Conclusion, its possible for the human body to have a VDOT score of 100 (90 + 10), which corresponds to a marathon performance just a tad north of "1 hour and 45 minutes" (over 18 minutes faster than the current WR
This is interesting, partly because of the shock of the number, but partly because it represents a much different route to breaking a record than is usually assumed. Specifically, this is a statistical
argument, asserting that if we wait long enough, we'll eventually see such an extreme outlier that the modern records will be shattered. On the other hand, it implicitly suggests that 2:05 will always be an elite time because most of the elite runners of the future would not be extreme outliers any more than they are today. (And given the current state of modern sport, would such an extreme outlier ever escape the suspicions of PEDs?)
Looking at the progression of world records (in practically any sport) over the past century shows a different perspective: once a record is broken, the entire sport tends to follow along, often due to advances in training and equipment. We now have hundreds of full-time marathon runners, so presumably training volume has basically been maximized. Shoe technology might not be completely optimized, but it's probably now affecting marathon times at the level of seconds rather than minutes. The two biggest things that I can come up with to knock the world record down a bit more are:
(1) Running surface: changing from the asphalt of urban streets to a rubber track surface could be worth a minute or two. Would some race ever be willing to foot the bill to lay down a couple hundred thousand of square feet of track in the pursuit of a WR?
(2) Drafting: the 2011 WRs featured large pacing groups out to 30K. At these running speeds, air resistance is a significant limitation on running economy. Future records are going to need a lot of cooperation.
So: let's start with an athletic outlier, not at the 1:45 level, but just fast enough to be the top dog in the sport under current conditions (say, with a low 2:03 WR). Then find a benevolent billionaire who pays for a ten foot wide rubber track along the Berlin course and pays the top 10 marathoners a million dollars each to run a 2:00 pace past 30K (with the outlier tucked in behind), perhaps with the help of some outlandish plexiglass windscreen on the pace car. On a perfect day, this could work.