Re: [BITS] ICS 230 Midterm: Social Construction of the Internet

Rohit Khare (rohit@godzilla.ICS.uci.edu)
Tue, 24 Feb 1998 03:45:56 -0800


> Hello Rohit -- I would guess several things from reading your mideterm.

Thanks for your reply -- I'm glad to take the debate a bit further. I
certainly stand guilty as charged on the factual matters discussed
herein. I can only claim in my defense that neatness seemed more
important than correctness at the time :-)

> I am not trying to discredit your writing. Just noting that it is
> indeed very difficult to impute history from recent memory.

I hope not! I'm still on a high from some of those turns of phrase. If
only I could pass the class on style points alone...

> 1. You have not read "Where Wisards Stay Up Late" by Matt Lyon and
> Katie Hafner, because they tell a drather different story.

Yep. I still haven't read it in detail. It's on the stack. Initial
sighting in the archive (and the FoRK brookstore!) at
http://xent.ics.uci.edu/FoRK-archive/summer96/0454.html

Katie Hafner, in her Amazon description of the book, highlighted
"collaborative work and simultaneous discoveries" throughout Internet
history, which is strong evidence of a 'high-commitment frame' in
social-constructivist-speak. Lots of people sharing the same ideas and
resources explore similar paths simultaneously. [I wonder if
'celluloid plastic' hackers would take similar exception to their
historical treatment! :-]

> 2. You did not live through it either, as my rememberences from
> living through it do not track with your story.

Well, that goes without saying -- which is where history takes over
from journalism. The magic of historiography is the conceit of
overlaying a compressed structure on messy events. In this case, the
contentious hypothesis I'll assume we're focusing on is:

* Was the Internet's technological development predictable?

And I think the warning of 'impending 20/20 hindsight' is entirely
appropriate. I *only* have hindsight to go on...

[Of course, I fully expect I could be proven wrong on this
hypothesis. It seemed like a fun position to take for a midterm.]

> For instance, did you know that IP was developed after TCP failed to
> work without IP? That was when Cerf and Kahn finally began to
> understand the fundamental issues with Internetworking by tunneling IP
> through multiple media types.

Personally? Yes, since that's why it's TCP/IP, after all. The circular
dependencies between their development belie the neat layer
diagrams. Fragmentation, encapsulation, and the end-to-end principle
were all learned iteratively through failed alternatives.

As it was represented in my essay? It did oversimplify, in order to
fit the arc into a single sentence. It *does* still seem reasonable to
view the Internet effort as first, a chance to build a common data
communications layer, and then as a common services layer. Your next
point seems to buttress that view:

> In the meantime, anxious to commercialize on the original ARPANET NCP
> protocols, X.25 was born without paying attention to the need to solve
> the Internetworking problems.

Now, the next observation you make seems like an *excellent* test of
the hypothesis. I think we could agree that in the traditional data
communications 'technological frame' (read: Bell-head) emphasizes
reliability, and hop-by-hop integrity.

> Remember that the original APRANET NCP included node-by-node
> reliability checks with retransmission between packet switching nodes,
> and delivered reliability to the HOST interfaces. This had to be
> abandoned to solve the problem with what Vint called the Catenet,
> which was just a network of tunneling gateways.

So here's the crux: is the 'discovery' of end-to-end (abandonment in
favor of reconstruction at the periphery) *inherent* in the 'Internet
frame' or accidental? I think I can be rightly blamed for circularity
in defining the end-to-end principle AS part of the Internet
frame. But within that perimeter, the path seems more fixed. The
emergence of security solutions for encrypting IP take the same
form. The application protocols like HTTP take the same form. Ideas
that don't -- intermediated email delivery and second-generation
routing of packet flows (which use state-in-routers) -- took longer to
really shake out and seem grudgingly accepted [the former, I'd say,
because of the reality of intermittent interconnectivity (we needed
mail relays) and the latter for intra-network allocation and billing].

> Etc, et al...

(hmm. are you implying I overlooked the same events or the same
people? :-)

> And, the ethos of the internet stems from the original work of Paul
> Baran in 1962, with his "Route Around Damage" concept for surviable
> packet switching, though the path of transfer of his work into the
> ARPANET was quite indirect. Over time. "route around damage" has
> transmogrified itself into "Work Around Problems".

I'm interested in two parts of what you said. The first is the
'nuclear warfighting' part of the story, which Hafner,
et. al. particularly highlight. I've felt the connection might be more
tenuous, since I didn't think the ARPAnet grants were expressly funded
for that purpose, that ARPAnet technology was used for that purpose
(i.e. MILnet wasn't *that* sensitive), and that high reliability in
the face of concerted attack was a community priority.

[If it *were* directly related, though, I can imagine some fun
shouting matches about 'patriotism' in support of IP: "if it's good
enough for our nukes, only Commies must like OSI!" :-]

Second, route-around-damage is indeed an organizing principle of the
Internet frame, from the wires-and-packets level to the social ("The
Net interprets censorship as damage and routes around it", John
Gilmore, NYT, 1/15/96). Now, another test of my bald assertion would
be to trace the evidence whether this was a founding principle, or an
innovation "from left field"?

> All this is now embeded in the IP net, in EMail store&forward,
> including all those acursed Gateways into non-Internet environments
> whic did not use the IP tunnelling model to deliver undamaged goods
> end-to-end. But, good old Internet working around problems had by
> then infused the core of developers to just make things work anyway.

Aha! A socially constructed technological pathway in action! We should
invite *you* over for a day in John King's class... I think this has
been a very legitimate inquiry into how believable the pat histories
in our textbooks really are in the face of events.

> Cheers...\Stef

Thanks -- I appreciate a well-armed gentleman/gentlewoman as a
sparring partner!

Rohit Khare
http://www.c2.com/cgi/wiki?RohitKhare

---
In that old Baudrillardian challenge, if one stages a bank robbery
and carries the performance all the way to the bank, fooling even the
teller and guards, one arguably has succeeded in redefining oneself
not as an actor, but as a bank robber.                -- www.suck.com