. Before handing a task over, ask what it tells you and how fast. If the answer is nothing much, the case for keeping it is weaker than instinct suggests. If the answer is quickly and honestly, that is a rare asset and worth defending.

What can be done about a wicked environment#

The environment can sometimes be made kinder, which is more productive than trying to make the person better at learning from a setting that will not teach.

What this does not establish#

The distinction is a lens rather than a measurement. No instrument scores an environment for kindness, and most real settings are mixed rather than one or the other. Hogarth's account is theoretical with experimental illustrations, not a body of field trials, and the claim that a given profession sits in the wicked half is usually an argument rather than a finding. Nothing here shows that making an environment kinder improves outcomes at work, only that the mechanism by which experience teaches is the one being described.

Key sources

Essay · SS-2026-319

Cite this page

Hirji, R. (2026). Kind and wicked learning environments. The SuperSkills evidence base, SS-2026-319. https://thesuperskills.com/research/kind-and-wicked-learning-environments. Last reviewed 26 September 2026.

An evidence review by Rahim Hirji, not peer-reviewed research. For a material claim, cite the underlying study as well; every study here carries its own permanent link.

How citations and IDs work
Questions answered on this page

What is a wicked learning environment?

One where the feedback a person gets from their own decisions is delayed, incomplete, biased by those decisions, or drawn from a sample their decisions censored. In such an environment experience still produces confidence, and the confidence is not matched by accuracy. The term is Robin Hogarth's.

What makes a learning environment kind?

Feedback that arrives quickly, is accurate, and covers the full range of outcomes including the ones the learner did not choose. Chess, weather forecasting and anaesthetics are commonly given as kind or nearly kind. Hiring, strategy and most management decisions are not.

Why does this matter for AI?

Because the tasks being handed to machines first are often the ones that returned the fastest and cleanest feedback, which means the repetitions being removed are the ones that were actually teaching. A professional can keep a job whose feedback is wicked and lose the tasks whose feedback was kind.

In this hub

Definitions

The terms this field uses, defined against their primary sources.

Ask the evidence
What does the evidence actually show?What should our board be asking about this?Where does Rahim disagree with the consensus?
Bring this into your organisation

If this describes something happening in your teams, say so.

Keynotes, board sessions and advisory work, drawing on research across more than 200 organisations in 30 countries. Tell me the room, the date and the shift you need. A reply within 24 hours.

Start a conversation

Topics and audiences  ·  All research

Box of Amazing

Rahim’s free weekly letter on AI and human capability

If this was useful, the weekly letter is where the thinking happens first. Most of what ends up on this site starts there. Weekly essays on AI, capability and the future of work. Read by 25,000 people, every week since 2017. Free, and one click to stop.

Opens Substack to confirm. No pitch in it, unsubscribe in one click, and nobody follows up because you read something.

Running an event, or responsible for how AI arrives in your organisation? Keynotes  ·  Advisory  ·  Boards  ·  Enquire