I think different levels of detail are hard to do (lots of modeling work, not that many 3D artists are active at the moment I believe). And LOD works well when objects are both close (need high detail) and far away (can be low detail), as you might have in a first person view. The top down view of a typical RTS game like 0 A.D. has to have most of the things on-screen at the same level of detail because those are equally distant from the viewpoint. It would only provide a benefit once people zoom in or zoom out quite far. So I don't know how strongly this game would benefit from LOD changes, although that also depends on the amount of mesh complexity reduction that could be achieved. Keeping the extra meshes in memory (ready to be swapped in and out) also comes with some cost in memory. Instancing would probably have a larger impact as the average player will have many times the same building on-screen (such as houses, towers and walls), the same goes for the environment (trees are a good example). I'm no expert so I wouldn't be able to guess how much benefit may be obtained and whether it's worth spending development time on that (and drop support for any non-OpenGL 3.1 compatible systems).