Whew!

I've been doing some preparation work for laying a new floor and while I'm waiting for cement to dry where the floor needed patching I've been fine tuning the tiller on the Molle'.
I've got a good full draw to the target 24", plus a tad as the brace is low still.
Damn thing is I can't see the draw weight from the video! The bow is so short I've zoomed in and the scale is off the bottom of the pic. I think it's about 45# but it's so nerve wracking flexing it I've quit while I'm ahead.
You can see I've got horn overlays on the nocks and the levers (especially the right) have been substantially slimmed. I think it will look stunning once it has a wipe of Danish oil on it

The curve on the limbs looks pretty good to me now.

Now this is where I don't get the whole Molle' argument.
Imagine I take a hint off the belly side of the levers so that they flex. The bow could be pulled so that the existing 'working limb' is in exactly the position it is in the picture. The tips would be back a bit further due to the flexing of the levers, so we'd have a longer draw and thus more energy storage surely?
When loosed the 'working limb' would recover to brace  in exactly the same time as it did before, or maybe a tad quicker as we've lost a bit of mass from the levers. So we'd have longer draw and less mass at similar poundage...
I can't see that the slightly flexing levers would somehow lag behind as they can be considered as separate short stiff limbs with a small deflection which would presumably recover before the ralatively longer working limbs which have a substantial deflection.
I dunno, a real analysis is beyond me... and that's why I'm making it. The proof of the pudding is in the eating. I suppose I could get performance figures and then reduce the tips to get 'em flexing, mind the change in performance would probably be less than the experimental error.
I'm not even entirely convinced a CAD system could be realistically programmed to give the real answer...