The precedent establishes that a prior decision in a past case must be repeated in a current case because both cases are similar. It is a content-independent argument: the anteriority of the decision in a similar case is important, not its quality. The precedent implicates the value of deciding, which is more desirable, for what it is assumed with the argument, than deciding correctly or according to the best consequences. If cases are similar, they must receive the same treatment – optimal or suboptimal. But when are cases similar? In one plausible account, cases are similar when there is internal justification in a syllogism in which the prescription from the past decision is the major premise; the facts of the current case are the minor premise and the conclusion is the result established in the precedent. Thus the precedent is a rule. In another possible account, cases are similar when relevant factual categories are mapped in both cases and it is possible to conclude that they are analogous. The precedent is an analogy. One account assures precedential constraint, but cannot adjust to distinguishing; the other can make room for distinguishing, but cannot offer precedential constraint. – that is a problem of allocating between stability and flexibility. The theoretical possibilities gives rise to empirical questions: is there an ordinary concept, deductive or analogical, of precedents? If the precedent is a rule, its extraction from the prior decision is manipulated by choosing different generality levels? If the precedent is an analogy, are there biases on the determination of relevance to the facts of a precedent case in order to get a desirable result in the current case?