What did we learn from a Scientific standpoint?
Kipchoge’s performance in the sub 2 hour marathon was amazing, far exceeding mine and many others expectations. While a lot has been said about the performance and how it translates to a legitimate marathon, what I want to explore is my initial reactions to what we learned from a scientific standpoint.
With so many interventions thrown in it’s impossible to delineate what mattered the most, but a brief analysis on what we can learn from this science experiment is warranted.
Performance is really hard
I’ll start with the cynicism or reality before going into the positive. Three men entered the race, and two didn’t make it to halfway on pace. But Tadese rana marathon PR, you might protest. Sure, but for a 58:@3 half marathon performer with his fueling dialed in for the first time in his marathoning career, 2:06+ is underachieving compared to his prowess in the half. It’s a solid performance overall, but not a hit it out of the park success. Desisa coming home in 2:14 was a rough day. So what you’re left with is one Success, one average performance, and one “failure”.
In such a situation, it’s tempting to focus on Kipchoge and to attribute all of the interventions as a success, but to me, that screams of a bit of survivorship bias. We have the best marathoner of all time, a man many considered capable of running 2:02 mid in a legitimate marathon, who thrived. Obviously, something got Kipchoge the other 90sec-2ish minutes.
But from a scientific standpoint, Desisa and Tadese provide interesting data. Did the tapering formula to get them perfect on race day not help them? Did the extra fuel not work? Did the shoes not benefit them to the same degree? Did any of these interventions hurt someone like Desisa?
We don’t know the answers to those questions. But it’s important to keep in mind, as we analyze the opposite. We can’t proclaim that the new tapering formula aided Kipchoge anymore than we can say it hampered Desisa. Did the fact that they took one of the world’s best marathoners (Desisa) and couldn’t get him to the starting line in position to even run half marathon near his best, mean that science failed?
No, it just shows that performance prediction and optimization are really freaking hard!
The Scientific Process of Evaluation
It’s tempting to write of Desisa and Tadese’s failure as a “well, we knew that these guys weren’t as good.” But that would be a mistake.The Nike group considered 60 elite runners, then whittling it down to 18. These eighteen men had VO2max, Running Economy, and other lab based values measured to determine who had the best chance. From these 18, using physiological data and race performances, they selected the 3 men we know. And in the end, only only one had a shot. We don’t know who was in their pool of 60, but what I think this shows is how far away physiological testing is from predicting performance.
During the broadcast, we heard the hype of the testing of these runners. But what we can state is that, selection based on physiological data doesn’t seem to help anymore than simply looking at race performance.
Case in point, Tadese has the best physiology on the planet, yet his marathon exploits have been sub par given a guy with his physiology and a 58:23 half marathon best. His VO2max and running economy values are historic. Yet, why hasn’t he ran faster than 2:10 on a legit course and 2:06 on the race course? Of course, with better pacing, he might have run sub 2:05 during the sub 2 attempt, but still. In idealized conditions, with a lot of extra help, you’d expect a man who has the best lab tests in history plus the best half marathon in history to run quicker.
I’m not disparaging Tadese, but it’s an interesting conundrum. Up until now, the blame was put on his fueling strategy. Yet, that was corrected and optimized, and he still made it to only about halfway before fading. Obviously the fueling helped, but it’s not the sole factor holding him back. Despite his physiology and half performances, what are the other factors we can’t measure that hold Tadese back?
Overall, I think the lesson on selection and measuring physiological data is clear. Despite what researchers might profess, lab data gets us only within 2-5 minutes of predicting someones time on average. You can do just as well by looking up an athletes times and taking a guess. We need to push farther as scientist, and attempt to understand and elucidate performance beyond the traditional VO2max, Running Economy, and Lactate Threshold variables.
Shoes: Gimick or aid?
The marketing deluge that we experienced during the attempt focused on the shoes. How much energy do they provide, how much time does it save?
No one knows. Are we to say that they helped Kipchoge, but hampered Desisa? That seems unlikely. But what we can tell from the slew of runners who have worn them in competition, that a variable effect seems likely.
