watches have their time and place in training. For example, these watches are great when you want/have to change up routes, but still want to make sure to get the intended amounts of miles in (always having to "pre-map" your run gets tiresome). And, it's nice to be able to get an idea of pace before you hit the first mile marker (and even better when there are no mile markers at all)... plus, being able to download your data post-run provides valuable training information. However, I don't foresee myself ever wearing my watch inside (even though I could get a footpod, I think I'll save my money and believe the treadmill) or for track workouts. I guess I'm "old school"... I'm used to figuring out pace using my cheap Target watch... and, some of the time, my calculations are more accurate than my GPS-wearing partners.
As it turns out, GPS watches are not 100% accurate all of the time. I know many of us have observed some of the phenomena discussed in this article (aka so how long did your watch say this run was? or hmmm... we suddenly dropped a min/mile going uphill?). So where do the discrepancies arise? Anything that disrupts satellite reception (like trees, buildings, and even clouds) will degrade accuracy. The watch will not be able to properly track distance (which is complicated by routes with a lot of twists and turns), and since it calculates pace by dividing distance by run time, this will be off too. If the watch temporarily loses you, it tries to extrapolate your pace, which can get interesting (like the 40mph situation that Mr. Henton observed). For a fairly comprehensive review of GPS watch accuracy (this guy writes awesome, well-documented reviews too if you're in the market for a watch), check out this two-part blog entry (part 1, part 2).