Although I understand your point, the particular example you pick is somewhat easy to fix. Using image recognition it would be fairly straightforward to check if there is a park and what amount of that park could be viewed (maybe using geolocation to check for more images of the park to compare)