That's all fine and dandy about real world efficiency versus specifications for "max", but doesn't the same thing apply to the XHP50? So, comparing specs for efficiency is not legit? The XHP50 spec is lower efficiency than the XM-L2.
As
@victorzamora noted, the efficiency might be different at different output levels. So, does the XHP50 actually have better efficiency than the XM-L2 when it's putting out 1000 lumens? At 2000 lumens? If so, then sure, you would be able to build a light that puts out the same lumens for the same or longer burn time. Or more lumens for the same burn time.
Do either of you know? I do recognize that that could TOTALLY be the case. I'm just curious if there is actual data available to support this idea, or if we are just speculating.