Comments on: “Anonymized” Data Really Isn’t http://www.discoveringidentity.com/2009/09/10/anonymized-data-really-isnt/ Just another WordPress weblog Tue, 24 Aug 2010 03:29:46 +0000 http://wordpress.org/?v=2.9.2 hourly 1 By: Mark Dixon http://www.discoveringidentity.com/2009/09/10/anonymized-data-really-isnt/comment-page-1/#comment-568 Mark Dixon Fri, 11 Sep 2009 04:20:28 +0000 http://www.discoveringidentity.com/2009/09/10/anonymized-data-really-isnt/#comment-568 <p>Thanks for sharing the story! Just another example of the flaws of anonymizing.</p> Thanks for sharing the story! Just another example of the flaws of anonymizing.

]]> By: mike waddingham http://www.discoveringidentity.com/2009/09/10/anonymized-data-really-isnt/comment-page-1/#comment-567 mike waddingham Fri, 11 Sep 2009 04:08:20 +0000 http://www.discoveringidentity.com/2009/09/10/anonymized-data-really-isnt/#comment-567 <p>(Forgive me if this story is included in your posted articles... didn't have time to read through all.)</p> <p>My favorite tale related to this was the story of a Yahoo or Google database of search queries that was 'anonymized' for research purposes. The data provider took great pains to remove any identifying characteristics from the records, for example they replaced the ip address with a scrambled one. The data went to a group of researchers, then the entire data set got posted online.</p> <p>Each anonymous user's data could still be grouped by the random ip address.</p> <p>After some days/weeks, the fatal flaw was found: users, while searching for all kinds of things, also did vanity searches on their names... So, in the grouped data you would see the vanity searches -- with their name in clear text -- right next to some fetish or decidedly politically incorrect search item!</p> <p>Mike</p> (Forgive me if this story is included in your posted articles… didn’t have time to read through all.)

My favorite tale related to this was the story of a Yahoo or Google database of search queries that was ‘anonymized’ for research purposes. The data provider took great pains to remove any identifying characteristics from the records, for example they replaced the ip address with a scrambled one. The data went to a group of researchers, then the entire data set got posted online.

Each anonymous user’s data could still be grouped by the random ip address.

After some days/weeks, the fatal flaw was found: users, while searching for all kinds of things, also did vanity searches on their names… So, in the grouped data you would see the vanity searches — with their name in clear text — right next to some fetish or decidedly politically incorrect search item!

Mike

]]>