I didn't read all this thread in detail but all I really have is a couple of anecdotes (LOL - that's especially for Jim) and thoughts. Tried to post yesterday but the forum died on me.
With regard to cleat placement "regualr" vs "mid-foot/arch":
For predominantly steady state cycling, there isn't much difference in performance.
When you need to have rapid accelerations, it is a disadvantage to have the cleats further back.
There are other considerations, including a requirement to change bike set up including the lowering of saddle height and adjusting the fore/aft saddle position and the resulting bar height/reach adjustments.
Due to the substantial change required it may or may not be possible to achieve such changes on a given rider's existing frame, or require additional cost for new components, or even perhaps make UCI legal saddle placement more difficult, especially if you are already very close the the UCI limit.
Also, it can result in significant toe overlap of the front wheel, which may or may not concern some riders.
I coach a rider who set world masters hour record who has/does train and race using both "normal" and mid foot cleat placement. His power output for such efforts is the same using either (after a reasonable adaptation period).
There is a shoe manufacturer, Biomac shoes, (Goetz Heine) that's been producing and promoting this for many years, they even have a patent on it. I met him maybe 6 years ago when he was demonstrating the SRM Torque Analysis tools in Steve Hogg's shop while a rider was using the shoes/cleat placement.
Their shoes are very stiff and light and allow for arch cleat placement. They are also very expensive.:
http://www.bikeradar.com/gear/category/clothing/shoes/product/review-biomac-bio-mxc2-shoes-09-34720/
With SPD cleats you can even fit two cleats on the shoe, although it's not all that practical since clicking out of one and into the other just makes your bike position "wrong".
Of course he proposes lots of benefits, but was unable to provide substantive evidence on the efficacy of this solution, relying on pedal force analysis to claim a more even pedaling output around the pedal stroke (by examining the raw torque data from an SRM equipped with torque analysis hardware), and improved "efficiency" (sic) which we know doesn't really mean much in terms of performance, and secondly, on celebrity "endorsement".
Sounds a pretty familiar tale really.
Not sure he was too impressed being asked some "difficult" questions, rather than have someone just take such assumptions/claims on face value.
My other anecdote is personal. As you know, I am now an now amputee rider after trans tibial amputation in 2007. On my right leg, I have regular cycling shoe and cleat placement (by Steve Hogg incidentally) and on my left, the cleat is placed effectively under where my heel would be if I had one, since it is connected to my prosthetic via a straight pylon.
My steady state power output (W/kg) for durations from 4-minutes and longer is the same (if not better) post-amputation than it was pre-amputation. My sprint power (peak and 5-sec power) has suffered a drop of ~ 250W (and that results in a drop in longer range average sprint power as well, but primarily because the initial high power isn't there. My fatigue rate isn't much different than before).
From a cycling performance POV, I think this is a solution looking for a problem that likely doesn't exist, however I am certainly open to inspecting additional real evidence should any be published.