Books written by Ray Sullivan

Thursday, 28 March 2013

Fingerprinting Your Route

We've all got fingerprints, and apparently they are all unique.  I say apparently because they are a bit like the common belief that all snowflakes are different - until we catalogue each and every snowflake it is just an assumption based on the observation that they all seem to be different.  Of course, if governments and police agencies have their way then there will be a time that all the fingerprints and all the DNA profiles in the world will be on one database or the other.

What isn't often appreciated is that although each fingerprint is a complex collection of individual whorls and whirls, peaks and troughs, a good analyst only needs a relatively small amount of markers to definitively identify a fingerprint assuming a perfect copy is already on record.  Since the 1930s it has been known that a good analyst needs only twelve such data points, which is a bummer for all those detective stories on the TV where they sigh and shrug about the fingerprints being a partial.  In all probability it is enough if the perp, if that's what law enforcement agencies really call alleged criminals - another assumption up there with fingerprint and snowflake uniqueness, I guess - has allowed his or her fingerprints to be left on a public record somewhere.  What isn't adequately admitted in these police dramas, by the way, is the labour intensive methods still needed to match fingerprints - sure some of the process can be automated but a real person has to sit and do some manual checking still.  And what's all that about, flashing up the photos of all the persons that the fingerprint doesn't belong to?  Why would anyone programme a computer to do that?

Anyway, if you think twelve data points is a trivial amount of data to identify a person uniquely, then you may be a bit surprised that the way you move around your home town can identify you in as little as four data points.  You see, we're all sharing information about ourselves to a number of databases, sometimes in real time.  If you have a cell phone, then it periodically locates your phone.  Use Facebook?  Do you post information about where you are.  Twitter - say no more.

But of course, the phone company knows who you are and if it knows where your phone is then there is a reasonable chance you are there too.  However researchers at MIT have spent an interesting few months looking at what is called 'anonymised' data, data where the identity of the persons associated with the data is suppressed.  They looked at 15 months' worth of anonymised phone data for 1.5 million phone users.  That's a lot of data to wade through - I still get a paper bill for my mobile phone, which I think I use lightly, and the phone, text and data usage information regularly falls across three pages each month. I dread to think what the records for 1.5 million phone users looks like.

By analysing this data they were able to identify mobility traces, as they call them, for the phone subscribers and once this had been done then it only took four data points to identify an individual.  The New York Times did a similar piece in 2006 using anonymised data released by AOL and actually tracked down a specific individual.  This is so Secret State it sounds like a bad plotline in one of those detective stories where they shrug their shoulders at the partial fingerprint, sigh and look wistfully at the flashing images of all the people who couldn't be the perp.

We share our location data because it allows our service providers the opportunity to serve us better, to provide meaningful weather forecasts, local theatre information, lists of local premises willing to charge an arm and a leg for coffee.  But civil libertarians may be a bit alarmed to learn that we're all very trackable and identifiable even when our identity is suppressed.  However all this location information becomes extremely necessary when we want our phone to navigate us to somewhere else, because knowing where we are now is a fundamental starting point for the mapping software.

And finding our way is something we've been obsessed with since our hunter/gatherer beginnings and Google has been helping a bit lately.  So has Apple with their mapping routine, but I think most of us can do without that kind of help.  Anyway, not content with mapping every road and street in the world, always on the day the garbage wagon is due judging by the amount of street view photos with black bin bags outside houses, Google is turning to map indoors.  Apple has responded by buying an indoor mapping company, Wifislam, to try and keep up with Google.

Wifislam uses existing WiFi networks to identify a phone's location to within a couple of metres indoors and also uses its technology to work out where your friends are relative to yourself.  Handy at the end of a pub crawl, I guess.

Neither Google or Apple are working with anonymised data, so it isn't unreasonable that their efforts will result in our locations being identified extremely accurately however, if MIT researchers are correct, it seems that our locations are reasonably predictable anyway.  Perhaps we'll only need three or maybe two anoymised data points to locate us in future?


I can be followed on Twitter - @RayASullivan
or on Facebook - use to find me

Why not take a look at my books and read up on my Biog here

Want to see what B L O'Feld is up to?  Take a look at his website here

Worried/Interested in the secretive world of DLFs?  Take a look at this website dedicated to DLFs here, if you dare!

No comments:

Post a Comment