[LINK] opaque sharing/scanning of social data

[Eric Scheid]

I read in todays SMH that the police are thinking of shaking down
Facebook/MySpace/et al with regard to scanning their profiles for unsavoury
types. Guess which LINK professor got their name in lights again ;-)

<http://www.smh.com.au/news/environment/online-networks-face-privacy-battle/
2007/09/13/1189276899605.html>

Anyhoo .. I stumbled across something the other day which is curiously
interesting, and that is the idea of using Bloom Filters for sharing social
network information. For those not familiar with them, essentially they are
a cryptographically hashed version of a data input with some interesting
properties ...(a) they can be of fixed size, regardless of input size, (b)
they are opaque, (c) you can test sample data against the filter, (d) you
won't get a false negative test, and (e) there's a small probability of a
false positive test. Property (e) is nice because it lends itself to
plausible deniability.

<http://en.wikipedia.org/wiki/Bloom_filter>
<http://loaf.cantbedone.org/about.htm>

Thus, it is technically possible for facebook et al to provide a means for
outside entities to search for known data, without actually revealing the
data they have, and without the outside entities disclosing their known data
to facebook et al.

To me, this has both positives and negatives. The positives are immediately
obvious, but the negative point that occurred to me is that this compromise
could be seen as a one more thin edge of the wedge towards social
acceptability of Our Dear Leaders prying into our lives.

Thoughts?

e.