It is now orthodox in economics to explain the desirability of government by pointing to the existence of public goods and the difficulties which the private market would have in dealing with them.' It is the theme of this article that the operations of the government itself raise a new and extremely difficult public-goods problem. Consider a federal judge who is making a decision on some case. The decision is a direct generation of externalities by him-the externalities falling on the participants in the case. In addition to these rather restricted externalities, he is participating in the production of a public good: law enforcement. If he should, for example, decide to bring in a decision which is not in accord with the law, he not only changes the situation in this particular case, but he also, in a sense, changes the law for the future. This latter phenol tenon is most important in those cases in which judges are actually making law; their decisions generate a pure public good (or public bad). Note, however, the judge himself has almost no private incentive to reach the "right" decision. Suppose that he reaches a decision which leads to an increase in the crime rate; the rise in the crime rate might have a significant effect on the country as a whole, but the likelihood that this particular judge or his immediate family would be victims is very low compared with the damage inflicted on others. The same principle would apply if the judge extended the definition of some crime to include acts which had previously not been regarded as within the statute. Once again, it would be extremely unlikely that he himself would suffer any damage from the act; hence, he has created a public good (or bad) without significant private gain (or loss). He cannot be removed front office, nor can his salary be reduced. In practice it is also impossible to even deny him pay raises which are given to other members of the judiciary. It is true that in some cases the judge's decision will not be final;