I was telling one of my athletes the other day that I have seen countless examples of people applying science to training, only to see it fail.
Instead, I prefer to anecdotally see what has worked with large numbers of runners, apply those same principles, and then see if I can figure out the science behind why its working.
In other words, don't do the training based on the science, do the training based on experience and then back out the science.
For example... doubles. I've rarely seen someone start to do doubles and get worse. So do doubles. Maybe it inhibits the formation of scar tissue? Maybe the second run "flushes out" free radicals? Maybe its just easier to get more mileage in that way, and the more mileage, the better? Does the science really matter, though, if it works?
The long run... I have seen more runners improve with an 18-30 mile long run (when talking about marathon training) then not improve. OK... Let's make sure you get that long run in. Now, I can look up the science... glycogen depletion, capillary density, increase in mitochondria, yada yada yada... but does it matter? As Malmo has pointed out (on the doubles), 100s (or 1000s) of runners have proven it works so what else do you need?
Another example. I have found through my own coaching successes and failures, as well as talking to several very respected coaches, that while training for a marathon, you need next to no V02 Max work. That its actually detrimental. Instead, a lot of 1/2 MP - MP work and easy running is the best formula to success. OK... let's experiment with that and then figure out the science.
Good coaches know the science, but don't coach based on it alone. They coach based on experience and trial and error, and then they hope the science backs them up. They also quickly find that different people respond to different things... especially based on their own individual talents and histories. This can completely turn the science part on its head.
Just some thoughts.