Florian Cramer on Mon, 10 Oct 2005 23:44:18 +0200 (CEST)


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

Re: <nettime> Who will own and control the Internet's infrastructure?


Alex:

> DNS is entirely unnecessary for the functioning of distributed
> networks such as the internet. It is simply a convenience: people
> prefer to read addresses as words rather than as numbers. 

This is not the only function of DNS. I would even argue that it's the
less important one. More importantly, DNS provides a second-level meta
adressing layer for the Internet, i.e. one that abstracts from the
physical locations. And what's more, there is no one-to-one, invertible
correspondence between domain names and IP numbers -  which would
complicate things further.

Let's take nettime as an example. "bbs.thing.net" resolves to the IP
address 64.115.210.15. I could, as you suggest, simply send this posting
to nettime-l@64.115.210.15 instead of nettime-l@bbs.thing.net. If you
reverse-lookup this DNS entry, then 64.115.210.15 resolves to the domain
name "static-64-115-210-15.isp.broadviewnet.net". I.e., Thing.net
currently runs via broadviewnet.net; its domain name and its subdomains
resolve to static IP numbers provided by this ISP.  Thing.net had to
change its ISP in 2003 after its old ISP Verio had taken them offline,
for political reasons: The Yes Men had issued a fake Dow Chemical press
release and spread it via thing.net (full story here:
<http://www.villagevoice.com/news/0304,carr,41320,1.html>).

If The Thing and Nettime wouldn't have had domain names (=DNS entries),
but only IP numbers, they both would have vanished from the net on the
spot. The damage would have been almost irreparable. 

> If DNS and the domain names disappeared tomorrow, the internet would work
> just fine. 

>From the view of the ISPs, yes, from "our" perspective, not, because we
would become the slaves of our ISPs and physical network locations. 

> We'd all be using IP addresses and the other name spaces already in
> existence. 

Another counter-example: Because of the current shortage of IP
addresses and a transition to IP v6 that never seems to happen, most
private Internet users don't get static IP numbers from their ISP, but
different IP numbers each time they log on. Thanks to the domain name
system and services like dyndns.org, it is still possible to host a
server, or log into your own computer from the Internet (via ssh, VPN
etc.) even if you don't know your temporary IP address.

Yet another counter-example: the domain name system allows you to map
different domains (DNS entries) to a single IP address and make them
appear as different servers. The apache http server can map different
subdomain address queries to different "virtual hosts". If you own the
domain "foo.org", you can create an arbitrary number of different
websites like "bar.foo.org", "xy.foo.org", "test.bar.foo.org" etc. under
a single IP address and without registering the subdomains in DNS.
Again, this takes control away from the ISPs and gives it to you.

> (I'm hesitant to say that control over IP number assignment means
> control over the internet, but I'd be interested to hear that argument if someone
> wants to make it.)

Well, it certainly does - as you wrote yourself, IP numbers are vital
for the technical functioning of the Internet, domain names are not. 
The shortage of IP v4 numbers and the slow transition to IP v6 already
has, as pointed out, negative consequences for Internet users with
standard ISP access. Yet things could be much worse. Imagine
ICANN (or whoever may be in charge of IP number assignment in the
future) would suddenly decide - disguised perhaps as a "security" policy 
- that IP numbers may no longer be assigned to non-corporate individuals, 
or, because of their shortage, would be available only for $10,000 per
address and year. 

The fact that IP number assignment (fortunately!) hasn't been
controversial yet doesn't mean that it will stay that way.  If IP number
and DNS control were shifted from ICANN to the United Nation's ITU, for
example, it might politically look good on paper, but could be
devastating in practical life considering how the ITU backed
monopoly-controlled telecommunication in the past.

> Additionally sites like google have made domain names more and more obsolete.

I am not sure which advantages it would have to give up domain names
that can be owned and controlled by individuals - and pass over that
control to an entity like Google.

> Another example: peer-to-peer technologies such as Gnutella have essentially zero
> reliance on the DNS. (Yes, domain names are conventionally used when bootstrapping
> with a web cache, but strictly speaking web caches are a convenience not a
> necessity, and IP addresses would work just as well.) Fully distributed p2p
> applications are widely available for most of the things we do online: email,
> chat, file transfer, etc.

But they all have to implement their naming and addressing schemes
internally. This only shifts the DNS issue to the level of the
individual P2P protocol, with even less public accountability for the
address assignment politics.

-F

-- 
gopher://cramer.plaintext.cc
http://cramer.plaintext.cc:70



#  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@bbs.thing.net and "info nettime-l" in the msg body
#  archive: http://www.nettime.org contact: nettime@bbs.thing.net