Five times mathematics discovered that the exception was the rule.
1.
In 1872, Karl Weierstrass presented a function that is continuous everywhere and differentiable nowhere. Continuous means no jumps. Differentiable means you can draw a tangent line. His function had no jumps and no tangent lines. Infinitely jagged at every scale.
Charles Hermite called it “a deplorable evil.” Poincare called such functions “monsters.”
In 1931, Banach proved that in the space of all continuous functions, the differentiable ones form a set of first category – topologically negligible. The typical continuous function is nowhere differentiable. The smooth functions mathematicians had been studying for two centuries were the aberration. The monsters were normal.
2.
The Pythagoreans discovered that the diagonal of a unit square cannot be expressed as a ratio of integers. They called such numbers alogos – without ratio, irrational, literally unspeakable.
The rationals have measure zero on the real line. If you could select a number at random, the probability of choosing a rational is zero. The “speakable” numbers occupy no space at all. The unspeakable ones fill everything.
3.
Cantor proved in 1874 that the algebraic numbers – roots of polynomials with integer coefficients – are countable. The real numbers are not. The difference is the transcendental numbers: pi, e, and almost everything else. Numbers we can define with polynomial equations are a countable island in an uncountable ocean of numbers we cannot.
Most numbers have no name and no finite description. They exist but cannot be specified.
4.
Turing showed in 1936 that most real numbers are non-computable. No algorithm can output their digits. The computable numbers – every number any program has ever produced or ever will produce – are countably infinite. The non-computable numbers are uncountably infinite. We will never calculate almost anything.
5.
Before Poincare studied the three-body problem, the assumption was that deterministic systems behave predictably. He showed the three-body problem is generically chaotic. Later work confirmed: chaotic behavior is typical for nonlinear dynamical systems. The integrable, predictable systems – pendulums, harmonic oscillators, Kepler orbits – are the special cases. The textbook examples are the exceptions.
The pattern: mathematicians encounter an object that violates their expectations. They label it pathological, monstrous, evil, irrational. They treat it as an edge case. Then it turns out to be the generic case, and the objects they considered normal are revealed as a vanishingly thin subset of what exists.
The labels – “monster,” “irrational,” “pathological” – encode the namer’s expectations, not the mathematics. The word “irrational” tells you that ratios were assumed to be standard. “Monster” tells you that smoothness was assumed to be default. The names are confessions of surprise, fossilized in the vocabulary.
The mathematics doesn’t care about the names. It was always this way. The monsters were always typical. We just hadn’t looked.