I've got 18 months of Moves data logged against corresponding tracks from my 910XT and the difference in consistency, accuracy, etc., is telling - the iPhone is frequently off by 10s (if not 100s) of metres and especially coming round the Millennium Dome is a joke - it has me jumping from the south to the north of the Thames every few hundred metres.

Moves iPhone 5S - http://imgur.com/kjYf8Tj Garmin 910XT - http://imgur.com/jnXuaQC

I bet your iphone is trying to use cell tower locations or wifi hotspots it's picking up around there and borking up the location. The area around there is kind of open and flat so I wouldn't be surprised if it's picking up location information from across the river.

It does seem to be WiFi related. But if I have to remember to disable the WiFi whenever I want to do some GPS tracking - and if I forget, I basically get random garbage - that pushes it well into "no bloody use" for me.

Surely there is an app for that? Android has plenty of apps that will automatically turn on/off your wifi based on various criteria, such as your location (I always turn it off as soon as I'm 50 yards away from home or from the office).

That's a feature, not a bug ;)

Uh, no, I'm pretty sure that's a bug. Unless you literally want to consider inaccurate tracking a feature.

I would bet it is the cycling of GPS that the app is doing. If GPS is just left on, a battery can get burned through in a couple hours (for one of the newer iPhones, much less an older one). Those wacky readings are some of the initial results after GPS is cycled back on.

One might look for a higher power setting if you are doing an hour / half hour run and not concerned about keeping battery life up.

Moves is on "battery saving" mode (which actually makes little difference - I get bugger all life out of my phone anyway) but that doesn't really explain why it's just around the O2 it has serious issues - that's only about 2-3km out of the whole 30km walk.

Probably put these apps should compensate for the fact that humans can't run that fast nor that they can run over water.

Do you use a GPS app in your phone often? This almost looks like it was falling back to AGPS / Skyhook while the almanac was still downloading, which can take half an hour or so.

Once the almanac is downloaded, the phone's GPS track should look much closer to the Garmin's.

If this was a recurring problem then I don't know what to say, other than it doesn't match my experience with RunKeeper on an iPhone 4S.

Well, I have Find My iPhone and 5 other motion trackers (Moves, Argus, Breeze, Human, Nike+) running 24/7 - it's a fair bet my GPS almanac is up to date since they should all be poking the M7 for updates.

Did you disable WLAN on the iPhone during the run?

That's what RunKeeper recommends, because the AGPS using WLAN-SSIDs lowers the accuracy of the tracking.

How long does the Garmin take to find you at the start of a run? I admittedly had an old model, but it was basically a dealbreaker with the forerunner. I was wandering around for 5-10 mins with my wrist in the air waiting to start.

I use the 910XT in a very cloudy part of the world (which I'm assuming makes things worse). I've just integrated the startup time into my prep.

Before I put my running stuff on, I turn on the watch and let it figure stuff out while I'm taping my nips, and if it's winter, gloves, tights, mentally bracing myself for the cold, etc.

Same here, I've got a Garmin watch and it's great once it finds satellites, but routinely takes well over 5 minutes to do so (in NYC). I've started using my iphone to track runs and bike rides instead, only using the Garmin for really long bike rides where I don't mind the wait and having my current speed on my wrist is worth the wait.

I used to have to wait for a fix as well. What works for me is that when I'm preparing for the run, I put the watch in a window for enough time to get a fix, which sometimes take a long time. Then when I go out to run later, it gets a fix pretty quickly.

The new models lock really quickly. I have a Garmin 620 and it locks within 30 seconds. Cloudy days might take a minute, but basically by the time I finish my warmup walk, it's locked on.

It's a lot better than the 310XT. I don't think it's taken more than a minute of standing still in the last few months. I do tend to start in pretty open sky'd places though.

I have found this with the phone, if I do a cold start. When I ride somewhere, start the tracking, the stop the tracking when I get to the run track, the iPhone is so much better at tracking me.

(Totally offtopic but running near the Whitehouse in DC is funny because they screw with GPS accuracy so you bounce all over the place)

How is that possible? GPS satellites are sending the same signal out to everyone, I don't see how they could possibly target a certain area.

The Economist[1] had a story about the London Stock Exchange getting jammed GPS-signal 10 minutes every day. I believe they are using something similar.

[1]: http://www.economist.com/news/international/21582288-satelli...

Gps satellites all send slightly incorrect information... A separate radio signal correction is transmitted to correct it. That way, during wartime (or in high value target areas) the US can basically mess up enemy gps enabled weapons by broadcasting misleading satellite and correction data.

That doesn't answer the question of how they could make GPs inaccurate only in a specific area.

Low power GPS jammer at the Whitehouse, perhaps?

A terrestrial transmitter spoofs or overpowers the GPS signal.

GPS satellites have a signal strength on par with a night light's, so it's easy to overpower in a small area using a low power transmission.

I don't know how they do it, but if you drive by the Pentagon with a GPS unit you'll see it go crazy.

On some you can load AGPS data onto it for the next few days, but the bigger issue is that Garmin no longer uses SiRF chipsets, which are vastly superior when compared gen-for-gen vs. the generic stuff Garmin's using now.

It's a shame, but the most accurate Forerunners are the older ones–same for Foretrex, etc.–which are often the ones with the shortest battery life.