A friend posted a screenshot of a fitness app I had not yet seen. Indulging my shiny object fascination of the moment, running and fitness apps, I had a look at the capabilities of the app. The app appears to be a good fit to what he is trying to do.
Some veteran runners like to keep running logs, track their running over the years. I ran for years without intentionally logging my runs. In 1998 Runner’s World sent a free running log, and I began to use the log to record fitness activities including running. I began to have data I wanted to retain. Data that has to be stored and conserved.
Some thirty-five years ago I had an Atari 800 computer. The simple programs I wrote had to be stored on a audio cassette tape. During that same era the 5.25 inch floppy disk provided removable storage for the Apple II computer. Within a few short years the 3.25 inch floppy disk would supersede the 5.25 inch floppy disk. This past week the last computer with a 3.25 inch floppy disk drive was retired from the division in which I work. I still have data on 3.25 inch floppy disks, but that data will never again be accessible.
In 1996 I created a web site in CompuServe. In 1998 the FSM lost the dial-up connectivity to CompuServe and switched on direct connectivity to the Internet. I had to recover the orphaned data from a hard drive mirroring my on line site, moving the data to GeoCities. Within three years I was again moving that data to college servers.
Over the years I have learned that with changes in technology come orphaned, abandoned, and sometimes unrecoverable data. Today data is often stored in cloud services, which is almost more vulnerable to sudden change than hardware based technology. A storage provider may suddenly decide to depart the storage market (Ubuntu One) or start charging fees for what was once free storage (Amazon Cloud Drive).
If the data is in a proprietary format, the loss of a software package may make the data unrecoverable.
When I first acquired a GPS enabled Android smart phone in late 2014, I looked for an app that could provide time, distance, and mapping capabilities for my running. Thirty years of rescuing and moving data is always on my mind when I make a data storage choice. I always want to know how a company plans to stay in business. If there is no clear way for the company to benefit from me or my data, then that is not a secure data storage choice. And the company should be big enough to fend off being acquired and taken apart by a competitor.
I opted for Google MyTracks reasoning that Google might just be big enough to not fail anytime soon. Google thrives on user data, MyTracks provides more information about me (where I am and where I tend to go) that has potential marketing value.
MyTracks provided only basic total time, distance, speed, and mapping capability.
During the summer of 2015 I became aware of a running app option that was functional in Micronesia, an option I had not seen in 2014. Although two of the market leaders, Runtastic and Runkeeper, are not available in the Google Play app store in Micronesia, Nike+ running was available. Runtastic and Runkeeper have a two tiered business model. The lower tier appears to be a lite version for free, possibly with advertising, and a paid subscription model with more features. Nike+ running is free – all features -and apparently an outgrowth of their abandoned fitness FuelBand.
Nike+ running provided data on splits, personal bests, and nice touches such as tracking total kilometers per shoe. All of the data a runner might want to access. Yet what of tomorrow? Nike has a business model that may be more sustainable than a small operation supporting a dedicated running app. Nike as a company certainly has the size and resources to survive. Yet the app is peripheral to their core business and could be shed like the FuelBand product. Whither one’s running data then? Where Google MyTracks could upload KMZ files to Google Drive, Nike+ has no clear path to data export nor extraction.
A dedicated app housed in a small independent software developer is equally insecure over the long haul. A company the size of a Nike or Google could buy the app or acquire the development team to enhance their own fitness apps should they choose to do so. Data may remain trapped in a proprietary format on cloud servers, inaccessible when the app no longer exists. A veritable 3.25 inch floppy disk of the modern era: without the drive the data is effectively lost. Without the app, the data generated by the app is lost.
Paper is not a particularly secure data storage option, but keeping a log in a notebook remains a viable option. I continue to echo run information into locally available composition notebook. That is the only way I have of knowing where my running data will be tomorrow. Where is your data sleeping tonight and will it be accessible tomorrow?
