Armchair Musher: Data are great, but mushing is magic

Wednesday, February 6, 2019

2-6-2019  7:11 PM

Well the front runners are already making preparations to leave, 36 hours goes by faster than you can imagine when you are in Dawson for the Quest. But based on the many photos and updates on kennels’ web pages it appears in general spirits are good, and lots of quality dog care was happening. While looking over the information on Trackleader for the first half of the race I once again called on my smartest techie friend Melinda Shore to talk about what I was looking at, and it morphed into a discussion about the value and limitations of data.

Which lead me to a great idea, let’s make the armchair a loveseat for a while and invite Melinda to be a guest commentator to share her own well informed perspective. I am so happy she agreed. In a near future post I will be using some of the data she put together to talk about how people ran the first half and what it might mean for the second. But before I do I would love for you to read what she has to say about the value and limits of just looking at data.

I cannot thank Melinda enough as she is a wealth of information, a constant help, a very active supporter of mushing, and has a wicked sense of humor to boot! 

Data are great, but mushing is magic

As Jodi has mentioned several times, one of the things that’s changed a lot over the past decade and a half or so is that distance dog mushing has moved online, with checkpoint arrivals and departures being posted on websites and GPS trackers being put on sleds. That means that we’ve now got access to a massive amount of data about how races are unfolding and how individual teams are doing, down to being able to look closely at things like run/rest schedules.

In the meantime, there’s been a data revolution in the outside world as well. With online advertising companies and large social media sites now being able to collect vast troves of data about individual behavior, they’re also developing the tools to analyze those data and identify the characteristics that, for example, lead some people to choose one brand of cereal and other people to choose a different one. The results are often subtle and surprising and astonishingly successful.

So, for some of us who combine, er, slightly nerdy tendencies with a profound love for the sport, there’s a natural tendency to look at race data and wonder if there’s something in there that can tell us why some teams do well and some teams don’t, and to consider the possibility of applying some of the data analysis techniques being used in other fields to the piles of data we’re accumulating in mushing with the running of every race.

A bunch of years ago an interviewer asked a musher - Jason Barron, if I remember correctly - about the trail for Iditarod that year, and he laughed and said that that particular trail is going to favor somebody, and a different trail would favor someone else. That’s absolutely true, and while we can have some general idea about who’s going to do well in a cold race, or a race with too much or not enough snow, we aren’t going to know how someone’s training has gone that year, or if they’re in a rebuilding year as their older dogs retire and younger ones start to get some experience, or if they’ve had a happy or sad winter and if they’re stressed or relaxed. We certainly don’t know how to collect that information systematically and encode it for analysis.

And there are the added complications of time and weather. There’s a saying that “you never fish the same river twice,” and not only is it true that you never run the same trail twice, it’s also the case that two people on what’s the same trail on the map are often running trails where conditions have changed in-between, and they’re effectively on different trails. For example, at race starts and on particularly gnarly sections of trail, the teams going through first may see really nice, pristine conditions and the teams towards the back of the pack may have to negotiate troughs that have been dug down the middle by mushers standing on their brake mats. Similarly, if someone’s out front breaking trail in soft conditions, the teams behind them are likely to have an easier time of it and their dogs won’t tire as quickly. Plus, weather can blow in fast up here, and someone can have calm air and blue sky on top of Rosebud while a team four hours behind them may run into a blizzard. Overflow forms and then freezes up. Things change so quickly!

Then there’s training. People train for the race they want to run and the schedule they think will work for their particular dogs. Someone who’s trained in a particular way is likely to run into trouble if they run someone else’s schedules, and we’ve seen that happen a few times — for example, a few years back someone in a different mid-distance race said they were trying to run Allen Moore’s winning schedule on the Copper Basin, and their team sat down and quit several times. We’ve seen teams try to pull off long runs they weren’t prepared for, and lose a lot of ground. If you see someone running a particular schedule and doing well, it doesn’t mean that someone else can do well on that same schedule.

So, right now I don’t think we’re able to identify all the pieces of data that would be needed to come to credible conclusions about a “formula” for winning 1000-mile dogsled races, and while you should never say “never” I think that the things that make the sport so magical and that contribute to good or bad performances are things that are going to resist more rigorous quantitative analysis. For now, I think it’s enough to be able to use the tracker and the checkpoint times to identify who’s booking a lot of rest, who’s struggling in the hills, who’s getting faster as time passes, who’s changed up their schedule from past races, and so on. The rest of it - the bonds between mushers and dogs, the amazing athleticism of our best buddies, the beauty of the trail and the vastness of the terrain - those are the things, I think, that matter the most and the things that can tell us the real stories of how a race is unfolding and what it takes to be successful out there.

And a reminder to all; there are some great videos about using the Trackers to get race data at the Mushing Tech Facebook page.

Author: 
Melinda Shore