The Pudding Asked 30,954 People to Mow a Virtual Lawn, Revealing That Humans Are Surprisingly Adept at Solving a Classic Computer Science

The experiment found that 52% of participants came within five moves of the optimal path, and 16% completed it perfectly. Despite 14,589 different approaches, the median person achieved 91% efficiency. The task, formally known as Coverage Path Planning, is related to the Traveling Salesman Problem.

Researchers noted that humans perform impressively close to computer-calculated optimal solutions, even with 70+ stops, where they are just 11% less efficient. The study highlights how people use heuristics—shortcuts for "good enough" solutions—when perfect paths become computationally impossible. For instance, checking every route for just 10 cities requires testing 3.6 million possibilities, and 20 cities balloons to 2.4 quintillion.

A detailed walkthrough of an average player's run showed that a critical early decision—whether to go down or continue right—often determined success. Players who completed the left section first faced an inevitable backtrack due to a dead end, leading to less efficient paths. The findings suggest that human intuition for path planning is remarkably strong, even as complexity grows.