Out riding last night and we had a little experiment with gps and Strava. I was using a Garmin 705 and my mate had a Garmin 305. We started a climb exactly together. For the first half he was on my back wheel, at half way we switched over and I was an inch or so behind him. The last 20 yards we completed side by side and we crossed the hypothetical finish line together.(I created the segment after the ride to avoid either of us missing the 'finish line')
When I uploaded the activity to Strava I was not a little surprised to notice that it had taken me an extra 23 seconds to complete the section.
As far as I am concerned, Strava is just a bit of fun but I started to think about the results out of 'academic interest' this morning i.e. instead of doing any work!
The only thing I could think was that the sampling rates of the various devices was different and this could have resulted in the error. If this has an impact on your gps data, surely it must then impact on the veracity of the data stored and so loving pored over by Strava fanatics.
I suppose one way of resolving this would be for them to write an algorithm to homogenise the different rates of each unit.
I'm sure I read somewhere that Strava data is inherently inaccurate, would this be because of different units, sampling rate or what (apart from blatant cheating)? Any thoughts among the techies out there?
Garmin usually only records a point when it thinks there's been a chnage of direction/speed/height etc to save memory so regardless of units it's the inherent randomness of the sampling that's the issue.
Do a google search or look on the Strava site - there's loads of info/discussion.
inherently inaccurate
GPS is only accurate to around 10m in the best (stationary, good view of the sky) case. Mostly it'll be worse, and worse at the start and end of a segment too....
Some of the tools that track comparative efforts on segments showed that - you'd find that someone would appear to be still for a few seconds, then jump ahead. iPhones seem particularly prone to it.
I imagine short sections of heavily wooded twisty singletrack are about as rubbish for GPS accuracy as you can hope for!
The trouble with the "segments" is you can use pretty use any route to get from the start is to the end of it & Strava will record your time from A to B.Click on a few of the KOM times to see what route they they've taken to compare it to yours & you'll see this.There's one at lower cliff at Cannock where the bloke has just gone straight down the hill in a straight line from the start then goes along the fire road at the bottom to the where the segment finishes...
Me and a mate did a 8 hour training ride last weekend, I was recording the ride with a Garmin Edge 305 and he with a 500. When we finished the ride the results were vastly different! Ok, he was getting to the top of the climbs a little faster than me but at the end of the ride he'd done something like 25 minutes and 10k or so less riding. My unit seemed the more accurate of the two as it was closer to my cycle computer stats.
@Andrewj Was this on Strava or Garmin Connect?
I have run Bikebrain on my 'phone and the 705 simultaneously (705 on the bars, 'phone in my back pocket) and have come out with quite different results. The distance is there or thereabouts but it is the time, speed etc that are different!
@slowjo We compared on Garmin as he doesn't do Strava. By the time I'd uploaded to Strava the difference was even greater in moving time, mine had increased by 8 minutes!
Are you both using wheel sensors?
Once a garmin is calibrated to the wheel sensors it uses that in preference to the gps to calculate speed, distance etc.
It may be possible that one of the garmins took an extra 20 seconds to get a satellite fix and so created the observed time difference.
The other possibility is that one of the models lost signal for a while which I've noticed happens on my 800.
You should repeat this experiment on an open straight road with clear sky view, and start/end the segment at a definate change in direction like an intersection.
I'd be interested to hear the result.
No wheel sensors on mine or my mates bikes.
FWIW, I've used Endomondo (on my HTC Desire) and a Garmin Edge 705 with calibrated wheel sensor on some 30k-ish road routes. The figures reported are typically within a couple of hundred metres of each other. Calorie estimates aren't much different either, though the Edge stops counting calories when I'm descending 🙂