Would it be possibly to implement such a system, or is it tooooooo much to ask for? Mind you, i really DO understand the extensive updating to the core functions and multiple corrections to the main code routines.
Here's my reasoning behind Hexagonal shapes for planet tiles;
-- For every 9 squares, you'd get 7 hexas (even could fit the same amount inside a group of four 'contiguous').
-- Most images of improvements could easily fit those also but with small overlaps in rare cases.
-- In particular, the overall terrain(s) capacity (or potential for efficient developments) would also benefit from a new framework since, generally speaking, the raw formats correspond exactly to what is beneath the virtual grid.
-- On any similar standard size and/or complete planet surfaces, the optimized grid may receive as much as 16(+/-4 to 8) new tiles just by stripping off outbounds of neighbor squares so that they can squeeze right inside a ratio of 30/40% (by calculating the geometrics of slightly smaller Hexas than the bigger boxes) more than their actual "wasted" areas.
-- A quality boost, variable effects, the feeling that MORE potential is already available, and so on.
Although the above is a suggestion... only.
- Zyxpsilon.