The author is describing a randomized search method, and not an evolutionary method. Stochastic optimization like this includes evolutionary algorithms but also other methods. <a href="https://en.wikipedia.org/wiki/Stochastic_optimization" rel="nofollow noreferrer">https://en.wikipedia.org/wiki/Stochastic_optimization</a><p>"Evolution is the change in the heritable characteristics of biological populations over successive generations", says <a href="https://en.wikipedia.org/wiki/Evolution" rel="nofollow noreferrer">https://en.wikipedia.org/wiki/Evolution</a> . The key part is <i>populations</i>.<p>That's why <a href="https://en.wikipedia.org/wiki/Evolutionary_computation" rel="nofollow noreferrer">https://en.wikipedia.org/wiki/Evolutionary_computation</a> also described evolutionary computation as "a family of population-based trial and error problem solvers with a metaheuristic or stochastic optimization character".<p>Hacking away at a single code base is not working with a population of code bases, and certainly does not match the description at <a href="https://en.wikipedia.org/wiki/Evolutionary_programming" rel="nofollow noreferrer">https://en.wikipedia.org/wiki/Evolutionary_programming</a> .<p>Even those hacking away at a single code base are not making random changes, and we only have second-hand evidence that the interviewee "started mutating random parts of his code, and rerunning, just to see if that somehow fixed things".<p>It's a junior programmer trying to build up an understanding of what the system is doing, but without the experience to know to go and re-examine the test input for validity.