Paradise lost: SatNav "destroys" local knowledge | Hacker News Reader

Paradise lost: SatNav "destroys" local knowledge

news.bbc.co.uk

I submit this for a few reasons.

The first is that when I use SatNav I arrive at my destination having no real idea where I am. I remember in a science fiction book reading a statement like "We claim to have explored space, but in truth, we have explored the small bubbles that surround individual planets. Of the great depths of true space we have no knowledge." I feel a little like that now when driving longer distances with the SatNav.

Secondly, it's interesting how often the SatNav produces "obviously" sub-optimal routes. The Travelling Salesman Problem is well-known to be NP-Complete, so it's not surprising, but some of the foibles are incomprehensible. It's no wonder some people distrust them, and quite surprising that others seem to have complete faith in them.

Finally, stop for a moment to consider the miracle that is the modern SatNav. Despite its short-comings it contains a device to tell you where you are to within a fathom, maps that know where you are and where you're going, enough power to compute a sensible (if not always optimal!) route between the two, a suer interface that non-technical people can usually drive, and all for the cost of a few hours work.

Addendum: GPS works by computing varied flight times of radio pulses. Light travels about a foot per nano-second, so a GPS receiver has to work with time differences less than a typical clock cycle (for the on-board computer). If you have a 1GHz computer, light cannot travel across the motherboard in a single clock-cycle.

Stunning.

PS: Yes, I do know the details, and I know the inaccuracies in the above.

What does the Traveling Salesman Problem have to do with typical GPS navigation? Typical use of a GPS device involves a start and an end point, and every once in a while a couple points in between, that ideally you may want in optimal order (though I doubt typical devices can do that for you). But there's nothing intractable about the scale of that.

Sorry, you're right. My recent work has involved pathological cases of dynamic spanning trees, and I've recently proved that a sub-class is NPC, and I mis-spoke myself here. The SatNav problem is obviously easier. Given the power of the processor, limited memory, and complexity of streets (in the places I typically deal with) I still find it surprising.

But you're right. Dijkstra's algorithm clearly solves this problem simply enough in O((|E|+|V|) log |V|) for the sparse case.

Apologies.

Interesting article, but I didn't follow the 'destroying local knowledge' part.

"Since before even the arrival of the car, people have worried that maps sever us from real places, render the world untouchable, reduce it to a bare outline of Cartesian lines and intersections."

I've never heard this before.

"The first is that when I use SatNav I arrive at my destination having no real idea where I am."

Then you're doing it wrong. Using a SatNav system has taught me a lot about my local area. Since my wife got her system, I've found three or four new routes just in my city. (Well, city-let, it's no New York.) They aren't all great, but they work and I didn't know that. (And it's not even my device; I don't drive with it all the time.)

Maybe you need to switch to map view and zoom out a step or two. It's like having psychic knowledge of roads you can't see and never would have normally gone down.

I think this is an interface design problem, not something intrinsic to GPS navigation.

Provided that I set the unit to show my position on an always-north-facing map, I arrive at my destination knowing exactly where I am. If, by contrast, I leave it in its default mode of rotating the map around an always-up-pointing car, then I get completely confused.

I also never use the spoken directions, precisely because I want to know where I'm to go, not when to perform atomic actions like turning; and the map shows me where to go. If I need to get on I-93 N, and the map shows me as being near I-93 N, then I'm smart enough to follow the signs and to reckon when to be in which lane, which are more likely to be correct than the map in the GPS unit.

A constantly changing, map-like rendering of what I see out the windshield accompanied by context-free commands is not helpful. A map that magically shows me my current position and direction is.

Pffft... GPS. All you need for navigation is a South-Pointing Chariot: http://odts.de/southptr/

I can't really complain about them. I used one this weekend and it saved me from having to buy a very expensive train ticket. In normal instances the shortest route between 2 points is a straight line. Which would have taken me through a very busy downtown. The Satnav had me drive 5 minutes out of my way to find an artery and saved me 30 minutes. I arrived at the station with 10 minutes left.

If I had chosen the logical route I might still be waiting for the next train.

Also, if you are in unfamiliar territory local knowledge is useless. And if you are local then the navigation is barely useful.

The technology for sat-nav, in other words, was around for several years before it was developed. All these different inventions simply needed to be brought together.

The computer revolution hasn't happened yet. It's still in progress. In any tech revolution, culture lags tech. Galois fields were around for well over a century before we figured out the practical use for them. The electric arc was discovered in 1802, and we still might find additional uses for it. (Relatively low-tech high-impulse+thrust drives using electric arcs to superheat hydrogen.)

Reply on news.ycombinator.com