Having seen friends using the Nike+ running app here on Pohnpei, I thought I would give Nike+ running a spin. This blog pulls together a series of observations I have made over the past couple weeks in other places.
I have been using Google MyTracks since October 2014. MyTracks is not specifically a running app, more a general purpose track and trace route tool. Nike+ reports total time, calories, average pace, NikeFuel earned, splits, and maps the route one has run if outdoors.
I’m using Nike+ running 1.7.2 on an LG Optimus II P715 Android 4.1.2 Jelly Bean rig. Note that the temperature reported in the screen above only appears after synchronization. I suspect that the temperature is the result of an online lookup on the server side, not a reading from the telephone.
The app also tracks total distance logged on each shoe, a statistic for which I have never had any information. I really do not know how many kilometers I rack up in a shoe before the shoe delaminates, and I always run a shoe until that delamination point. Way beyond loss of cushioning in the EVA foam.
Nike+ has running specific features including reporting of splits, a capability MyTracks does not have. Determining road splits in the past usually meant hauling a Garmin eTrex, a piece of paper, and a pen, to record the splits. Having them automatically available feels luxurious. I have used splits in my statistics class, data the students can use to calculate 95% confidence intervals and other metrics. This is also the first time I have known splits for joggling. When I run, I usually juggle three tennis balls. I cannot carry a GPS and juggle, so splits while joggling were completely unknown.
The LG P715 has an accelerometer, which the Nike+ running app can use when indoors on a treadmill to estimate distance based on steps. The app apparently uses a generic step distance to calculate distance. The uncalibrated app credited me with 1.83 km (1.137 miles) for 1.00 miles displayed on the treadmill (14% long).
I ran the treadmill mile at 5.4 mph. Note the mile measure is based on the treadmill as seen below, not the app. The app is set in kilometers, the default for Nike+ running is in miles.
1830 meters by the app for 1806 steps on the treadmill suggests the app was using 1.01 meters per step, or about a meter per step. If the treadmill is correct, then my actual stride on the treadmill was 0.89 meters. This difference is also roughly 14%.
On the fourth of August I attempted a calibration run with the Nike+ running app. In retrospect I suspect I should have done this with the settings set to miles not kilometers. I used a Garmin GPS to run 1.61 km on the track. I do not now recall the app distance, but the app was short at about 1.53 some kilometers. I used edit run to enter the correct distance, but on closing that screen the app reported a different value than I typed in. Eventually I found I had to enter 2.0 km for the app to report that I had run 1.61 km, as seen in the screen shot. I suspect some sort of faulty conversion process in the edit run module of the app.
The Nike+ running app feeds data to the nikeplus web site, providing another way to pull up data and statistics.
The full site includes an animation of the most recent run route.
A satellite view option is also available on the desktop version of the web site.
One feature of a running app is typically the awarding of badges and trophies, along with tracking personal bests. The awards for NikeFuel levels are based on Nike’s proprietary NikeFuel score. Nike wanted to be able to combine all possible athletic activities into a single metric. Working with Glen Gaesser, an exercise and wellness professor at Arizona State University, Nike generated a correlation between wrist acceleration and oxygen uptake. This was then turned into a dimensionless value referred to as NikeFuel. The original work was aimed at a wrist based accelerometer, but the cell phone based apps also generate NikeFuel scores. The Nike+ Fuelband is claimed to be a valid and reliable estimate of physical activity energy expenditure.
Whether the FuelBand is still available is unclear to this author. I suspect the $149 retail price, the same price as an Android phone that can run apps that also generate NikeFuel, was probably beyond what the market would bear for a dedicated fitness device. NikeFuel is an attempt to “gamify” fitness, find a way to generate a video game like score that one can then tout to one’s friends, but the “black box” nature of NikeFuel makes this hard to do. Runners talk in terms of distances run, times, pace, these are the “smack talk” of running. I have trouble seeing how an effectively dimensionless black box of a number will compete with existing physical metrics.
