Flutterby™! : Ground Truth 2006-12-02 16:52:05.104327+00

Ground Truth

2006-12-02 16:52:05.104327+00 by petronius 9 comments

The Times of London has what is becoming a more common story: A British ambulance crew transferring a patient to a hospital 8 miles away set their GPS unit and proceeded to follow its instructions to Manchester, some 200 miles away. Fortunatly the patient was not in extremis and arrived in good order, 5 hours late. The GPS database is being reviewed, as are the brains of the drivers.

The story mentions some similar cases in the UK, and I heard of one recently where an elderly German on the Autobahn followed the voice mode on his GPS right into a sandpile, more than a mile down a very clearly marked "Closed for Construction" offramp. I realize that we expect the mighty computer to give us accuracy, but at what poiont does common sense kick in? And would these folks have been so trusting in written instructions? I notice that in my area Mapquest has a habit of putting you on the local expressways if they are within a mile even on short trips. I ignore such idiocies, but only because I know the territory.

I think that one of the difference between those of us who know computers and the 99% of the rest of the world is that there's this notioin that computers, and this probably extends out to generally trusted information sources in general, like newspapers, are infallible.

It's kind of like when we (well, I do this, I don't know about the rest of y'all, but I'm assuming...) catch errors at cash registers and such: I'm trying to constantly fact check the world, keep a running tally of what things should be, do sanity checks. Most people are really surprised when you say "there's no way four turkeys can be $13. Can you check the receipt?", they just assume that the computer got it right.

Geospatial Data Accuracy is a major, major issue in my world. Entire conferences are help to discuss how to manage, document, and convey error to the map (re: GPS) user.

One big conclusion is that every GPS navigation devise should have a big, huge disclaimer stuck on it. The GPS may be easy to use but it's far from infalable. One of the easiest examples is the two-way street becoming one-way. If the database in the GPS hasn't been updated, then you could be going the wrong way down the street!

Further, people don't understand how inaccurate GPS can be. At best, it's about 15 feet off. that's the width of a road! Throw in a little cloud cover, some ground reflection, maybe a tree or two and you could be 1/4 mile away...

About 25 years ago, our neighborhood changed all the streets so you jog over a block, rather than going straight through (we have a chunk of 'hood that's got two through-streets two blocks apart, and people were "cutting through" the residential streets at 40mph). There's no way to go straight through anymore.

Seems one of the GIS vendors had "corrected" their database. Yahoo! Maps, Google Maps, and Mapquest still show this incorrectly at lower zooms (they all share NAVTEQ data). It's only at the highest-available zoom level that you can see the chicanes.

Also of interest, try a auto-based GPS in the Santa Cruz Mountains sometime. Like on Ice Cream Grade. Due to the heavy tree cover and cliff walls making for funny reflections convinced the GPS I was either driving in the Pacific Ocean or similarly far off the road. "Please return to the highlighted route!" It also got a little confused about which way I should turn on this curve.

And Dan, there's an account of a century in the SCM that might interest you that I stumbled onto by googling for "ice cream grade santa cruz mountains california".

Oooh, that looks like a cool ride. I love the area around Felton, although I might have to get me a compact crank for those climbs.

The neo-luddite in me says that this is just another symptom of excessive use of technology, but it's a pervasive and longstanding one. Just as CAD users thought that it was a good idea to specify dimensions to the ten thousandth because the software they used defaulted to that, approaching mapping issues, even if that's just following directions, without at least a basic understanding of the flaws of the technology is itching for darwination.

I'm also reminded of my grandfather telling me that my uncle used his GPS to hit the mouth of a creek while coming in from the surf in his dory, in the fog. From what I know about GPS it's both reasonable that it worked most of the time, repeatabililty during a resonably small time period is actually good, but the idea of trying to hit a 20 or 30 foot wide creek mouth coming in with surf off the Atlantic with 10 foot or less visibility and a device that under the best conditions is good for roughly +/- 10 meters seems bloody insane to me.

People forget a little "personal responsibility" in so many things regarding technology.

GPS's are incredible tools. When some people are handed a hammer, they bust a finger open playing with it. Some people build a house. - Same for GPS's.

Yeah, I like that part of the SCM. Helps that I have a friend in Boulder Creek to visit when I'm out there. Although next trip I might need to zip up your way, since I haven't much explored north of the bay.

As for the grades, the nice thing is that for just about every one of the steep climbs, there's an alternate shallower (and longer) route you could take if you don't have the equipment for the climb. But they'd be pretty darned spiffy on the downhills!

The problem with biking in the SCM are the minivan commandoes who head out for an afternoon of wine- tasting and then start bombing along the single-lane mountain roads faster than their tires can handle (and driving well beyond what they can see around the hairpins). I nearly got driven off the road by one when I was out there in my Camaro in '96, and every time I drive out there I see fresh gaps in the trees where others weren't as lucky.

Dave, sounds like some of those downhills are "overheat your rims and risk blowouts" steep and curvy. But if you bring your trike out here I'll be happy to put some asphalt under us with ya in the north bay.

And on the risks of bad map data: James Kim has been found dead. It's unclear which map the family misread, but it's pretty clear that information presentation and mapping was what got them into that situation, even though it was a judgement call[1] that lead to his death.

[1] I can't say that I would have made that call any differently other than to say that every time I've heard of such a thing, leaving the car was a mistake, but after 9 days it's a coin flip.