Showing posts with label anonymous. Show all posts
Showing posts with label anonymous. Show all posts

2009-09-08

High SNR Sentences: Identifying You in Data

It was found that 87% (216 million of 248 million) of the population in the United States had reported characteristics that likely made them unique based only on {5-digit ZIP, gender, date of birth}.
The sentence was from a CMU researcher working in 2000 on 1990 census data. Unfortunately the paper is behind a wall, but a different, available paper with a decent methodology puts the number at 63%. I was not comforted by the lower estimate. Tip of the hat to Ars Technica for an interesting discussion on "anonymizing" data.