It’s worth recalling important passages from James Madison’s famous Report of 1800 in light of the many uninformed criticisms of nullification (e.g., “Why, the courts are our infallible judges!”):
“The resolution of the General Assembly [the Virginia Resolutions of 1798] relates to those great and extraordinary cases, in which all the forms of the Constitution may prove ineffectual against infractions dangerous to the essential rights of the parties to it. The resolution supposes that dangerous powers, not delegated, may not only be usurped and executed by the other departments, but that the judicial department also may exercise or sanction dangerous powers beyond the grant of the Constitution; and, consequently, that the ultimate right of the parties to the Constitution, to judge whether the compact has been dangerously violated, must extend to violations by one delegated authority, as well as by another; by the judiciary, as well as by the executive, or the legislature.
“However true, therefore, it may be, that the judicial department, is, in all questions submitted to it by the forms of the Constitution, to decide in the last resort, this resort must necessarily be deemed the last in relation to the authorities of the other departments of the government; not in relation to the rights of the parties to the constitutional compact, from which the judicial as well as the other departments hold their delegated trusts. On any other hypothesis, the delegation of judicial power would annul the authority delegating it; and the concurrence of this department with the others in usurped powers, might subvert for ever, and beyond the possible reach of any rightful remedy, the very Constitution which all were instituted to preserve.”
The “parties to the constitutional compact” being, of course, the peoples of the states.
To the claim that Madison didn’t really mean what he obviously did mean, and what everyone at the time took him to mean, or that he later claimed he had never meant such a thing, I refer readers to my book Nullification (specifically, pp. 288-290).