[clug] Cleanfeed. Other Comsequences.

David Tulloh david at tulloh.id.au
Tue Oct 28 00:25:50 GMT 2008

jm wrote:
> There are other methods than pure lists involved. Similar techneques
> to spam filtering can be used, eg, key word or bayesian filters. These
> are more likely responsible for these false positives/negatives. Just
> try to send an email to anyone with the word "free".
I'm not sure about this.

The software that was trialled certainly seemed to do content analysis
and it's easy to see how this would cause delays and false positives
(imagine designing a robust filter to stop sexual content but allow safe
sex education or breast self examination techniques).

However if you look at what has been announced it seems like it will be
lists.  The fact sheet [1] that the government took to the election
discusses the ACMA black list and expanding the black list by linking it
with US and European equivalent listings.  The letter from Mark Newton
[2] also discusses the dangers of widely spreading such a list.  Based
on some of the things Mark Newton has said, I have been assuming that he
has seen some of the proposed implementation details.

So the Government has certainly talked about intelligent filtering but I
haven't seen anything that indicates that they have chosen to go with
such a system.  The uncertainty around this is another indicator of just
how little information they have released.


1: http://www.alp.org.au/download/now/labors_plan_for_cyber_safety.pdf
2: http://users.on.net/~newton/ellis-2008-10-20.pdf

More information about the linux mailing list