www.nettime.org
Nettime mailing list archives

<nettime> 'IANA' to revoke .su ccTLD? digest [pope bradley elloi elloi]
nettime's_deja-vu on Sat, 26 Oct 2002 04:03:47 +0200 (CEST)


[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

<nettime> 'IANA' to revoke .su ccTLD? digest [pope bradley elloi elloi]


Re: <nettime> 'IANA' to revoke .su ccTLD?
     Ivan Pope <ivan {AT} ivanpope.com>
     Rick Bradley <roundeye {AT} roundeye.net>
     Morlock Elloi <morlockelloi {AT} yahoo.com>
     Morlock Elloi <morlockelloi {AT} yahoo.com>

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Date: Fri, 25 Oct 2002 20:49:39 +0100
Subject: Re: <nettime> 'IANA' to revoke .su ccTLD?
From: Ivan Pope <ivan {AT} ivanpope.com>

Please, there are many forums for Domain Name discussions. Surely nettime
has taken this as far as it should? Thanks, Ivan

> From: Ronda Hauben <ronda {AT} ais.org>
> Reply-To: Ronda Hauben <ronda {AT} ais.org>
> Date: Fri, 25 Oct 2002 10:59:06 -0400
> To: nettime-l {AT} bbs.thing.net
> Subject: Re: <nettime> 'IANA' to revoke .su ccTLD?
 <...>

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Date: Fri, 25 Oct 2002 13:47:20 -0500
From: Rick Bradley <roundeye {AT} roundeye.net>
Subject: Re: <nettime> 'IANA' to revoke .su ccTLD?

* Morlock Elloi (morlockelloi {AT} yahoo.com) [021025 08:22]:
> > The *real issue* as I understand it is that the IP numbers are critical
> > for the tcp/ip protocol and those have been put into ICANN's hands.
> > The IP numbers must be unique for the messages to have a destination
> > on the Internet.

> The computing resources needed for distributed addressing are here. Witness
> 802.11b developments - self-configuring mesh networks a la wikiwiki and
> similar. The point is to use whatever pipes are available to run addressing
> system, but not the one managed by pipe owners. 
>
> This is crucial. Pipe owners should do routing within their domains, moving
> bits from A to B. Like dialing a telephone number, the telephone system doesn't
> care about the name of the person you call. That is resolved in your phone
> book.

Let's be clear about some real distinctions here:

 - *ANYONE* can assign any address to any network device already.

 - *ANYONE* can use any (known or heretofore unknown) networking protocol
   on any network device, as he wishes.

 - *ANYONE* can use any name assignment system he wishes, already.

 - It is only when we wish to communicate effectively, to share resources, to
   identify and access other systems on these networks that we have to 
   agree on interoperable routing. 

 - To access a remote resource we need meaningful addressing.

 - To access an addressed resource we need useful, reliable routing.

Pipe owners already do routing within their domains.  Network owners (down to
the 2-device network) do routing within their domains.  Many people do their
own addressing, naming, and routing with private schemes and protocols (though
TCP/IP has become a de facto lingua franca between many devices -- personally I
feel this is a feature, not a bug).

If, however, a network needs to access another network using resources the
network owner *cannot afford to own*, then they must interoperate with
providers willing to provide access to them at affordable (for the customer)
prices.

The providers with the most resources have agreed to be bound by certain
conventions -- primarily because these conventions are practical and effective
means of solving the technical problems inherent in long-haul communication.  A
sizable chunk of the administration for the real world portions of
interoperability under these conventions, for better or for worse, goes through
what is being called IANA.

When you speak of using mesh addressing and routing technologies to escape the
addressing schemes of "the pipe owners" ("RA! RA! Fight The Man...") let's be
clear here:

 - you (via use of technologies in the "unlicensed" 802.11 spectrum)
   effectively *are* the pipe owner

 - you are then using the addressing scheme of the pipe owner (bear with
   me and see how this grows into a significant statement)

 - when more than a couple of you get together to communicate (else this
   network is of little practical benefit) you will have to work out a 
   system whereby you can route to one another, address one another, and
   (almost certainly) name one another

 - if your decisions are made well the network will grow -- otherwise some
   other network will grow and this discussion applies to the network of
   relevancy

 - as a successful network grows there will be contention for resources --
   access to high capacity routing, access to valuable information
   repositories, access to desirable names; otherwise some parts of the network
   will have poor connectivity; otherwise there will be a number of (e.g.)
   "Indymedia"s visible on the network; otherwise a device will not be
   reachable from some areas due to duplication of addressing, etc.

 - in order to manage this contention coordination must take place -- some of
   which must be centralized, otherwise a rogue connected subnetwork may ensure
   that the network at large is not reliable or useful

 - when access to legacy resources ("www.cnn.com" on the Corporate Owned
   Internet) is desirable (and it will be) you will have to administer
   connection points to the outside network to interoperate, and routing
   on your non-legacy network will have to provide dependable results to
   its users as well as to outside resources

 - at some level maintaining a useful network (even when "you" "own" "the
   pipes") will require centralized administration and decision-making --
   otherwise the network will become useless.  If you don't think so, wait
   until the spammers find your anarchic utopia...

