Saturday, 14 January 2012

A quiet pint

and which one are you?

A day off, well part of a day off anyway. Proposal documents don’t abide by the usual tempo of the working week. Anyway a few hours to read, relax and tune out the mind. In short I could go to the pub. Hopefully to avoid the pillock from a local council with his insane ideas to inadvertently humiliate or enslave humanity in the name of road maintenance.


See this complete nimrod wants to fit every car with an accelerometer and a very accurate GPS device so that as cars drive around the town they can inform him of every pot hole and road defect automatically when they get in range of their home wi-fi. Thousands of remote sensors telling him the exact state of the road surface daily. It’s a bureaucrat’s wet dream, well a road maintenance bureaucrat’s wet dream anyway. I said to him, “but you’ve just uploaded everyone’s car movements up into a public data set? What about the privacy implications?” predictably he scoffed “ah but I’d be anonymous data, there’d be no privacy risk”.

The usual response from someone who doesn’t understand the power of analytics and very very cheap computing cycles.

If you combine a large anonymised data set with a small but explicitly personalised dataset you can draw very detailed conclusions from it. Say you had someone’s credit card records, you could draw conclusions from their purchases of petrol and certain journeys, you could note a specific journey always occurred after a petrol purchase, within a margin of error of course. But you could say, hold on track #123456 on any given day is probably Ed, look there it is, a journey 2.5 minutes after a BP petrol purchase down to his house. And what conclusions could you draw if that predictable journey changed? Look  track #987654 (journey 2.5 minutes form credit card event #314159) now pauses outside number 37 Railway Cuttings for 3hrs before going home? Would that be a cause for investigation by an interested party?

Of course the buffoon from the local roads dept blusters, “but who’s got your credit card number, who would know you are having an affair?” (guilty conscience there perhaps?) 

“Well apart from your partner” I suggest gently “and who could be more interested?” 

“oh this is improbable Ed, simply improbable”

“what when you’ve a massive public dataset and you can get an Amazon EC3 stack to crunch the numbers for less than £2?”

The classic retort form the petty apparatchik comes next

“well if you’ve nowt to hide, you’ve nowt to fear lad” say’s he.

I give up, “how’s the wife?” I enquire.

“she should be here” says our man “she’s just nipped off for some petrol, running a bit late”…..

I move away and finish my pint in silence.

1 comment:

  1. Just for fun take a look at http://www.aclu.org/ordering-pizza for a nightmare scenario of data mining!

    ReplyDelete