I personally feel like I wouldn't be able to write software while on a treadmill as I already find it hard to do sometimes while standing up at my desk.
If one doesn't have that specific problem, should one use a treadmill desk? Dunno. He presents several graphs of daily mileage, is that evidence for or against usage of treadmill desks? Dunno. The calorie count is small (compared to what?) when he calculates out, does that mean treadmill desks are not helpful? Dunno.
More relevant would be stuff like: how many emails did he send each day correlated against mileage? How many words did he write in his novel, or how many words did he edit? Did he at least lose some weight? Was his mood better on treadmill days? Did he consider running a simple randomized test (just flip a coin for each day whether to use a treadmill) for anything? Heck, what was his subjective experience, even, about the effects of treadmill - did he feel unable, as many people do (not just you), to do anything requiring real concentration?
OP reminds me a lot of Stephen Wolfram's big fancy QS post about metrics like typing: he had measured an impressive number of things, but nothing that ever mattered.
I find a solution in splitting my time and do complex coding while sitting grunt coding while on the treadmill. The switching is distracting. I don't get as much done this way comparing to sitting all day, but overall I think the health benefit is worth it.
There is also the possibility my pace is too fast, though.