Some years ago, I drove an ageing and rather temperamental British sports car (as though there were any other sort).
It was handmade out of fibreglass – rather carelessly, if the truth be told – and excitingly unreliable. But it went like crazy and looked quite the Badger’s. Parisians in particular used to drool over it. But no matter.
This car had an old-school speedometer – the sort that has a rotating cable running from the gearbox into the cockpit, and works by means of a spinning magnet and eddy currents.
It worked smoothly and accurately up to 50mph, but from 60mph to 80mph it just lurched happily, if rather drunkenly, between those readings.
Over 80mph (for example, on certain German freeways), it would steady out and work normally again, though you didn’t want to try that too much lest some car part work loose and fly off, such as the gearbox.
But for all the wavering of its needle, I was never caught out or confused by this irreverent speedo.
Mostly, the law required me to be at 50mph or below, where the instrument was precise. Above that, if I needed precision I could just glance at the position of the rev counter or the rate at which the fuel gauge was going down. (The motor had a bit of a thirst, bless its heart.)
In other words, the speedo was self-documentingly imprecise. It reported with unambiguous visual clarity that I was between 60 and 80. In that sense, its accuracy couldn’t be faulted.
Imagine, however, that it had been a digital speedo of similarly British automotive cantankerousness.
On a motorway journey, it would have swept smoothly up to, say, 56mph, and then jumped to 69.865443mph and stayed there until the next fuel stop, 145 miles later.
I wouldn’t actually know at any moment if I were under 70mph, the UK motorway limit, and I wouldn’t know that I didn’t know.
This sort of false precision is a real problem in today’s instant-facts-and-figures society.
If I ask Google, for example, for the “latitude and longitude of Sydney”, it tells me that it can be found at 33.8683°S 151.2086°E.
To me, those co-ordinates imply a tiny area, expressed to with 10 metres: the sort of detail I’d need to describe exactly the lobby of a small, boutique hotel in the crowded part of the CBD. Which is exactly what it is.
But it’s a misleadingly precise answer for a question that isn’t about an individual hotel, but instead about a metro area of well over 10,000km2 that’s home to nearly 5,000,000 people.
They simply couldn’t all fit into the Vault Hotel at the same time, and anyway they’d run out of beer almost immediately.
So spare a thought, if you will, for Wayne Dobson, of Las Vegas, who’s a repeat victim of what you might call “precise imprecision.”
It seems that at least one mobile phone company has a truly dramatic flaw in its geolocation database. Whenever it knows that a phone is in his part of the world, but can’t be more accurate than that, it says so in an unnerving way: it pinpoints his house.
It doesn’t draw a little circle on the map to say, “That phone’s probably in a 2km radius of here,” or a jagged polygon to say “It’s somewhere inside this grid of lines joining the following five transmission towers spread over an 8km2 area.”
It as good as says, “Head to Casa Dobson. You’ll find the phone in the kitchen, next to the kettle, under this morning’s newspaper.”
The problem is that this precisely imprecise data isn’t just a theoretical hassle for telecommunications engineers in the field. It also gets shared with the sort of tracking software that you or I can put on our own phones in case they’re lost or stolen.
So, aggrieved members of the public keep turning up at Dobson’s house, shouting that they “have proof” that he’s stolen their phone and ranting at him to give it back. No-one’s popped a cap in his trousers yet, but you have to think that it could happen.
Worse still, the police have turned up at his house, answering other people’s emergency calls. In this sort of case, the cops are between a rock and a hard place. They have to attend, and they have to start by trusting the official data they’re given by the dispatcher. They can’t simply say, “Ah, that’s Wayne’s place – it won’t be him.”
And that brings us to the critical question: what to do about this sort of thing?
I don’t have a glib answer, I’m afraid.
But I do urge you to be responsible and respectful about how you present your own data. False precision leads to false authority, and that is the sworn enemy of computer security, because it can lead you off on wild goose chases implementing policy that is not actually evidence based – it just seems to be.
In short, don’t be afraid to use the words “about” or “approximately” in your answers, and do your best to quantify your imprecision, as engineers do when they write something like “250mm, plus-or-minus 5mm.”
My current GPS location:
PS. The above location, seemingly precise to within 1% of a nanometre, and reported apparently without irony by my Android GPS app, is actually off by about 100m.