We know that a slew of runners have used the same shoes or similar prototypes in marathons. Rupp, Hasay, Cragg, Bekele, Flanagan, the 3 men in this attempt and more. What we’ve seen is a few who have overperformed with the shoes and many who have performed within prior expectations. For instance, do we think Rupp or Bekele would have run minutes slower at Boston or London without the shoes? I certainly don’t.
On the other hand, Kipchoge exceeded expectations and Jordan Hasay, once she slipped on the shoes, ran performances that far exceeded anything she’s done in her life. Did the shoes benefit Hasay to a much higher degree, or was there another factor?
Our best conclusion at this point is that the shoes seem to have a variable effect, perhaps giving some runners no advantage, others a half-percent, and a few outliers a large effect. Nike’s own testing has suggested variation between 2-7% changes in economy, showing that the shoes help some more than others.
The question is why.
Drafting and the Psychology
Drafting and pacing seems to matter. A whole lot.
Up until now, even with rabbits, pacing has been erratic during every marathon attempt. In this one, with the help of lasers and a car, the 5k splits varied by only a few seconds. This remarkable consistency undoubtedly played a role.
In addition, the drafting formation and car seemed to provide a substantial assistance. The energetic savings of drafting runners has been explored elsewhere. The effects of the car being that close have not.It should be noted that elite cyclists and researchers have commented that the car would provide a slipstream effect, even if the wind was still.
One last thing to consider is the psychological benefit of having pacers with you the entire race. Noel Brick and his group has done work on the psychological benefits of pacing. What has been hypothesized is that with rabbits, the decision on pacing is taken away. We literally get to “shut off our mind” and engage only in following the rabbits. By taking away this mental process, we free our minds up to focus on the task at hand.
Having pacers the whole way means that we no longer have to engage our minds on both the effort and pace once the rabbit drops, like in a traditional marathon.
Between the car and the flying V of runners, it seems to have made a significant effect.
Fatigue’s Ugly Head
The more interesting cases are of Desisa and Tadese to me Why? Because we got to see fatigue in an extreme circumstance. Seldom do runners go for broke and try to hang on. In Desisa, you got to see an example of extreme fatigue and Tadese a more moderated version of it.
It’s interesting to me that these two guys, despite all the gadgets, pacing, drafting, and so forth, either didn’t make it to halfway or barely did.
What we saw was a great demonstration of how fatigue occurs when you press as long as you can, ignoring signals to slow. Once they break, it’s a gradual or drastic slow down, depending on how far they delved into the depths of their fatigue. Desisa didn’t run out of glycogen at 10 miles obviously. His body sensed he was over the red line for finishing a marathon at the current effort, and despite resisting that information for as long as possible, his body and mind shut him down gradually.
Fatigue is a complex thing. It’s not only physical but also psychological and emotional. In this attempt, the runners stayed on a pretty set pace, attempting to run at 2hr marathon pace for as long as they can. What this means, is that the runners have to ignore the internal cues and sensations that are informing them that they need to slow down in order to make it the entire 26.2. In a natural race, the brain and body might head this warning early, and we’ll see an ever so slowing runner as fatigue ramps up.
In situations where we override and tune out this feedback, we carry on for as long as we can, all the while, our body amps up the signaling. Our effort increases, sensations of pain go up, and we might have an increase of emotional freakout moments where doubts shoot into our mind. The longer we ignore our bodies prediction on what we are able to do, the louder the signals get. Our body is trying to protect itself. What happens if we simply ignore and ignore and ignore? Well, the last 1km of Joshue Cheptegei’s race provides a glimpse. Perhaps spurred on by the home crowd, Cheptegei ran far past the limits his body set, and in the end, left it no choice but to essentially shut him down. Not even the thrill of winning before an entire home nation could propel him forward, past fatigue. He shut down, staggering across the finish line.
