Random Privacy
Eric Cordian
emc at artifact.psychedelic.net
Sat Sep 21 10:25:31 PDT 2002
Tim wrote:
> Not a new idea. Ted Nelson (IIRC) wrote about using coin flips to
> randomize AIDS poll questions. ("Have you engaged in unprotected sex?"
> Flip a coin and XOR it with your actual answer.) I remember talking to
> Eric Hughes, Phil Salin, and others around 1990-91 about this.
[snip]
> The idea is that the
> pollee XORs his answer with a random bit. His answer then doesn't
> _implicate_ him, but overall statistics can still be deduced from a
> large enough sample.
Uh, excuse me?!
I can see how such an idea works if you add a random variable with a known
mean to the data. A researcher could do this before storing the data, in
order to protect the confidentiality of individual respondents, and still
be able to compute aggregate statistics.
However, if you XOR a bit with a random bit, you have something equally
likely to be in either state. Even a large collection of yes/no responses
XORed with random bits is indistinguishable from random data.
So I am afraid I must give the prior message my coveted "Silliest Thing
Said On the Internet This Week" award. <Chortle>
--
Eric Michael Cordian 0+
O:.T:.O:. Mathematical Munitions Division
"Do What Thou Wilt Shall Be The Whole Of The Law"
More information about the Testlist
mailing list