Naive set theory treats a set as any well-defined collection of objects satisfying a property, formalized as the unrestricted comprehension principle: for any predicate P(x), the collection {x : P(x)} is a set. Developed by Cantor in the 19th century, this approach successfully handles finite sets, infinite sets of numbers, and transfinite arithmetic. However, the system is logically inconsistent: certain self-referential predicates generate outright contradictions, as Russell famously demonstrated. Axiomatic set theory was developed to preserve the power of naive set theory while eliminating these inconsistencies.
Begin by working through Cantor's basic constructions — natural numbers, rational numbers, and the reals as sets — to appreciate what naive set theory enables. Then study the specific paradoxes (Russell's, Burali-Forti's) that reveal its limits. The contrast between what naive set theory can build and why it fails motivates every subsequent axiomatic choice.
When Cantor developed set theory in the late 19th century, the underlying principle seemed obvious: any well-defined collection of objects is a set. If you can state a property clearly, you can collect all objects satisfying it into a set. This principle — unrestricted comprehension — is the heart of naive set theory. It is powerful enough to construct the natural numbers, the rationals, the reals, ordinal arithmetic, and much of classical mathematics in a unified framework.
The formal statement of unrestricted comprehension is: for any predicate P(x), the collection {x : P(x)} is a set. This is not vague informalism — it is a precise axiom schema. It lets you form sets like {x : x is a prime number}, {x : x is a real number}, or {x : x is a set with exactly three elements}. Cantor used it to explore infinite sets and developed groundbreaking results about different sizes of infinity. For several decades, this seemed entirely adequate.
The crisis came in 1901 when Bertrand Russell wrote to Frege with a devastating observation. Consider the predicate P(x) = 'x ∉ x' — the property of not containing yourself as a member. Unrestricted comprehension says this defines a set R = {x : x ∉ x}. Now ask: does R contain itself? If R ∈ R, then R satisfies the defining property, so R ∉ R — contradiction. If R ∉ R, then R satisfies the defining property, so R ∈ R — contradiction again. Either assumption leads to a logical impossibility. This is Russell's paradox, and it is not a subtle or avoidable mistake — it is a direct, unavoidable consequence of unrestricted comprehension.
The paradox revealed that naive set theory, despite its intuitive appeal and mathematical power, is logically inconsistent: you can derive a contradiction from its axioms, which means every statement is provable in the system (ex falso quodlibet). This does not mean Cantor's results about infinity were wrong — they survive in the successor theories. But the foundations needed to be rebuilt. Zermelo and Fraenkel developed ZFC axiomatic set theory, which replaces unrestricted comprehension with restricted separation (you can only form subsets of existing sets) and adds carefully chosen existence axioms to recover what mathematics needs.
The lesson of naive set theory is that mathematical intuition about 'collections' is not automatically safe. Self-referential constructions — sets that talk about their own membership — are especially dangerous. Axiomatic set theory's main technical achievement is designing a system expressive enough to do all of mathematics while preventing the self-reference that leads to paradox. As you move into ZFC, each axiom can be read partly as a targeted response to the failures of naive comprehension.