When you race, the pace naturally ebbs and flows, even in a time trial situation. In the sub 2 hour marathon attempt, it was a trial of staying on a pace for as long as possible, with little variation. This may be efficient, but it’s also a recipe for hanging for as long as possible and then blowing up. Which, more or less is what happened to Tadese and Desisa. The longer one pushes into the depths of fatigue without heading the warning signals, the bigger the blow up occurs.
And that’s what we saw. Pushing past the signals that normally inform us to slow, until it was too late for those two. On the other hand, Kipchoge was able to ride that line, staying just on the right side of those signals.
In the end, we like to think we understand fatigue, but the reality is we have many theories on fatigue, but not one agreed upon understanding. Which is amazing if we think about it. We’ve got all this research, and yet, we still have what I’d describe as a rudimentary understanding of why people tire and slow down during a race.
Lastly, the researchers seemed to have only minorly adjusted the training. One point of emphasis by thr group was the use of a tapering formula by Dr. Phil Skiba, a man whose work I admire. For Kipchoge, he suggested that his coach adjust the taper by 2 days. We’re celebrate his performance as an indicator that it worked!
While I am interested in and applaud modeling of tapering, how do we know it worked? Did it fail with Desisa, who certainly looked like he might have been overtrained or needed a bigger taper? Did it not work for Tadese?
Or put another way, how do we know that it aided Kipchoge? Here’s a man who has a coach who has expertly made Kipchoge show up on race day every single marathon he has run!
Look at his track record, 2:05:30 in his debut for a win, 2:04 wins at Chicago, Berlin, and London, a 2:03 win in London, Olympic champion, and on and on. His slowest marathon that’s he’s lost in was the 2:04:05 debut in Berlin!
If that isn’t expert peaking, I don’t know what is. Coaches dream of having their athletes show up like that every single marathon. Given the nature of the marathon, it’s almost unbelievable he hasn’t had a bad day.
So, did the 2 day adjustment help his peaking? Well, I’d err on the side of history and say that whatever Kipchoge and his coach do, they show up on race day without a formula!
What did we learn from a scientific standpoint?
Did science provide the answer to running sub 2? It’s tempting to paint the picture of a triumph of science. But, I think that would be a mistake. We have the worlds greatest marathoner, a man, who demonstrated incredible physical and psychological gifts and determination. The drafting and pace setting seemed to allow him to get the most out of himself. The shoes probably helped, but who knows how much.
Did the fueling, training analysis, and so forth impact him? Perhaps, but where was the help on the other two? Did the algorithm for training miss that Desisa was overtrained or needed a few extra days to taper?
We don’t know. Take what worked, adjust, but don’t sell the house, just quite yet.
If the initial calculations on 90sec-110 seconds off the car/pacing/drafting are correct, and we assume Kipchoge would have run right around 2:02:00 without them, then how far outside of Kipchoge’s abilities is that? As I said, most of us felt Kipchoge was good for 2:02 mid in a legitimate marathon, perhaps faster.
So, if the pacing data is correct, did the shoes, fueling, algorithims, etc. contribute to another 10sec, 20sec, or even 0 seconds? We don’t know. It’s possible that all of those extra gadgets, while scientifically hyped to a large degree on the broadcast, mattered little to none.
And that’s why I think this project should not be a celebration of our scientific understanding, but instead a push forward In sports science. We need to understand how to predict performance better. We need a clearer understanding of how fatigue occurs. We need to know how these ideas translate to performance out on the track or road.
Kipchoge’s performance is to be celebrated and analyzed. What makes him different? It’s not the physiology that we measured, as it appears to be worse than Tadese’s. What’s the performance factor, that has shown up in races for over a decade, that doesn’t appear in the lab? Perhaps psychology will provide some answers.
In the end, let’s not be fooled by the survivorship bias, and instead push forward in our knowledge and understanding. It’s the same reason why as a coach, it’s really easy to convince myself I know what I’m doing after a great race, but more often than not, I need to remind myself of how little I know, and how lucky I am. Performance is difficult. There is always a human component, even if we track, measure, and “optimize” everything