Sure, I cracked off 18 NikeFuel points per minute for thirty minutes three times in one week. But what does that actually mean? Even my children laughed at me when I tried to brag about being supersonic. Without a clear link to the physical world, NikeFuel is ultimately doomed to irrelevance.
Error and uncertainty are central to statistics and physical science. One cannot measure anything twice and get the same result.
I ran on the road with two GPS equipped devices on the same day as the calibration run above. A Garmin eTrex Venture HC held in my hand produced 8.18 kilometers. Nike+ said 8.94 km for my cell phone mounted on the same arm. Now I suppose that my upper arm could have traveled 760 meters further than my hand, but as both are still attached to me, I doubt it. Seven runs to the airport gate and back 2009-2010 average 7.898 km using a Garmin eTrex, the dock gate is a tad farther. Out to the dock and back has averaged 8.42 km on prior runs with my phone, which is reasonable in light of the 2009-2010 data. 8.94 seems a tad long.
Middle numbers are safer choices in measurements, thus I suspect the Garmin was short and the LG P715 was long. The trace from that day suggests that there was an error in measurement.
As noted in the caption above, the distance from two to three kilometers is longer than from six to seven kilometers. As I was running the same route out and back, that is a significant difference and an indication of some form of error. The Deketik gate out-and-back run is somewhere out around the 8.42 kilometer average.
By default I leave my GPS on, even when running inside. On the ninth of August I set Nike+ running to indoor. The next day I switched back to outdoor. At some point in this process the app must have switched off my always on GPS, as I discovered it was off after I came in from my run on the tenth. So switching from outdoor to indoor may inactivate the GPS, but switching back to outdoor does not reactivate the GPS.
The Nike+ running app can apparently disable the GPS but not re-enable the GPS. This makes sense as the GPS adds load to the battery – unloading the battery is less protected than loading the battery. Google MyTracks used to ask permission to re-enable the GPS. Noting that I did run a calibration of the Nike+ running app, the distance is in concurrence with other distance numbers for that route. Seven runs with a Garmin unit in 2009-2010 averaged 7.923 km on that route. Google MyTracks reported 8.47 km on 03 November 2014. This fits a pattern where the LG P715 is generating longer distances that the Garmin eTrex units. Lesson learned: manually re-enable the GPS in the Nike+ running app after running inside.
Note that the Nike+ running app cannot automatically post to FaceBook for those of us in the 691 telephone area code. The issues is probably the security and verification systems in FaceBook. FaceBook cannot send an account verification code to a 691 number, and thus the Nike+ app is blocked from automatically posting (no auto-post to an unverified account). That also means a 691 runner cannot post to FaceBook while running and receive cheers in return.
A runner in the United States with autoposting to FaceBook activated. If liked or commented on during the run, the app cheers for the runner.
There are some other caveats for the Nike+ runner in the Federated States of Micronesia. The app appears to award badges and count weeks of activity based on the time zone of Nike headquarters in Oregon. Thus I do not get a weekly based badge until Oregon gets over into the new week. I also earn Night Owl run badges for running between midnight and 0300 hours. I do not run at those times here in GMT +11, but the time over in Oregon at GMT -8 is the small hours of the night.
Meanwhile my children will laugh at me even harder when they learn that I have exceeded supersonic.
First light is an astronomical reference to the first light to fall on a telescopic lens or mirror. The article is an early days overview of the app, first use by the author.
Post-script: NikeFuel on the Nike+ running app might not be a completely black box. There are high correlations to the distance run (r = 0.9995, n = 22) and the duration of a run ((r = 0.9907, n = 22). Pace is not well correlated (r = 0.4724, n = 22), what correlation does exist can be explained by the distance and duration.
Distance is in miles, pace in minutes per mile, and duration is in minutes. NikeFuel values are on the x-axis. NikeFuel for the running app may be based solely on distance as duration is dependent on distance for a runner at a fairly constant pace. Note that some runs were done on treadmills. All but one treadmill run was post-calibration.
Post-script, three months of use, update on trophies only