Don't fool yourself into thinking that some sort of wiki-inspired 802.11 meshes
are some anti-establishment panacea.  Eventually they will become
establishment, with all the socio-political problems adherent thereto.


> Imagine if AT&T and baby Bells managed names instead of numbers ... and if
> Ronda Hauben didn't pay the monthly tax she would become unreachable.

The analogy doesn't work.

If Ronda Hauben doesn't pay the monthly tax (i.e., her phone bill) her service
is disconnected.  The registration (directory) is tied tightly to her telephone
account.  To have a directory listing you must have a paid telephone account.
When the account is unpaid Ronda is unlisted (modulo database update and
publication frequency).

There are a multitude of telephones which are unlisted, actively or passively.
There are a number of phones with multiple listings.  Similarly there are a
multitude of network devices which have no DNS name entry.  There are a number
of network devices which have multiple DNS entries (I'm sitting at one right
now, and mailing through a chain of 3 or 4 more).  [ This discussion completely
ignores the massive number of "telephone" "lines" used for voice and data
services which bear little functional resemblance to the colloquial conceptions
about what constitutes "telephony".]

The advantage the telephone system has in naming is that people generally look
up name entries based on locality.  Currently this is not the case for network
devices (nor, actually, would this be particularly desirable for a number of
reasons).  Note that 800 numbers (and their 8* cousins) occupy a limited space
which is under contention similar to TLD names.  800 numbers also have other
similarities to DNS-listed network devices.

> Imagine unmarked streets and roads and no house numbers, but if you pay you get
> directions "... on third street turn to the left, enter 5th house and knock on
> the second door from the right."
> 
> Same thing.

When was the last time you made an active decision in how your packets were
routed across the Internet?  Network devices on TCP/IP networks typically
specify their next hop for routing purposes, and packets carry only source and
destination information (for the technically savvy stickler, find me a
nontrivial public network which allows TCP/IP source routing...).  Network
intersections (routers) are often named, but this is more for administrative
and diagnostic purposes than anything.

The problems with the driving/walking vs. network routing analogy are

 - We don't typically expect to be routed from our doorstep to our destination
   with nothing more than "www.bobs-house.com".  Were that the case we would
   not need access to street signs, landmarks, etc.  Only the taxi driver (if
   you will) needs a system to navigate us.  It is because the average
   individual must do the navigating that street signs exist.

 - On public TCP/IP networks we *do* expect someone to route our packets from
   our doorstep to our destination.  We pay a connectivity fee which filters
   through our ISP to (hopefully) eventually cover the costs of administering
   routers on the backbone.  We are not allowed to route across someone else's
   network, because this is technically inefficient (not to mention egregiously
   insecure) to the point of being unworkable.  

 - We don't pay separately for routing advice, so I don't know how that is
   supposed to fit into your analogy

Now if an edge user of the Internet has consistently better real-time routing
advice than the OSPF-, BGP-, and RIP-aware routers on the backbone I'm sure the
router maintainers would be interested in knowing about it.  For the remaining
100% of us we leave the routing decisions to core routers and their admins, not
because we're the oppressed masses, but because that's the only workable
mechanism.  For technical reasons.  In the Real World.  Hello.

> Right now, the addressing system is closely intertwined with the routing. What
> is needed is public knowledge of routing endpoints, and then each member of the
> public maps addresses over that. Like road signs.
>
> Transistors are small enough and cheap enough to handle this today. The reason
> for centralised addressing is no more technical - it's only a tax-collecting
> issue. A toll gate.

See above.  This line of thinking is fundamentally misinformed.  I'm not even
going to ask why the size of transistors (which has been small enough since the 
early 80's to handle real-time routing decisions) has a bearing on these
matters.

> This is important to understand - technology is at the point where the biggest
> concentration nodes and huge corps based on them are about to become
> irrelevant. We will drive our own bits the way we see it fit.

Good luck.  Note that I say this with a liver full of bile and a mouthful of
venom for the incumbent carriers.  They are mismanaged, disorganized,
politically corrupt, customer-hostile, and in bed with every political lever
from the FCC to the Commerce Dept.

That said, the argument presented is, and the arguments relevant are, on
technical merits.  The one presented is fundamentally flawed.

> In the meantime, transportation companies are smashing private cars. As more
> users learn how to drive that will become the needed annoyance point.

(Wtf?)


Where do I agree?

1- The potential for distributed routing and network formation is awesome
   and intriguing.  The ability to escape LEC control via disengagement from
   land-line access is powerful and will (rightfully) upset telco business
   models.  Hopefully this will work for the greater good.  Do not, however,
   discount the staying power of large corrupt businesses.

