Friday, June 10, 2005

The Press, GPS, and road tolls

Why is it that nobody ever gets it right about GPS? Every time the magic navigation system is mentioned in the national press, someone always tells us that the satellites "track" the user. This time, it was the Grauniad, and worse still, it was its normally fantastic science supplement, Life. In a report about the government's delightful road-charging scheme, they said this:
"Darling wants to levy varying fees from 2p to £1.30 a mile according to the distance, time and type of road. The idea is that a black box in each car, roughly the size of a DVD player, will be tracked using global positioning systems (GPS). Many family cars already have GPS, and Ellen McArthur used a similar device to navigate the high seas. The in-car box is typically monitored by several satellites, which combine and "triangulate" to give accurate coordinates."
No, the black box will not be tracked. No, it is not "typically" monitored by satellites. It is never monitored by satellites. They nearly got it right in that it involves triangulation, but again, the satellites do not triangulate: the box does. But that wasn't it, though...
"New Galileo technology will improve this by using perhaps eight or 12 satellites to pinpoint each black box."

Let's get this straight. The NAVSTAR satellites that make GPS work do not track, monitor, pinpoint or do anything else to the receiver. They are essentially highly accurate clocks. The whole system architecture is exactly the opposite of this: for a good reason. Like the internet, it's a stupid network - all the intelligence is at the end-point. There's a damn good reason for this, too. GPS was designed to work anywhere and at any time, even in a nuclear war. If the satellite control centre in Colorado Springs or wherever is blasted down to nothing, the system functions just as well, because it all happens on the user device. What happens is that the user device contacts the satellites and records how long it takes for them to send back a signal, which tells it how far away each one is at that precise moment; where the circles intersect is your position, so long as you're within the satellites' orbit. The receiver tracks the satellites. Monitors them. Not the other way around. Is that completely clear?

It's an interesting question why they always get it wrong. If it was purely ignorance, you'd expect the mistakes to display a normal distribution. Given that there are essentially two possible options, this would suggest a fifty-fifty split. In practice, there will be a small but significant chance of someone saying it works with skyhooks, fairies and Miracle Whip, so it won't be exact, but being right or wrong would be equally probable. But in this case, they always get it wrong in the same way. (Pity about the rational expectations hypothesis, eh.)

I wonder if the idea that it tracks you, not the other way around, is some kind of unconscious reflection of one's fears of total surveillance? You won't be surprised to learn that I think Alastair Darling's scheme is poison, especially as he's already pledged to make an equivalent cut in fuel duty. This is so stupid as to be unbelievable: the one policy lever that gets right to the problems caused by cars (i.e. pollution, climate change and resource depletion), has been demonstrated to work, and requires absolutely no new administrative apparatus or technology, and he's carefully ruled it out? Under Darling's idea, drivers of monster SUVs in the suburbs will actually benefit.

Another point is that the proposed box is vulnerable to fiddlers precisely because of the way GPS works. It tracks the satellites to work out where it is. Then, every so often, it rings up and says where it's been (just like the electronic tag used on criminals). Of course, that gives the hacker a week or so to break into the thing. Now, a navigation device that did track the user would be far better, as it would require only a very cheap client and would collect the information centrally and in real time. But there aren't any at the moment. It's conceptually possible, but I doubt it would be better enough than GPS/Glonass/Galileo to make it worthwhile, especially as centralisation in computer networks means complexity and cost.

Stupid idea, stupid government, stupid press.

No comments:

kostenloser Counter