[Air-l] A definition of the internet

[elw at stderr.org]

> I pretty much concur... with the caveat as you mention the slow but 
> supposedly impending move to IP v 6...  and with the idea that the 
> functionality of TCP/IP actually being replaced.... by an equivalent or 
> better protocol/protocol defining mechanism.


I think the first book I saw on IPv6 was circa 1994; even then, it was 
heralded as the "next big thing", soon to be coming to a network very near 
you.

We are, in fact, much closer to IPv6 now than then.  At that point, the 
software on the edges of the network - on the clients, built into the 
daemons running on servers providing service to the edge, built into the 
routers that make the network RUN - just wasn't there.

In a very real sense, the technological underpinnings have taken well over 
a decade to build.  The addressing scheme has been worked out for ages; 
the "experimental" deployment networks for IPv6 have now been shut down. 
People are very successfully using IPv6 in small-scale networks - think 
building-level, just-above-small-business level - where they have a 
relatively strong network staff and good gateways set up between IPv4 
(outside) and IPv6 (inside).

For new corporate deployments, IPv6 is starting to make a whole lot of 
sense.  [In a lot of cases, ipv4->ipv6 gateways happen to also be 
corporate border firewalls or proxies.] For home - not so much, yet. 
Earthlink and others are trying to encourage home users along the path 
with the development of things like their custom firmware for Linksys 
802.11g access points.  ;)


> Another way we risk this shift away from the Internet is in the calls 
> for building more "Intelligence" into the network (sometimes with 
> 'security' being the driving wedge).  Reading David Isenberg's work on 
> the "Rise of the Stupid Network" we see the virtues of the TCP/IP based 
> networks as one where the intelligence of the network is by design at 
> the edges.

Most of the very oldest services surviving on the network - as well as 
some of the newer ones that are gaining what will be difficult-to-dislodge 
popularity - take intelligence-at-edge very seriously.

Oldsters:  mail, usenet.  [~35 and ~26 year old services, respectively...]
 		shell connections.

Middle-aged services:  http [~13 year old - just coming into its teen 
years]

Young services not doing so well:  "walled garden" instant messaging 
programs where a "smart" server is controlled by some corporation.  AIM, 
MSN, Yahoo! IM, etc all fall into this camp.

[I would be very, very impressed if any of those three corporations were 
to commit to providing their IM service in perpetuity forever.  I just 
don't see it happening.  Feel free to correct me, vendor representatives 
;) ]

Young services doing really well:  Bittorrent, Jabber-based federated IM 
[google talk et al..], a few of the other p2p applications.

One of the marks of a relatively successful internet service - I think - 
is that service's resistance to being destroyed or completely disrupted by 
acts of god or acts of congress.

There's not a whole heck of a lot that can be done to disable a service 
whose spread only requires a couple of operators to exchange IP addresses 
in order to federate / communicate with each other's servers.

The whole DNS could implode, and folks would fairly quickly replace it. 
There's just not enough "special magic" for ICANN or any other body to 
completely wreck things.  Maybe disrupt for a while -- c.f. the current 
court battle over spamhaus -- but the replacement engineered would likely 
be much less prone to a repeat of the same attack.

[Hey, even if the central DNS roots went away -- enough of the data is 
cached *all over the place* that things would not immediately go awry.


> I'd also point to conceptual work of Garth Graham - taking the TCP/IP as 
> a form of social contract... where we are all peers.  Imagine a 
> reconceptualization of politics in such a frame... I know a number of us 
> are.


It used to be that the social contract was that people behave relatively 
sanely in order to get Jon Postel (in his IANA guise) to allocate IP 
addresses for them.  I find that I think those days were superior to the 
current ICANN boondoggles.


--elijah