2- There are scalability issues with the current DNS, and IP allocation 
   systems.  I don't agree that there are currently scalability or control
   issues with Internet routing over the backbone.  Propose a viable
   complete alternative to getting my packets from my hub to your inbox and
   I'll listen.

3- The entities in charge of administering IP and DNS details have too little
   oversight and have a strong incentive to profit at the expense of the 
   users of the Internet.


1 will take care of itself, particularly if we, as experimenters and end-users,
play with the technologies in our own backyards.  Do what we will.  Don't ask
permission, just do.

2 will be aided significantly by widespread deployment of IPv6.  When any
customer has a dedicated 32+ bit address space of his own to play with the bulk
of address crunch issues become irrelevant.  Don't like the administration of
the mainstream DNS system?  Promote use of reliable alternate roots.  Frankly I
haven't found a stably run alternate root myself, but I don't think mainstream
DNS is completely broken either; and that's not to say there isn't or can't
be a viable alternate root.

3 is the major issue which desires action.  Unfortunately it is an issue which
is orthogonal to the technical matters discussed.  There will be contention for
resources and eventually coordination must take effect to allow any large
network to function.  Address the socio-political issues and don't pretend that
some new whizbang gizmo will dislodge The Man from your shoulders.

Rick
-- 
 http://www.rickbradley.com    MUPRN: 438    (64F/64F)
                       |  very action. I
   random email haiku  |  think they were calling them pink
                       |  contracts or something.

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Date: Thu, 24 Oct 2002 01:43:24 -0700 (PDT)
From: Morlock Elloi <morlockelloi {AT} yahoo.com>
Subject: Re: <nettime> 'IANA' to revoke .su ccTLD?

> The Internet perceives ICANN as damage and routes around it?

Back to my original argument, there is very little damage.

If you are not a peddler of something that needs to be easily addressable from
Uzbekistan and Tasmania at the same time, you don't really care about global
dictionary. Like your phone book, your address book is your DNS. People you
care about will let you know when they change address.

In my phone book, "joe" may translate to a7746 {AT} 212.23.201.33 ... who cares ? I
have friends that serve http from 212.23.201.33:81 or some other non-standard
port.

Now, for pathological marketoid cases and blog celebrities that have 3000
online "friends and contacts" ... tough shit. They represent irrelevant
percentage of 'net users.

"people" don't care about ICANN because ICANN caused zero problems to them so
far. They don't even know that it exists. The nature of human communication
limits the number of close contacts to a small set quite manageable in a very
old fashioned way, where ICANN cannot cause damage.

However, the moment you want to have your name on a global billboard, be
prepared to be gunned down by folks with bigger guns. There is nothing unfair
or immoral about it. Either buy a bigger gun or stop whining.

A piece of a global namespace cake is expensive, like any other. There are no
entitlements. One doesn't get global exposure for free. It's easy to understand
once one gets over barlow-induced dellusions.


=====
end
(of original message)

Y-a*h*o-o (yes, they scan for this) spam follows:

__________________________________________________
Do you Yahoo!?
Y! Web Hosting - Let the expert host your web site
http://webhosting.yahoo.com/

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Date: Fri, 25 Oct 2002 13:54:08 -0700 (PDT)
From: Morlock Elloi <morlockelloi {AT} yahoo.com>
Subject: Re: <nettime> 'IANA' to revoke .su ccTLD?

> Let's be clear about some real distinctions here:

...

You need to read the post before replying to it.

I'll help and summarize it now, but will not bother in the future.

1. normal communication addressing needs can be handled locally. Most people
don't care about global visibility. Think of it as source routing if that
helps. I want the map visible, I'll compute the route, and road owners don't do
my routing for me, and they can't influence visibility.

The growth and aggregation of small nets that you mention is globalising - and
no wonder that centralised mechanisms are the only solution that you see. "More
than few" is patently false as the size limit. uucp based nets with forward
routing handled hundreds of thousands of addresses in 4.77 MHz 8086 times.

I wonder why you oppose the current working one. It's also funny that you point
out "unlicensed" attribute of 802.11b. Is the only legitimity the one that is
enforced by concessions granted by men with guns ?

2. When I mentioned AT&T handling names, that was the *illustration* of what
would happen if telephone system was handled as DNS. We all do know well that
ATT handles numbers.





=====
end
(of original message)

Y-a*h*o-o (yes, they scan for this) spam follows:

__________________________________________________
Do you Yahoo!?
Y! Web Hosting - Let the expert host your web site
http://webhosting.yahoo.com/

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

#  distributed via <nettime>: no commercial use without permission
#  <nettime> is a moderated mailing list for net criticism,
#  collaborative text filtering and cultural politics of the nets
#  more info: majordomo {AT} bbs.thing.net and "info nettime-l" in the msg body
#  archive: http://www.nettime.org contact: nettime {AT} bbs.thing.net