This paper addresses the need for empirical research on the processes by which students create algorithms. I analyse the collaborative work of three high-school students on a contextualised graph theory task, in which they created an algorithm for maximising the happiness score of a seating arrangement. The group found an optimal arrangement but created an algorithm that did not fully account for this arrangement. The group's written algorithm reflected only the properties of their optimal arrangement that they explicitly noticed after creating the arrangement. And, these explicitly-noticed properties aligned with the group's predominant contextual considerations.