[l3th4l]

think of it this way:

the whole scene is rendered on a coordinate system (x,y,z) relative to the camera position with z being the depth. now imagine there are only a limited amount of positions for each axis (depending on how many bits are used to store them (2^n)).
whenever two objects (polygons) occupy the same space you get that flickering effect since the engine doesn't know what to render in front and what behind.
increasing the clipping distange effectively cuts part of the scene in the forground and trades the z values used there to be used in the rest of the scene which increases the distance at which objects start to flicker because more z space becomes available for it. it's always a tradeoff so you always loose some stuff in the foreground when you try to increase the rendering accuracy in the background...