This is interesting, but sounds consistent with what some others have done (e.g., Farah at NOP). Note that this isn't a reduced mileage program, but a shift of some work up to a higher pace. Farah has claimed that a similar shift had been key in his recent breakthroughs. The question, of course, is if and how we can generalize this information. The data points are small, and as with all training, there are very heavy individual responses to training. Magness addresses this in his book, noting that people who are relatively fast-twitch for their event seem to benefit more from larger disparities in pace (i.e., slower recovery and long runs and slower paces and more space between intervals), while people on the other end of the FT/ST spectrum (relative to their event) benefit from less disparity and more LT paces. It seems possible (although I haven't seen any empirical data on this) that someone like Hall, who already has a massive aerobic base (and appears to be on the ST side of the spectrum for the marathon), can further adapt and improve by shifting up his recovery paces and/or shifting some of that mileage to longer tempos. This might not work as well (or be counterproductive) for a FT runner or a runner who has not yet established his/her aerobic house.