Research case library

31 cases to look differently at knowledge and AI.

Start with a story that interests you. Then explore what we know about it, possible explanations and what AI might change. Each case includes sources and questions for further research.

Read the cases in Dutch or English. They bring together sources, arguments and proposed tests; independent assessment of the content and translations is not yet complete.

01Chinese RoomCorrect output is not the same evidential claim as understanding.Read the case 02Simpson’s ParadoxA figure can be statistically correct and still answer the wrong management question.Read the case 03Abilene ParadoxA group may display complete consensus on a choice that none of its members privately wanted.Read the case 04Goodhart’s Law & Campbell’s LawOnce a measure steers behaviour, it changes from an observation tool into part of the system.Read the case 05Ironies of AutomationThe more routine work automation takes over, the harder the remaining human work may become.Read the case 06Ellsberg ParadoxA precise number can make uncertainty easier to grasp while also concealing which parts of the probability model itself remain uncertain.Read the case 07Ship of TheseusRoutine output may look the same while rationale, provenance and recovery capability have meanwhile been replaced piece by piece.Read the case 08Information CascadesMany agreeing answers may originate from a single information source and therefore contain less evidence than they appear to.Read the case 09Streetlight Effect / Drunkard’s SearchAn analysis may be extremely precise about exactly the part of reality that happened to be easy to observe.Read the case 10Braess’s ParadoxAn additional smart route may help each individual choose while still making the overall system worse.Read the case 11Mary’s Room / Knowledge ArgumentComplete information and practical competence are different claims — without this resolving Mary’s philosophical argument.Read the case 12Berkson’s Paradox / Collider BiasAn association may arise in your dataset precisely because you selected on a common effect.Read the case 13Volunteer’s DilemmaEveryone may agree that something needs to happen and still wait because one volunteer would be enough.Read the case 14McNamara FallacyWhat is hard to measure may disappear from management’s view without becoming any less important.Read the case 15Jevons Paradox / Rebound EffectMaking a task cheaper may increase total use so much that some of the efficiency gain disappears.Read the case 16Prisoner’s DilemmaCall a knowledge-sharing problem a Prisoner’s Dilemma only if not sharing really is the individually dominant local choice.Read the case 17Sorites / Heap ParadoxRemoving one grain seems unable to turn a heap into a non-heap. Repeating that step conflicts with the clear endpoint: one grain is not a heap.Read the case 18Pluralistic IgnorancePublic agreement may be a poor estimate of what group members actually believe in private.Read the case 19Chesterton’s FenceFirst understand which problem the process element was meant to solve. Then test whether that problem still exists and whether the element still helps. Only then decide whether to retain or remove it.Read the case 20Tragedy of the CommonsKnowledge itself is not consumed; verification, attention and stewardship can be.Read the case 21Brain in a VatA body of information may be perfectly coherent internally yet lack an independent route to external reality.Read the case 22Friendship ParadoxYour network contacts are not a random sample: highly connected nodes naturally appear more often.Read the case 23Newcomb’s ProblemYour choice can provide evidence about an earlier allocation without changing that earlier allocation.Read the case 24Cobra Effect / Perverse IncentivesA reward changes not only how much of the desired behaviour you get, but also which new strategies become rational.Read the case 25Ultimatum GamePeople do not necessarily accept every positive outcome when they perceive the allocation as unfair or illegitimate.Read the case 26Monty Hall ProblemWhat you see depends not only on the world, but also on the rule that made particular information visible to you.Read the case 27Trolley ProblemWhat the majority chooses in a dilemma is data about human preferences, not an automatic justification for machine policy.Read the case 28Birthday ParadoxA tiny probability per pair may become large at system level when the number of possible pairs grows quadratically.Read the case 29Dunning–Kruger EffectAssess how much stronger the observed pattern is than patterns produced by an appropriate null model with measurement noise and the same analytical method.Read the case 30Fermi Paradox / Great FilterNot finding something is evidence of absence only if you know how likely you should have been to find it had it been present.Read the case 31Map–Territory ProblemThe map is not the territory. A description is not the process. A knowledge base is not the knowledge.Read the case

How to read the library

A different perspective

A case helps you recognise a pattern and discuss it. Whether that pattern applies to your situation needs closer examination.

From story to research

Each case distinguishes the story, the available evidence and questions that remain open. This helps you judge how far the comparison holds.

See it differently?

Know a better source, a counterexample or an experience that casts the case in a different light? Let us know.

Challenge the research

From a case to a research question

Want to investigate a question further? Explore K-point and the other research programmes.

Explore the research programmes →