But in their experiment, the researchers were able to prove, mathematically, that such answers wouldn't get someone any closer to inferring information
about any single person or business that might violate privacy regulations — such as whether an employee held a job at a
particular workplace, or precisely what fraction of a
company's workforce belonged to a
certain race or had a
certain level of education.
What is perhaps different
about Beagle is that the focal point of the work is not thousands of documents that are to be examined for one
particular case, but rather helping a
company examine
certain types of document over a long time period.