From: Scott Brickner <sjb@universe.digex.net>
To: “E. ALLEN SMITH” <EALLENSMITH@ocelot.rutgers.edu>
Message Hash: b2e46d1e1a95e0b80959d1e5c21030c9b5102036746cd9eff35fe1bd38d8c558
Message ID: <199605061824.OAA10053@universe.digex.net>
Reply To: <01I4A37X6Y0W8Y56P8@mbcl.rutgers.edu>
UTC Datetime: 1996-05-07 04:49:02 UTC
Raw Date: Tue, 7 May 1996 12:49:02 +0800
From: Scott Brickner <sjb@universe.digex.net>
Date: Tue, 7 May 1996 12:49:02 +0800
To: "E. ALLEN SMITH" <EALLENSMITH@ocelot.rutgers.edu>
Subject: Re: PICS required by laws
In-Reply-To: <01I4A37X6Y0W8Y56P8@mbcl.rutgers.edu>
Message-ID: <199605061824.OAA10053@universe.digex.net>
MIME-Version: 1.0
Content-Type: text/plain
"E. ALLEN SMITH" writes:
> I am not quite certain if the model of
>[content provider]-[ISP]-[Phones]-[ISP]-[ISP]-[user] is going to work much
>longer. That routes the material through quite a few too many bottlenecks,
>among them the phone lines. I could reasonably easily sign up with two ISPs
>and start myself as a router (with a good computer and the right software),
>from what I know of the subject; with ecash routing of messages, this might
>get quite common (and profitable).
> When you've got a few large organizations doing the routing, what
>you've said is _probably_ correct. When you've got a lot of people doing it
>out of their garages, then it isn't.
> -Allen
The problem is that it requires the cooperation of both of your ISPs.
You'll never receive packets to route from either of them unless you
have some sort of contract in place. In the scenario I outlined, the
"common carrier" status of the ISPs is contingent on their following
the censorship protocol, so their contract will require that you, too,
follow it.
The network layer isn't the geodesic Bob H likes to talk about. That
doesn't happen until the transport layer (one higher). It's a
heirarchical star, with a relatively small number of big ISPs acting as
the hub, several groups of regional ISPs acting as local arms, and many
local ISPs acting as the end-points.
Even in the face of a "digital silk road", this isn't likely to
change. The cost of operating a router is proportional to the number
of connections it has. The vast majority of traffic doesn't have
stringent enough delay requirements that it'll be willing to pay the
additional cost of going through a very highly connected router.
Therefore the hierarchical star configuration is near-optimal for
normal traffic (and pretty much all of the stuff that they claim they
want to censor).
Return to May 1996
Return to “Scott Brickner <sjb@universe.digex.net>”