DiscoGary wrote:
coachy wrote:
Just out of curiosity I did a bit of my own on my own watch the Garmin Forerunner 35. OK it's some non traditional research. Since I have done many road certifications in the past, when I run races I am ALWAYS in search of the tangents because I know I can save myself at least 30 seconds over 5K utilizing them.
Here distances from my watch on my last 10 road 5K races, all courses are certified and all on the same watch: 3.13, 3.12, 3.18, 3.16, 3.12, 3.13, 3.15, 3.13, 3.14, 3.10
At RunningLane, I measured it twice. Once the day before and the other in the community race and my two measurements were 3.17 and 3.18.
Just a little food for thought.
We haven't talked about the effect of trees and buildings on GPS. You should try to rank these results based on the amount of overhead interference each route had. That might reveal something interesting.
Oh, I did. I forgot what page. But that is called obfuscation. It can happen in broad day light too. I gave a rubber ball attached to a string theory -- an LRC original -- to explain how the signals can bounce within the triangle you are pinpointed as well.
I work in the field. I program on the device level through map tracking interfaces via mobile app and website. I know the statistic like the back of my hand - 15m/s^2 is the accepted margin of error for commercially available devices (your watches, your garmins).
Way over the Bamans and MileIdiots is the Ramer-Douglass-Peucker algorithm (1972) which takes a complex linestring (your GPS route) and simplifies it to store less datapoints.
Example:
Before RDP: linestring has 6000 vertices
After RDP: linestring has 50 vertices
These milesplit people, calling out Cory Mull and the Texas A&M chick, do not supersede opinions of science.