GPS Accuracy and the Automation Paradox

It's been a busy summer for UK's MRTs. Not a week has gone by without someone getting lost in our hills, without yet another call to learn how to use a map and compass and not to rely on phone apps. This in turn elicits other comments that the problem is not in the use of digital tools per se, but in not being able to navigate. True as this is, the calls for learning traditional navigation should not be dismissed as Luddite, for not being able to navigate competently and the use of digital technologies are intrinsically linked.

GPS Accuracy

Before getting onto the bigger problem, the question of GPS accuracy is perhaps worth digressing into. Our perception of what the GPS in our phone can do for us is skewed by our urban experience. We use mapping applications daily to locate street addresses, and we have got used to how accurate these things are in that context.

However, many of us do not appreciate that because GPS does not work well at all in cities, mobile phones use so called Assisted GPS. With A-GPS the accurate location is derived from the known positions of mobile phone masts and the presence of domestic wifi IDs, which street mapping vehicles collect and store in massive databases. And, obviously, A-GPS only works in cities and with a working Internet connection (which is why your phone will complain when you use the GPS while in the air-plane mode).

So how accurate is GPS alone?

First, there is the accuracy of the GPS service per se. This is the simple part: the US Government undertakes to operate the service in such a manner that a user in the worst location relative to the current position of the satellites can achieve grid accuracy of ±17m and altitude accuracy of ±37m in 95% of cases. You can often get better results, but need to allow for even bigger error 5% of the time.

Then there is how well the device on the ground can access and process that service. The above numbers assume a clear view of the sky down to 5° above horizon, allowing for the acquisition of 6 different satellite signals. They assume no weather interference, and a good quality receiver that makes full use of all the available information.

In the hills the real conditions often are nowhere near optimal, and the tiny GPS devices in watches and phones, with their tiny aerials, are not of the requisite standard. The real errors will be, possibly significantly, bigger (e.g., I have seen an error of some 200m on an iPhone 5 on one occasion in the upper Glen Nevis).

So how good are these numbers from the perspective of mountain navigation?

The altitude resolution error is potentially around a major contour line difference and deteriorates rapidly as the number of visible satellites drops. As such GPS estimated altitude is not much use for accurate navigation. (Altitude is very useful for mountain navigation, and a much better resolution is achievable with a barometric altimeter, when used correctly.)

But if you compare the location accuracy to what a moderately competent navigator in moderately challenging circumstances will be able to estimate without the GPS, the GPS wins hands down; this is what it's designed for. Nevertheless, the GPS based location cannot be assumed to be pinpoint accurate. In complex terrain the errors can be navigationally significant, and are not good enough to keep me safe -- there are many locations in the mountains where if I overshoot my target by 30m I will die.

This is, of course, no different than following a compass bearing. Neither the compass nor the GPS are magic bullets that will keep me safe. But with GPS we seem to be conditioned to trust the technology more than it merits. Competent navigation comes down not to the tools, but to making sound judgements based on the information provided by the tools, whether it's map, compass, or GPS. And that brings me to the Automation Paradox.

The Paradox of Automation

The Automation paradox can be formulated in different ways, but it comes down to this:

Automation leads to degradation of operator skills, while, at the same time, the skills required to handle automation failures are frequently considerably higher than average.

In an industrial field, the introduction of automation largely replaces a workforce of skilled craftsmen/women with a low skilled one. This is unavoidable; the craft skills come from practice of the craft. The automation of the process does away with the practice, and doing so removes the opportunities for practising the skills.

But the bigger problem with automation is this: when automated processes fail and require manual intervention, they tend to do so in atypical, complex, corner cases which require higher level of skill to handle, skill that the workforce does not have. In industrial fields this leads to the development of a small number of exceptionally skilled (and highly paid) experts who get called in when the automatic process fails.

Navigating by GPS is subject to the Automation Paradox; it takes away the grind of reading maps, taking bearings, pacing and timing distances. This is great while it goes well, for it leaves more time to enjoy the great outdoors, and so we do. But in doing so it deprives us of the opportunities to develop the rudimentary navigation skills.

But when it fails, there is every chance it will not be on a nice sunny day with cracking visibility. It will be when the weather is awful enough to interfere with the radio waves, or in a location where no satellites are visible. The competent navigator will simply turn the unit off and carry on, for she has other tools in the bag and knows how to use them. The rest will have to call in the experts to get them off the hill (assuming their phone has a signal).

And this is the problem with the GPS. It's not that it's not a useful tool, it is, in the hands of a competent navigator. But that competency is developed through deliberate and ongoing practice of the basics. What it does not come from is the following of a GPS track downloaded from somewhere on the Internet, and let's not delude ourselves, that's how the GPS gets generally used.

P.S. If you live in the Central Belt and don't know where to start, I offer Basic Mountain Navigation and Night-Time Navigation courses.