brief thoughts on privacy issues

[DJ Adams <dj adams pobox com>]

Hey all, just joined this list, replying to an archived post forwarded
by Nat yesterday. The privacy concern is valid, but needs to be tempered
(imo) with thoughts on how the backends can and do work. To take the
credit card / ISBN example, a sensible backend is not going to try and
take a 16 digit credit card number and try it against an ISBN match,
knowing that ISBNs are 10 digit. Once a clue has been moulded with a
clue type, either in the frontend, or, increasingly, in the backend,
backend matchers should only do type-sensible things with that clue
text.

I could see the concern in a backend free-for-all, where every clue data
was match-tried against every source, but that doesn't happen right now
(and shouldn't happen anyway, as the backend processing would get rather
crazy).

As I said, it's a valid concern, but there's scoping of the idea of
'privacy' to be thought about, too. What if I'm a user on a *nix system;
access to system-local files might not necessarily be an indication of
"safe". Also, using the IRC/AIM analogy - if data is sent over one of
those mediums (media, blargh), then it's "public" and to be deemed so.
But what about Jabber? Is that public or private (seeing as you can have
local Jabber servers)? It's not cut and dry, is all I'm trying to say.

Cheers
dj
http://www.pipetree.com/qmacro