Our network location implementation’s 3D trilateration is using far too much CPU so we’re going to temporarily downgrade back to 2D until we heavily optimize it. We’ll make an efficient Rust implementation to replace the initial Kotlin code and we’ll see how fast we can make it.

3D is useful to take advantage of Apple’s network location data having altitudes for a lot of networks. It helps a lot with estimating a position around buildings with more than a couple floors. Upgrading it to 3D helped a lot with some downtown areas but it’s much too slow now.