CONCEPT
The Probabilistic Method
Paul Erdős’s audacious technique for proving that mathematical objects with desired properties exist by showing a randomly chosen object would have them with positive probability—the moment randomness became a tool for reaching certainty, and the conceptual ancestor of machine learning.
The probabilistic method, introduced by
Paul Erdős in a single landmark paper in 1947, inverts the standard picture of what a mathematical proof is. Ordinarily, to prove that a mathematical object with some property exists, you exhibit one: you construct it, point to it, here it is. The probabilistic method refuses to try. Instead, it shows that if you choose an object at random from some collection, the probability that it has the desired property is greater than zero—and therefore at least one such object must exist, even though the argument hands you nothing you can point to. Existence is established by chance, not by craft; the proof is non-constructive. In the canonical 1947 example, Erdős applied this to Ramsey numbers, showing that colorings avoiding certain monochromatic structures exist for graphs far larger than anyone could construct by hand, simply by computing the expected number of forbidden patterns under a random coloring and showing it