A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
A theory is stable if it has a notion of independence (forking) satisfying certain axioms; instability is witnessed by the existence of an order property or independence property. Stable theories have strong model-theoretic structure (categoricity in all sufficiently large cardinals), while unstable theories can be much more complicated. This stability/instability divide is the fundamental classification in modern model theory.
From your study of forking independence and stability theory, you know that stable theories support a well-behaved notion of independence — an analogue of linear independence in vector spaces — satisfying symmetry, transitivity, and other algebraic axioms. The stability/instability dividing line is the foundational answer to the question: which theories admit such a notion, and what structural properties does it unlock? This classification, developed by Shelah in the 1970s, is one of the deepest organizing principles of modern model theory.
A theory T is unstable if it has the order property: there exist a formula φ(x, y) and elements aᵢ, bⱼ in some model such that φ(aᵢ, bⱼ) holds if and only if i < j. In other words, the formula φ can be used to define a linear ordering of elements — the theory can simulate order. The theory of dense linear orders (like ℚ or ℝ with <) is paradigmatically unstable. Intuitively, order introduces combinatorial complexity: if you can rank elements, you can build exponentially many distinct "profiles" of relationships, which defeats the bounded-type-count that forking independence requires. The order property is the *witness* of instability — its absence is what makes a theory stable.
In stable theories, the absence of the order property has profound structural consequences. The number of complete types over a parameter set A is bounded: a stable theory has at most |T|^ℵ₀ + |A| complete types over A (rather than the maximum of 2^|A|). This bounded type-count is not merely a counting curiosity — it is what guarantees the existence of prime models, saturated models, and the machinery of forking independence. Morley's theorem — that a theory categorical in some uncountable cardinal is categorical in *all* uncountable cardinals — is in essence a stability theorem: categoricity forces stability, and stability provides the tools to classify models up to isomorphism. Stable theories are "tame" in the technical sense: their models can be systematically analyzed and, in the best cases, completely classified.
The stability/instability divide has since been refined into a rich spectrum. Superstable theories are those where forking has the best-behaved independence theory; ω-stable theories (categorical in ℵ₁) are even more structured. Moving outward from stability, NIP theories (those lacking the independence property) include all stable theories plus ordered structures like (ℝ, <, +, ·) and the p-adic numbers — they are unstable but still tame in important respects. Simple theories have a weaker independence notion. Shelah's classification program (or "stability spectrum") aimed to draw all possible dividing lines between tame and wild theories, with stability as the first and sharpest. Whether a theory falls on the stable or unstable side of this line determines, at the deepest level, whether its models can be classified or whether they form an unstructured zoo.