Russell's Set Paradox

The paradox stated precisely

Naïve set theory says that any property defines a set: the set of all red things, the set of all prime numbers, the set of all cats. Most sets do not contain themselves as members. The set of all cats is not itself a cat. But some sets might: the set of all non-cats is not a cat, so it seems to contain itself.

Now consider R, the set of all sets that do not contain themselves. Does R contain itself?

If R contains itself, then by its own definition it should not contain itself (since it only contains sets that don't contain themselves). Contradiction. If R does not contain itself, then it meets the defining condition, so it should be in R. Contradiction. R can neither contain itself nor fail to contain itself without contradiction.

What it did to Frege's program

Gottlob Frege had spent decades constructing logicism: the project of deriving all of mathematics from pure logic, using set theory as the bridge. His Grundgesetze der Arithmetik laid out the formal system. In June 1902, as Frege was completing the second volume, he received a letter from Russell.

Russell wrote with characteristic care: "There is just one point where I have encountered a difficulty." He then described what is now called Russell's Paradox. Frege immediately saw the problem. He added an appendix to the volume, acknowledging that one of his core axioms, Basic Law V, generated the contradiction. He wrote that the discovery had shaken the ground on which he stood.

Frege never fully recovered his program. Russell's letter arrived too late to prevent publication but in time to attach an admission of failure to the work.

The solutions

Two durable solutions emerged. Russell himself developed type theory: a hierarchical system in which every set is assigned a "type" or level. Sets at level n can only contain members at level n-1. This makes self-membership, and therefore Russell's set, syntactically meaningless. You cannot form R because "set containing itself" is a type error.

Ernst Zermelo, independently working on the same problem, produced an axiomatic framework later extended by Fraenkel into ZF set theory. The key move is the Axiom of Separation: rather than forming sets from arbitrary properties, you can only carve subsets out of sets that already exist. There is no "set of all sets" to begin with, so R cannot be formed. Both approaches work by building walls around the naive intuition, not by explaining it away.