Swarm intelligence in a race

Or unintelligence? Not sure.

A friend recently posted on Facebook a video of starling murmurations over a river in Ireland. In these murmurations, thousands of starlings fly in uncanny coordination, relying on their neighbors for information to avoid collisions; basically, if your neighbor moves, so do you. What's so interesting is that this patterned movement is best described by equations of "critical transitions." (Who knew The Birds of Hitchcock were just going through phase transitions!) "Critical" in the case of starlings means "always ready to optimally respond to an external perturbation, such as predator attack." And, survival seems to be the main reason for these beautiful murmurations, as shown by one 2011 study in which scientists followed a peregrine falcon chasing a murmuration. "Most of nature's beauty has to do with one of two things: sex and being eaten alive." Another study (2013) showed that the benefits of this kind of "swarming" behavior to animal prey (i.e., causing predator confusion) is sufficient for the behavior's evolution. There is also the related, decades-long work of Thomas Seeley (Cornell University) on swarm intelligence of honey bee colonies. A basic finding of Seeley is that, even though an individual bee in a swarm has limited information and intelligence, the swarm as a whole is far smarter (crowdsourcing!).

All this got me wondering--naturally--about how swarm movement and swarm intelligence might be connected to running. (There is always a connection, right? After all, running is the perfect metaphor for life--or, is it the other way around?) An imperfect connection might be to a group of hunters practicing persistence hunting. The latter has been hypothesized, in the running man or endurance running theory of human evolution, as being critical to the survival of early humans. A closer connection might be to a field of runners (the "swarm") in a race. Throughout a race, but particularly at (1) the start, where runners are discarding clothing, adjusting paces, smiling at cameras, and generally sorting themselves out and (2) each water/food stop, where there is--if one were to actually analyze it--quite a complicated set of decisions being made by runners and volunteers in their various interactions (runner-runner, runner-volunteer, and volunteer-volunteer), individual runners are making decisions--sometimes smart, sometimes not--and acting on them, based on limited local information (e.g., other runners nearby, an upcoming turn). But, what about the swarm of runners as a whole? Is it smarter than any individual runner? I'm not sure. Given situations like the course congestion in the 2013 Marine Corps Marathon (e.g., in the new out-and-back section along Rock Creek Parkway), maybe swarm unintelligence is more the reality! Still, a better understanding of running swarms in races could provide better guidance for a runner in negotiating the start of a race, the water/food stops, and other crowded sections.

Go slow to run fast

... but, what is "slow"?

Ever since my pace fell apart in the second half of the Wineglass Marathon, I've been trying to narrow down the range of paces that I can currently sustain over 26 miles. Plugging my PR half marathon time of 1:46:34 into one of those race times predictors, I get the following chart of predicted vs. actual race times, which are quite accurate for the shorter races. But, the predicted marathon time is clearly an outlier (predicted time of 3:42:11 vs. my actual PR time of 4:22:06). But, what does that mean? Is it an outlier because these predictors just are not accurate at longer distances? Or, is it an outlier because my training just has not yet gotten me to my current potential? I definitely like the latter explanation better! Runners like Ed Whitlock, who was the first person over 70 to run a sub-3:00 marathon, show what my theoretical potential is, especially given that I still have quite a few more years to go before joining that age group. :)

Closely related to what pace I can currently sustain over the course of a marathon (and how far this pace is from that needed for my BQ) is what pace at which I should currently be training, especially for the weekly long runs, so as to be able to sustain my BQ pace for the upcoming March marathon. This topic has been, naturally, well discussed within the RunYourBQ online community--on pacing in general and on the benefits of slow paces for base runs. Likewise, Runner's World has many articles on this topic, e.g., "Easy does it."

Here's the multi-race pace chart I'd previously shown, in my MCM race report. The main revelation for me from that race was that an even pace at the fastest speed I can sustain for a particular set of race conditions should yield the best results.

My "modern" PR (4:22) was run at the March 2013 Rock 'n' Roll USA, where I was able to sustain a ~9:30 pace through around Mile 21. For MCM, I ran at around a 9:45-10:00 pace, and I was able to sustain that pretty much the entire way, with an overall slightly negative split. But for the rather crowded MCM course, especially through the early miles, I probably would have had a new PR. Given these two race results, no wonder my Wineglass effort, running the first half at a few seconds faster than my BQ goal marathon pace (GMP), turned out to be a mini disaster! My training for Wineglass just wasn't enough for sustaining that kind of pace for the entire race.

So, one of the main lessons I took away from Wineglass is that I need to do more miles in my long runs at or faster than my GMP. In training for Wineglass, I did most of the long runs at about 60-90 sec. slower than GMP, which was what I'd considered to be my LSD (long slow distance) pace. In hindsight, I think that pace was too slow. In training for the March marathon, I've modified my long run paces as follows: Front half warm-up miles (or half + 1 if the run is of odd miles) at GMP plus ~30 sec; rest of the miles at GMP, except for the last two miles, which are at GMP minus 15-30 sec. So far, at the mid-point of the current training cycle, with long runs at 18 miles, I've been able to run at these modified paces. I think just as important, though, if not more so, is that I'm gaining confidence in being able to shift into a faster gear towards the latter part of a run (or race)--and, to sustain it.

I don't disagree with "go slow to run fast." There's much research behind it. But, in some cases, maybe one needs to go fast to run fast!

Review of NB Minimus 10 - Updated

I had posted a review of the New Balance Minimus 10 last July (2013), in which I'd noted that it was generally a good shoe but had this particular stitching design that created a hot spot, particularly at the top outside of my left foot. That problem, I just want to update, went away after some weeks of running. As I'd indicated at the time, for me, the NBM is no longer a minimal shoe (as advertised), and I do definitely favor the latter kind (e.g., Vibram). But, as I'd also indicated, the NBM would be good for really long runs and races and for trail running. Since the hot spot problem resolved itself, I've run one marathon (MCM in October 2013) and numerous long runs in the NBM. One thing I didn't note in my earlier post, because I'd not yet worn the shoe long enough, is that the sole material wears out fairly quickly. This seems like something on which New Balance could improve. I'm already having to use Shoe Goo to keep up with the wear.

Overall, though, I would still consider the NBM in the future.