I was recently thinking about this... We've been building houses and other structures using plum lines and water levels all the time before afordable optics came in play. This kinda means most of our buildings are actualy polar rather than cartesian. Surely enough given the size of earth the error is quite tiny. But it's funny thinking about how the room i am sitting in right now is shaped like frustum with spherical floor and ceiling, rather than block. Despite what architecture drawing says...
If the floor and ceiling (and walls) were leveled and flattened and brought to plumb with a straightedge scraped in with the three-plate method [1] (Popularized by Whitworth in the 1830s, but the ancients made straight edges and flat plates too), then they were actually not 90 degrees at the corners!
[1] https://ericweinhoffer.com/blog/2017/7/30/the-whitworth-thre...
There are very long and narrow wave pools used for research and testing and they are long enough that the surface of the water curves measurably vs extending perfectly straight lines from the center out.
Long bridges, like the Verrazano Narrows in New York City, have plans that account for Earth being a sphere. The towers at either end are not parallel, but tilted apart so that each is aligned with its local gravity.
i imagine imperfections in construction dominate this effect
If you have two buildings 4km apart (about the length of Central Park), that’s about 1/10,000 of an earth circumference so 0.036° change in ‘up’. If the buildings are 300m tall, 300*sin(0.036°) = 0.188m
That’s less than those buildings are probably expected to sway in a strong wind, but probably outside the tolerances for modern construction so theoretically measurable as an average deviation.