on Tue, Dec 16, 2003 at 09:09:18PM +0000, Colin Watson (cjwatson@debian.org) wrote:
> On Tue, Dec 16, 2003 at 01:34:03PM -0700, Wesley J Landaker wrote:
> > On Tuesday 16 December 2003 1:08 pm, Joerg Rossdeutscher wrote:
> > > A mailserver can harm _others_.
> >
> > I totally agree. Which is why I'm all for only allowing arbitrary
> > entities to determine who can and can not run a mail server. What we
> > need is more control, more censorship, more penalties, and less
> > interference from subvertive terrorists who try to route their mail
> > around the system. The only reason they have to be doing something like
> > this would be if they had something to hide. I believe that their
> > computers should be confiscated and their citizenship revoked.
>
> Let's turn this around: why should *I* be forced to accept mail coming
> from a dynamic IP, when statistically such mail appears much more likely
> to be spam or viruses? Who are you to tell me that I have to accept such
> mail?
Statistically, mail from any arbitrary source is more likely to be spam
or viruses, than not.
Statistically, mail from the US is more likely to be spam or viruses
than not.
> (If it's not obvious why direct mail from dynamic IP addresses is a
> favourite tool of spammers, it should be.)
Spammers will abuse what they can get their hands on.
My own response is:
- IP-based discrimination is at best a blunt instrument. Where
applied against specific netblocks based on known history, it's at
least actionable. Even whole-country blocking works as a goad to
encourage countries to start getting serious about securing their
domain -- or downward-delegating such responsibilities. These days,
you can expect to find abuse@ and postmaster@ addresses to work for
many domains in China, Korean, Hong Kong, and Taiwan. Vastly
improved over a couple of years ago.
- Dynamic IP blocking is per se unaccountable. There's nothing the
owner of a particular address can do to secure the address which
impacts the listing. This isn't a listing based on behavior of the
address. It's a listing based on an independent attribute of the
address. Some users may be able to get their provider to remove
IPs from residential/dialup lists, but not all.
- There are highly specific filters and methods which can effectively
discriminate between spam and non-spam content. Activity-based
lists, Bayesian and content-based filters, reputation systems,
teergrubbing, rate-limiting, and the like.
> This is *not* censorship, by the way.
No. It's arbitrary discrimination.
And for your own personal email configuration, it's your call.
This isn't acceptable for general-purpose communications, however. And
I'd suggest you look into common carrier laws as well (I'm somewhat
familiar with US statutes) as to showing preferences by customer. I see
little distinction between this practice and the illegal real-estate and
insurance underwriting practice of redlining neighborhoods.
> Censorship is when the government represses your speech.
NB: Not strictly true.
> > Oh yes, and blacks to the back of the bus, please; just be happy we
> > let you on at all.
>
> It's a weak argument that requires a comparison to racism to be heard,
> not to mention that it demeans the plight of those affected by racism.
The similarity is this: a secondary indicator is being used to
establish an absolute preference for or against a specific activity.
Despite the known invalidity of this indicator in a large number of
cases. And the existence of more specific, accurate discriminators.
Peace.
--
Karsten M. Self <kmself@ix.netcom.com> http://kmself.home.netcom.com/
What Part of "Gestalt" don't you understand?
Backgrounder on the Caldera/SCO vs. IBM and Linux dispute.
http://sco.iwethey.org/
Attachment:
pgp_br2lV_oku.pgp
Description: PGP signature