A bright line and a rule of thumb fail in opposite directions

Two things that get called the same word

A rule specifies what happens without consulting the particulars: no purchases over a stated amount without a night to sleep on it, no email after a stated hour, never invest in something you can’t explain. A heuristic is a shortcut for producing a judgement quickly: take the option with the fewest unknowns, weight the most recent evidence heavily, prefer the arrangement that fails visibly.

The difference is what happens when the case is unusual. A rule that bends for unusual cases has stopped being a rule, because every case feels unusual to the person inside it. A heuristic is designed to bend, and its value lies in getting close to a decent answer cheaply rather than in binding anybody to anything.

What a rule is actually buying

A rule buys freedom from the judgement of a compromised decider. The person who wrote it had no stake in Tuesday afternoon; the person facing the choice does. Whenever your judgement in the moment is predictably worse than your judgement in the abstract — under fatigue, in front of a persuasive seller, at the end of a long negotiation — a rule made earlier is a better instrument than a fresh assessment.

It also buys enforceability. A bright line can be checked by somebody else, applied consistently across cases, and defended without relitigating the merits each time. This is why institutions run on rules despite everybody involved knowing that rules produce absurd results at the edges: the alternative isn’t perfect case-by-case wisdom but inconsistent judgement applied by whoever happens to be in the chair.

What a rule costs

Every rule is a bet that the cost of the cases it gets wrong is smaller than the cost of the discretion it removes. Since the rule was written without knowing which cases would arrive, it will certainly be wrong sometimes, and the wrongness will be conspicuous when it happens. That visibility is what makes rules unpopular in exactly the moments they’re doing their job.

The characteristic failure is a rule that outlives its reason. Conditions change, the reasoning that produced the line is forgotten, and the line remains because removing it would require somebody to take responsibility. Writing down why a rule exists, alongside the rule, is a small habit that prevents a surprising amount of this — it converts the rule into something that can be argued with on its own terms rather than defended out of inertia.

When a simple rule beats a careful analysis

There is a genuine research literature arguing that simple decision rules using very little information sometimes outperform elaborate models, particularly when data are scarce, conditions are unstable, or the model has been fitted to a past that won’t repeat. The proposed reason is that a complex model has more freedom to fit noise, and that flexibility becomes a liability out of sample.

The claim has been argued about vigorously, and the fair summary is that it holds in some environments and not others rather than being a general truth. What isn’t contested is the underlying trade: complexity buys accuracy when you have plenty of relevant data and a stable world, and costs accuracy when you do not. Knowing which of those you are in is the substantive question, and it is not answered by preferring simplicity or sophistication as a matter of taste.

Choosing between them for a particular decision

Three questions do most of the sorting. How often does this decision recur — a one-off does not repay the cost of writing a rule, while anything that arrives weekly does. How reliable is your judgement at the moment of choosing, compared with now. And how expensive is a wrong call in the unusual case, since a rule that occasionally produces a disaster is worse than one that occasionally produces an annoyance.

The answers point at a spectrum rather than a binary. Hard rules for recurring decisions where in-the-moment judgement is known to be poor and the exceptions are cheap. Heuristics for recurring decisions where the cases genuinely differ and being roughly right quickly is worth more than being precisely right slowly. Full deliberation for the small number of decisions that are rare, large and hard to reverse.

The exception clause is where this lives or dies

Most personal rules die of a well-argued exception, and the argument is always good, because the person making it has both the motive and all the details. The repair is not to forbid exceptions, which produces a rule so rigid it gets abandoned entirely, but to specify in advance what an exception requires: a named condition, a second person, a delay.

A rule with a costly, pre-specified escape hatch is far more durable than one with none, and this is the same structure that makes a good commitment device work. What matters is that the price of the exception is set while you are calm and paid while you are not. A rule you can suspend silently, on your own authority, in the moment, is a preference with a formal name.