From: Brad Templeton (brad@templetons.com)
Date: Fri Jan 19 2001 - 14:38:23 CST
On Fri, Jan 19, 2001 at 08:49:00PM +0200, Kai Henningsen wrote:
> brad@templetons.com (Brad Templeton) wrote on 19.01.01 in <20010119101231.C27968@main.templetons.com>:
>
> > On Thu, Jan 18, 2001 at 10:22:00PM +0200, Kai Henningsen wrote:
> > It was my understanding that Netscape posts from the server you are
> > reading from. You can of course have multiple servers, but you can't
> > as I recall read a newsgroup from one, then say "post" and have the posting
> > to to another.
>
> Well, that was certainly how I imagined people to do this.
Well, I tend to want to read newsgroups from the server that gives me
best performance, which is of course the one in the server room at my
ISP, over my multi-megabit cable modem connection to them.
It would be OK if I could read from that server and when posting post via
another server that protects my privacy, that would not be inconvenient,
though it would be annoying to have to pay extra fees for that.
However, most news software does not support this. Some support multiple
servers, but you post from the same server you're reading the group from.
> Well, I'm not quite up on modern anonymizing servers, but I sure *thought*
> there were some around that worked for free. That was certainly how
> anon.penet.fi worked.
And at the first major trouble, it closed and this is to be expected because
when you do it for free, what's the point of putting up with a lot of
trouble?
The ISP is not doing it for free. They do it as part of the service I pay
them a monthly fee for. They have reasons to be responsive to customers
if the software lets them do it.
>
> > But let me clarify what privacy means. It doesn't simply mean creating
> > new identities to post under. Many people use an address which does not
> > have their real name as their _sole_ identity on USENET.
>
> A practice that, incidentally, I have rather little respect for, unless
> they are in Usenet *only* for the few groups where there is a real need
> for it.
> Mind you, I do thing that it's their *right* to do that - but I don't
> think it's a good idea.
I agree with you. As you may note, I post with my real name, and with a
real E-mail address which gets to me (though a spam filter.) That's more
than most people do.
I even agree with you that it makes most groups better when people don't hide.
But I believe that there are some groups which are made better by letting
people be anonymous, including groups around medical disorders and other
private problems. Groups about sex. Groups for gays. Groups for any
counterculture. Groups where people want to be open about their industry
or employer without revealing they work for that given employer. Such groups
are a big part of what USENET is about.
My point is it should be a per-group or per-hierarchy policy decision as
to whether people can post under pseudonyms. Not a netwide decision.
Thus this standard shouldn't encourage the use of personally identifiable
information in audit trail or spam-blocking headers, and the authors of
injector software shouldn't have done this by adding nntp-posting-host by
default.
They made a netwide *policy* decision that it would no longer be possible
at most sites to post to any group at all without revealing personally
identifiable information. That changed the character of the net.
Now the question before this group is how to standardize such efforts.
Though the nntp-posting-host header has existed for several years, that
does not make it a standard we now have to uphold. When introduced it
was a change, and it turned out to be a change in netwide policy. This
body should not ratify a policy change, but rather standardize how
local groups and admins can make and execute policy decisions.
> Ah. So all you want is sham privacy - the kind that won't deter someone
> who really wants to find out. Reading nana-u, it sure looks like there are
> quite a number of those around.
No, it's not sham privacy at all. It's your ISP not disclosing your real
identity to the general public. When I go to the bank, they know all my
financial secrets and I expect them not to give them out except under
proper court order. That isn't sham privacy, it's normal privacy.
Anon servers, which make it impossible to get your identity at all, like that
from ZKS, are extremely high level privacy. There are many lower levels and
that doesn't make them shams.
> > It doesn't protect them from their own ISP but yes, it stops random people
> > finding out their IP address from their USENET postings.
>
> Their IP address being just about the most unimportant thing to know about
> most people, them being behind a dynamic IP.
But I have a static IP, and so do millions of others.
>
> Given I'd expect the vast majority of Usenet readers to *not* have local
> access, I sort of fail to see how that could possibly be a point.
>
> And no, an ISP's news server is *not* local access any more than a typical
> specialist usenet provider's is.
It's not local-local but nor is it remote. For the millions who have DSL or
Cable modems, it does make a difference if their news server is in their
own ISP's server room or out past the congested peering points. A big
difference.
>
> > Why not design it centralized, the administration
> > is a lot simpler.
>
> Admin of a centralized system serving several millions of newsreaders,
> *simple*?
Yes, compared to having 300,000 individual admins handle it.
>
> For most people, this *is* a system for use from remote servers.
There are 3 levels then, One is your own LAN, the other your ISP and
then the cloud beyond. I use remote to refer to the cloud beyond. Most
people do not read USENET from a remote server.
>
> > But then, why have injectors at all? Why have hundreds of thousands of
> > admins administer injectors and have dozens of software authors write
> > and maintain them, when in fact a few injectors centrally located would
> > do the job. Just tell people when they want to post they should
> > switch servers and post there.
>
> I think you need to go and talk to some people who understand load.
I do. Every full-feed USENET server already handles every article.
Injection is more work than handling an article in batch, but how much
more work? Twice? 10 times? I doubt it is 10 times, but if it is,
it means that 10 servers capable of handling USENET feeds could handle
the entire injection load of the net. Or 100 or whatever factor more
difficulit you think injection is.