In the 1980s, Canadian IT press brought to light a brewing controversy around use and potential misuse of our Social Insurance Number (the 'SIN' - Canadian equivalent of USA's Social Security Number) by organizations of all types seeking an easy-to-use unique identifier of individuals. Everyone agreed that the SIN was indeed the best identifier that satisfied the need for easy data processing applications. The controversy was of course centred around the potential for misuse of the SIN. One recalls the arguments advanced back then. On one side of the issue, all about big government and industry using SINs to track, surveil and control citizens in their daily lives a la 1984. On the other was touted the enormous benefits of accurate unique identification of individuals throughout myriad administrative processes like taxation, healthcare, banking and more. And there were lots of nuances and corollary arguments too. As often happens, neither side was wrong.
Fast forward to today's discussions around big data, analytics, unprecedented surveillance by government and industry - the principles remain the same, all the opinions and feelings continue to surface, propelling the same arguments. And my response to it all remains the same as in the letter I wrote to the editor in 1981, to wit: "There is nothing inherently wrong with using SINs as identifiers, same as we might use name, address, picture, fingerprints, voice prints and so on - they're just more efficient. The big questions are about whether individuals or organizations: 1) collect data accurately, 2) store and use collected data ethically or unethically by accident or by design, and 3) have any need to collect the data at all.
I very much like the approach of Information Accountability Foundation www.informationaccountability.org and its 'A Unified Ethical Frame for Big Data Analysis'. Further, Omer Tene's paper 'People Like You' where Tene argues - in our ecosystem driven by analytics fed by observational data - it is inferences that raise the major risks rather than identity.
Inferences may be a good thing, so long as they are drawn accurately, ethically and based on good data. And that's the edge of the slipperiest of slopes. One only has to look as far as published poll results where potential error rates and probabilities are always front and centre with the results. Woe betide any of us adversely affected by data that is incorrectly interpreted, unethically disclosed by accident or design, or should never have been collected at all. With luck, Big Data vs Privacy debates will be vigorous well into the future, with the best interests of the people prevailing.
Rest, Assured. First-World.com