Aren’t hexagon bins functionallyy equivalent to triangular bins (subdivide each hexagon in 6 equilateral triangles)?
And then to square grids (look at what the hexagons subdivided in 6 triangles create:
- this is just like a square grid, skewed to 60° instead of 90°, but further subdivided in two parts along a single diagonal direction
- or it looks like a stack of rectangular bins (rows) that are interleaved with their cells shifted midway if you further split the rectangular cells in two parts, you get again a regular rectangular grid.
In other words, these grids have the same use. I just think that hexagonal shapes needless obscures more the data, and a basic square grid is simpler to compute: the only relevant difference is the length of cell sides, i.e. the rendered resolution (in surface per cell: all those grids are equivalent at only a constant scale factor, so you can get the same visual impact using one grid pattern or another, just by changing the cell size).
I bet that square grids are just simpler to compute and other grids (regular hexagons, regugal triangles, interleaving rectangles) don’t offer any advantage in terms of readability. square grids also offer more regular aspect and avoid blurings effects or color artefacts along cell borders, which pollute the global visual impact.