The Case of the Speluncean Explorers

Five explorers are trapped underground with no food and little chance of surviving until rescue. They agree to draw lots. One of them will die so the others can live. Four eventually return to the surface and face trial for murder.

Lon Fuller's fictional case has lasted because the facts are simple and every judgment feels dangerous.

Apply the statute literally and the survivors are guilty. Treat the cave as a condition outside ordinary civil society and their agreement may create its own rule. Judge by consequences and four saved lives may outweigh one death. Follow public sentiment and mercy may look more legitimate than punishment.

Each argument can be made coherently. They do not converge.

Analysis reveals the conflict

Many difficult decisions are described as information problems. We assume that one more report, one more expert, or one more model will reveal the correct answer.

The cave case shows a different kind of difficulty. The conflict remains even when everyone agrees on what happened. The disagreement concerns which principle should govern the facts.

Should law remain stable when the result feels unbearable? Should an emergency change the rules? Should consequences outweigh procedure? Should a judge interpret the law or ask another institution to show mercy?

More information can sharpen each argument. It cannot make the principles identical.

This distinction matters outside law. A career decision can be analyzed through income, probability, accumulated experience, future options, identity, and the cost of delay. A product decision can be analyzed through growth, user autonomy, safety, and commercial survival. The analysis is useful because it tells us what each choice costs.

It may never tell us which cost we must prefer.

AI can make indecision look intelligent

AI is exceptionally good at generating the arguments around a decision. It can build scenarios, find counterexamples, simulate critics, and expose a tradeoff that one person missed.

That is valuable. It can also extend avoidance.

If the real conflict concerns values, another round of analysis may produce a more sophisticated map without producing a choice. The person feels busy because the reasoning improves. The decision remains untouched.

This is one of the limits of outsourcing judgment. A model can describe the consequences of a choice. It does not have to live inside the life, institution, or relationship that follows.

Responsibility cannot be generated on behalf of the person who will carry it.

A mature decision names its loss

We often call a decision rational when its disadvantages have been hidden well enough. A more honest standard is whether the decision maker can name the principle being protected and the cost being accepted.

Choosing stability may mean giving up speed. Choosing a discontinuous path may mean surrendering the value of accumulated experience. Protecting individual autonomy may make a product less efficient. Maximizing efficiency may reduce the space in which a person can still exercise judgment.

There is no need to turn every choice into a moral drama. Many decisions are reversible and should be treated lightly. The hard ones become clearer when we stop asking for an answer without loss.

The explorers' case offers no comfortable verdict. That is its use.

A decision can be well reasoned and still leave something valuable behind. Maturity begins when we stop treating that remainder as proof that the decision was wrong.