A computer science student turned a part-time supermarket job into an algorithmic experiment. Faced with simply sweeping the Albert Heijn floor, they instead modeled the store as a grid graph, built a visual editor, and wrote a C++ path optimizer using simulated annealing. The goal was to solve the Traveling Salesman Problem for an everyday task, applying complex heuristics to find the most efficient route to cover every tile.

The initial results were technically perfect but practically insane. The algorithm produced a path that was brutally short on distance but filled with sharp, dizzying turns. This happened because the system optimized for one variable: total path length. It completely ignored human factors like momentum, physical comfort, or the simple desire not to look like a malfunctioning robot. The solution worked flawlessly for the wrong question.

Realizing the flaw, the student added a 'turn penalty' to the cost function. This forced the algorithm to balance distance against the physical cost of turning, producing smoother, more walkable routes. This same principle explains why social media algorithms drive outrage and why AI models hallucinate. They optimize for engagement or confident-sounding text, not for truth, happiness, or reality. The lesson is universal: technical perfection is worthless if you solve the wrong problem.