I got paid minimum wage to solve an impossible problem

4 months ago

The author, a Computer Science student, attempted to optimize the task of sweeping a supermarket floor by turning it into a grid graph and using simulated annealing to find the shortest path.
Initial optimization focused solely on minimizing distance, resulting in a path with impractical sharp turns, making it unusable for a human.
The author realized the mistake was optimizing for the wrong metric (distance) and adjusted the algorithm to include a 'turn penalty', resulting in smoother, more practical paths.
The experiment highlighted a broader issue: algorithms often optimize for easily measurable metrics (like engagement or profit) that don't align with real-world needs (like happiness or sustainability).
Key lesson: Technical correctness is meaningless if you're solving the wrong problem. The real challenge is defining what to optimize for in the first place.

Hasty Briefsbeta