willcadell/gist:9e81cfcea15bc5e18202

## gistfile1.txt
Assumptions lead to inaccuracy. This is as true for devices as it is for people.

I love skiing, all kinds of skiing. I am lucky enough to live in a place where I can do quite a bit of skiing of one variety or another. One of my great pleasures is to take to the XC trails after work (in the dark) and zip around the woods with the snow gently falling around me.

I'm also a technology guy, and I've been using Map My Fitness (MMF) on various different iPhones for years. This has been a great combination for me to track my adventures, but it's not really an 'activity tracker'. It turns out that life in general is also an interesting thing to measure. This year, Santa upgraded me to a Fitbit Surge, primarily because of its GPS abilities which would allow me to track workouts that happen in an ad hoc manner (my life seems to be getting progressively more ad hoc!)

I'm a geospatial guy, so GPS is always interesting to me - distance, elevation, speed are all good  things to measure. In XC skiing these can change wildly with conditions. I was very excited to put the Surge to the test.

Now I know, XC skiing isn't a summer-time activity, and in real terms doesn't have the popularity that running or biking might have, so not having the XC setting on the Surge was forgivable. Instead I used the hiking mode, it activates the GPS, so in my mind was a reasonable surrogate. I went skiing last week and measured my Surge against MMF on my iPhone 6s.

MMF - 12.29km
Fitbit - 9.87km

Interesting! Now, I started the workouts at slightly different times, but the vast bulk of each workout was the same, there should not have been a +2km difference. So I downloaded the .tcx data and took a look. Well, a little GPS Babel and Google Earth later I could see that the Fitbit measured line was 11.7km (incidentally the MMF line was measured at 12.2km). Now that 0.5 km difference would be easier to swallow.

So, what's Fitbit doing to carve off  10% of my workout, when the actual geographic data says otherwise? The devices's GPS is evidently ok. It was something in the Surge firmware logic that determined my workout was shorter.

I went skiing again yesterday with my 7 year old daughter, this time without MMF to independently verify. Like I said, ad-hoc life...

Fitbit workout - 6.5km
Measured line - 9.41km

No wonder she was tired. In this case, I even felt the distance was wrong having skied these trails frequently. Another big difference in this workout was hills.

Why are hills important? Even if we consider that Google Earth  *might* have been measuring flat distance, and Fitbit was taking into account elevation, then the calculated distance should be longer (think through your Pythagoras...)

No, I think hills are important here because of speed and shoogling (that's a Scottish term for shaking). In fact, a XC skier typically tucks when skiing down. Thus, they go quite fast, and don't shoogle as much as someone hiking.  This would mean fewer steps recorded and a weird condition for the watch to consider. The watch would be thinking "why is this hiker travelling so fast, whilst not stepping?". I am taking a wild guess that this error would be noted as not part of the designated activity, that of hiking, and thus could not be part of the workout.

Tomorrow when I go skiing again, I will be using the Bike mode to test my theory. Interestingly, in terms of a Fitbit workout, I posit that the bike has more in common with the XC skier than the hiker (or i imagine the runner) would. The Bike mode, will care less about steps and more about GPS, and I would hope it will account for a wider speed variation.

What's the lesson here, though? I believe that the lesson is that as technologists we can be too clever. We can assume too much, when being more generic might satisfy a great deal more use-cases. I would love to see Fitbit add a XC mode, but in lieu of that, if they were to add a Generic GPS workout for those of us who do distance activities which might not be included in the list of those presently supported.

As a consumer the lesson is that I shouldn't assume that my device will understand what I'm doing, especially when I'm not actually doing what I said I was going to.

Edit: I did contact Fitbit support on this issue (no confirmation as yet)
	Assumptions lead to inaccuracy. This is as true for devices as it is for people.

	I love skiing, all kinds of skiing. I am lucky enough to live in a place where I can do quite a bit of skiing of one variety or another. One of my great pleasures is to take to the XC trails after work (in the dark) and zip around the woods with the snow gently falling around me.

	I'm also a technology guy, and I've been using Map My Fitness (MMF) on various different iPhones for years. This has been a great combination for me to track my adventures, but it's not really an 'activity tracker'. It turns out that life in general is also an interesting thing to measure. This year, Santa upgraded me to a Fitbit Surge, primarily because of its GPS abilities which would allow me to track workouts that happen in an ad hoc manner (my life seems to be getting progressively more ad hoc!)

	I'm a geospatial guy, so GPS is always interesting to me - distance, elevation, speed are all good things to measure. In XC skiing these can change wildly with conditions. I was very excited to put the Surge to the test.

	Now I know, XC skiing isn't a summer-time activity, and in real terms doesn't have the popularity that running or biking might have, so not having the XC setting on the Surge was forgivable. Instead I used the hiking mode, it activates the GPS, so in my mind was a reasonable surrogate. I went skiing last week and measured my Surge against MMF on my iPhone 6s.

	MMF - 12.29km
	Fitbit - 9.87km

	Interesting! Now, I started the workouts at slightly different times, but the vast bulk of each workout was the same, there should not have been a +2km difference. So I downloaded the .tcx data and took a look. Well, a little GPS Babel and Google Earth later I could see that the Fitbit measured line was 11.7km (incidentally the MMF line was measured at 12.2km). Now that 0.5 km difference would be easier to swallow.

	So, what's Fitbit doing to carve off 10% of my workout, when the actual geographic data says otherwise? The devices's GPS is evidently ok. It was something in the Surge firmware logic that determined my workout was shorter.

	I went skiing again yesterday with my 7 year old daughter, this time without MMF to independently verify. Like I said, ad-hoc life...

	Fitbit workout - 6.5km
	Measured line - 9.41km

	No wonder she was tired. In this case, I even felt the distance was wrong having skied these trails frequently. Another big difference in this workout was hills.

	Why are hills important? Even if we consider that Google Earth might have been measuring flat distance, and Fitbit was taking into account elevation, then the calculated distance should be longer (think through your Pythagoras...)

	No, I think hills are important here because of speed and shoogling (that's a Scottish term for shaking). In fact, a XC skier typically tucks when skiing down. Thus, they go quite fast, and don't shoogle as much as someone hiking. This would mean fewer steps recorded and a weird condition for the watch to consider. The watch would be thinking "why is this hiker travelling so fast, whilst not stepping?". I am taking a wild guess that this error would be noted as not part of the designated activity, that of hiking, and thus could not be part of the workout.

	Tomorrow when I go skiing again, I will be using the Bike mode to test my theory. Interestingly, in terms of a Fitbit workout, I posit that the bike has more in common with the XC skier than the hiker (or i imagine the runner) would. The Bike mode, will care less about steps and more about GPS, and I would hope it will account for a wider speed variation.

	What's the lesson here, though? I believe that the lesson is that as technologists we can be too clever. We can assume too much, when being more generic might satisfy a great deal more use-cases. I would love to see Fitbit add a XC mode, but in lieu of that, if they were to add a Generic GPS workout for those of us who do distance activities which might not be included in the list of those presently supported.

	As a consumer the lesson is that I shouldn't assume that my device will understand what I'm doing, especially when I'm not actually doing what I said I was going to.

	Edit: I did contact Fitbit support on this issue (no confirmation as yet)