Exactly this!
They should say more clearly up front what is the definition of efficiency.
If they want people to think of a particular metaphor, but there are critical differences, those differences must be clarified up front. It seems the person designing the game/test has never really mowed actual lawns (at least lawns s/he actually cared about).
I took the approach of efficiency meaning minimizing turns to save time.
Instead their definition of efficiency is fewest moves. Their "optimum" has many more turns than many other paths, so would be far more inefficient use of time (and apparently nevermind that path also lacks the necessary overlap to do a good job of actually mowing a lawn).
It was also unclear up front that the squares with rocks/features could not be mowed into, but this is unrealistic, as you need to mow around the feature edges not merely the square nearest them.
Overall, it is a less-than-good choice of metaphor for the test, and with the poor explanation of what "efficient" means by his measure, it is not the slightest bit surprising many got different results, "sub-optimal" by the page's standard.
It seems the test itself is so sub-optimal so the results are meaningless — ench test-taker's goals were likely significantly different from the goals the test-maker assumed.