Posted: 20/11/2014 at 12:24
Banksy -- imagine that each time the GPS records a point, it is randomly out by a few metres. Now go for a run in a straight line. Your true track is straight; the GPS track jigs from side to side a little due to the random error in each point. And the total length of that slightly jigging line is longer. That's why I reckon they tend to overestimate a little, in general. I've also noticed my Garmin assuming a slightly wider track round my park loop corners when I'm going faster, which could be an algorithmic thing.
Then sometimes, you get a complete freak-out, usually adding a lot of distance artificially. Can be due to high buidings etc, but sometimes for no apparent reason at all (maybe a lorry driver with a GPS jammer going past or something). Sometimes that wil result in an underestimate, e.g. if it has lost satellite lock for a while and assumes a straight line from the last known position.
Overall I think GPS watches are brilliant, as nothing else can give such good accuracy (usually within 1 or 2%) as one takes any old route you fancy. But they're not perfect, and occasionally very wrong.
As for official course measurements: obviously they're more accurate, but I remember reading something once about them ensuring that the course is long enough rather than spot on? And I've had the same problem with mile markers (including the VLM balloon arches) being a bit out for the convenience of where they can place them. (For VLM at least, more accurate marks are on the road, but not always easy to spot.)
I use the tip I learnt here of having a Garmin field of "Time -- average lap", and pressing the lap button manually at each mile marker. That field then gives you an accurate average min/M so far that doesn't depend on GPS. Have to press the button twice if you realise you've missed one though!
10M steady for me this morning. Once again my weekend race has left my legs too creaky to do any quality work in the week, though I suppose it counted as speed training in itself.