Coaching Error: Parsing bad data

This post is controversial.

When trying to formulate a plan for reaching a goal, it's natural to look at the work of those who've come before, to find the commonalities between successes. This has some drawbacks.

The 8-2-5 syndrome:

You plan 8, do 5, call it 5.

Most runners overestimate what they do, though a few dramatically underestimate. You'd think that GPS would greatly improve things, but I'm currently doing a stair workout in a ravine and the error in distance is 100%, the error in elevation (notoriously bad with GPS) 200%.

Runner's schedules when reported are often an ideal week, rarely if ever achieved and give no indication of how it was approached or if there were recovery weeks. Sometimes coaches and athletes simply lie about what they do, not wanting to give away their "secrets" or trying to convince their competition that they do much less or much more than they really do.

Runners are not independent. Just as you are looking at what others have done, so have they. Looking at marathoners, you may see that all the successful ones ran a long run every week and think that that's a necessity, overlooking that there have been exceptions. Grete Waitz set a world record in her first marathon, never having run more than 13 miles in training before that; later she changed to more "traditional" marathon training and improved by a few minutes, about what one could expect just from gained experience. The first volume of Fred Wilt's "How They Train" (long out of print, but partly covered in Noakes' "Lore of Running") showed that runners used extremely varied approaches to achieve similar results, but by the 3rd volume, they became more homogeneous. As information spreads, varietydecreases; with the internet, information spreads rapidly.

There is a common error in believing that this increased homogeneity is the result of improvements in training knowledge. New records are smaller improvements than in the past, so we must be getting ever closer to perfected training, right? Instead, I believe that it is a statistical phenomenon; there are simply more performances by more individuals, so there's a greater likelihood of more extreme performances. The marathon record keeps lowering, but the 10 mile record stays relatively constant, as there are few races at that distance, few who do them and there's little importance given them, as the races have no prize money and are not an Olympic distance.

Jason Koop's "Training Essentials for Ultrarunners" has an anecdote about coaches arguing about minutiae in an athlete's training. One pointed out that they were arguing about less than 0.5% of the total. Another pointed out that 0.5% is the difference between an Olympic gold medal and no medal at all. The implication was that they had perfected training down to 0.5%. This is preposterous. The placebo effect is 30%, so 30% of winning is the belief that you are going to win because you are the best and have trained best. That's what the coaches were doing in the anecdote - convincing each other and the athlete that they have it all worked out - the placebo effect.

Looking at a week's or a month's training is like looking at a rainbow. There's a continuous spectrum, but we instantly divide red from yellow from blue. If you ask where red ends and yellow begins, you insert orange. If you ask where orange becomes yellow, you might create yellow-orange and orange-yellow. If you look extremely closely at the spectrum, there are indeed gaps, due to atomic spectra being based on quanta. Similarly, a week has discreet workouts, but several years form a continuum.

Or should.