In reading a description on Usenet of a NIST competition for selecting a standard cipher, I read:
Consider that the best currently known methods for factoring use randomization: Construct enough cases at random, and eventually you can paste them together into a set of factors. This notion of an algorithm is very new - at best, 40-50 years old, though in terms of actual practice, perhaps no more than 25 years old. Mathematics in the past has dealt with proofs, which may be constructive or non-constructive. Constructive proofs have historically been given as deterministic algorithms.
Emphasis mine. I know for a fact that the concept is much older than that, as there are numerous algorithms named after people from classical antiquity. This makes me wonder how old the concept of the algorithm really is. Wikipedia merely says that it has existed for centuries, and cites the Sieve of Eratosthenes as an early example. Here I define algorithm as an unambiguous series of steps and procedures to solve a specific problem, particularly a mathematical problem. Note that I am not asking about the terminology, but about the earliest known instance of a well-defined algorithm.