Punish or prevent

A law forbidding an act does not accept that the act is normal. It accepts that the prohibition on its own is not enough.

People know they must not steal or kill. Yet knowing a moral rule does not remove crime. The causes of behaviour include gain, impulse, group pressure, addiction, desperation and confidence in getting away with it.

Punishment performs several possible functions: it restrains a dangerous person, deters others, expresses public condemnation, compensates for damage, creates an opportunity for reform and satisfies the wish for revenge.

These functions do not coincide. A system can isolate well and reform badly. It can punish harshly and prevent almost nothing. It can display activity after the event without reducing the causes of the next one.

So the question “is the punishment severe enough?” is often framed too narrowly.

What matters for prevention:

• the probability of detection;
• the speed of response;
• how clear the rule is;
• living conditions;
• access to treatment;
• education;
• protection of potential victims;
• the possibility of behaving lawfully.

The police do not prevent every crime and do not solve every theft. That does not prove they are entirely useless. The fire brigade does not arrive before the smoke either.

A public service, however, should be judged by what it actually does, not by its uniform, its ranks and the number of its reports.

Punishment is justified not because the suffering of the guilty restores some cosmic balance. It is justified where it protects people, reduces repetition and stays proportionate.

A dog genuinely does not understand the criminal code. A human understands words, but that does not yet mean words alone are enough.

The difference between severity and probability is visible in a warehouse. A director losing stock can announce that theft means dismissal in disgrace and a report to the police — and go on losing goods off the delivery notes. The same director can introduce a weekly stock check and a camera over the loading bay. The first raises the price of being caught, the second raises the chance of it, and usually works faster. After the stock checks start, theft drops without a single dismissal — simply because the gap between the loss and its discovery shrinks from a quarter to a week.

Severity is cheap for whoever announces it: it requires no accounting, no time and no money. So systems that are bad at detection almost always compensate with the volume of the sanction.