As I understand it, the problem lies in the size of the dies. The tiny XP-E die face makes for a more pinpoint source of light which results in an accurate gathering of it’s output by the parabola for emissions in a tighter profile resulting in increased throw by default of the smaller hot spot. An XM-L die is significantly larger, the point source is spread out such that the parabola does not see a single source of light but instead a conflicting broad flat panel with emissions diverging from sources too far away from center to be accurately collimated. Hence, a much larger hot spot with less concentration of output. To answer the broader source, a much larger parabola is required to minimize the effect of the point source being so large.
Bucket is working on creating his own reflector and has experimented with vast research on the subject, in an effort to effectively focus the broad emissions from the large MT-G2 emitter with hopes of enhanced throw characteristics from an heretofore floody source. With a new process in place the expected results should surface soon, I believe he’s working on this even today. His thread” DIY Reflector”:DIY Reflector is an interesting read at the very least, and pictures showing his progress are invaluable in accessing ideas pertaining to this subject.
As in so many other discussions, there are opinions a-plenty. 