[e2e] It's all my fault

John Day day at std.com
Thu May 17 18:56:50 PDT 2007

At 18:33 -0400 2007/05/17, Tom Vest wrote:
>On May 17, 2007, at 4:58 PM, Vadim Antonov wrote:
>>On Thu, 17 May 2007, Greg Skinner wrote:
>>>DARPA-funded research provided computing resources upon which email,
>>>USENET, Unix, etc. were extended and popularized.
>>DARPA didn't create those resources from the thin air. They took it from
>>somebody first. Your statement is an example of "What is seen and what is
>>not seen" fallacy.
>Okay, what is seen:
>The universe of telecom facilities was owned and operated through 
>the vehicle of adjacent, non-overlapping territorial monopolies for 
>at least 4-5 decades leading up to the 1970s -- either as the result 
>of a market outcome (e.g., in the US), or of subsequent movers 
>observing how things played out in the earliest telecom markets 
>(e.g., in the US). That said, the raw inputs required for the 
>Internet to emerge (telecom facilities, technology, clever people, 
>etc.) were widely distributed throughout the world in the 1970s and 
>1980s. The same laws of physics that permitted T-carrier technology 
>and 4ESS switches to work in the United States from 1975 on also 
>applied everywhere else on Earth.
>In the US alone, the advent of of the latter technologies was 
>accompanied by regulatory changes (60 FCC 2D / 1976, which compelled 
>the incumbent territorial facilities monopoly owner to sell T-1

I would dispute this.  The technologies were not developed in the US 
alone.  The US may have been the one place where quite by accident 
there was an organization willing provide a place for it and grossly 
overprovision the technology so that its flaws were not immediately 
apparent.  The US may have also had sufficient domestic facilities 
for the technology to reach a critical mass.  But all the smart 
people were not in the US.

>circuits to 3rd parties even when those parties intended to use them 
>for commercial purposes) which made it possible for someone other 
>than an incumbent facilities owner to provision telecom 
>"infrastructure", manage it independently, and use it for any 
>purpose that they saw fit. In the US alone, the incumbent facilities 
>owner's efforts to squelch the "invidious bypass" that this new 
>technology made possible (i.e., the Consumer Communications Reform 
>Act of 1978, aka "the Bell Bill") were quashed.

Be careful here.  This is a pretty sugar coated version of what 
happened.  None of this would have happened if ARPA, NPL, and IRIA 
had not proven that the technology worked by 1972 and there was 
another way to do networks.  This lead every major phone company to 
show that they could do it too.  So that by 1977, there were several 
commercial packet networks in the world.

Had they been left to their own devices the phone companies would 
never have built a packet network.  When they were they still would 
not have built one friendly to computers.  That was forced on them by 
the computer companies.

The computer companies saw this as a way to go after market that the 
phone companies were ill suited to pursue and the phone companies 
very quickly saw that the organization that was being adopted 
relegated them to a commodity business.

>Eventually -- sometimes many years later -- other regulatory 
>jurisdictions followed a similar path, and the Internet started 
>growing in those places as well. Eventually -- generally decades 
>later -- in places where such changes never occur(red), some green 
>field bypass telecom facilities (e.g., wireless) began to provide 
>(at the moment, grossly inferior) options similar to those created 
>by the aforementioned regulatory interventions. In the mean time, 
>much of the Internet service that was available in the 
>"unreconstructed" parts of the word arrived in the form of " service 
>imports", i.e., services provided by offshore operators based in one 
>of the infrastructure-friendly jurisdictions.
>Something else that is seen:
>Empirical evidence supporting the story above is visible in the 
>global distribution of autonomous systems (using the country code 
>and/or org fields to localize each to a particular country). Since 
>ASes are tools for managing multihoming, and multihoming is only 
>technically possible where telecom facilities are overlapping, or 
>fungible (i.e., available in fractional bits and pieces as 
>"infrastructure" that can be managed independently from the 
>facilities provider), and available on commercially reasonable 
>terms, this distribution makes perfect technical sense. Places with 
>more ASes generally have more Internet users, devices etc., all 
>things (population, GDP, geography, number of years providing 
>Internet service, etc.) remaining equal.
>What is unseen?

I am afraid that you have fallen into a trap that historians are 
quite familiar with. Looking at the events and seeing the sequence as 
almost foreordained as if it was a very deterministic sequence that 
could not have turned out any other way.  I am afraid that a closer 
reading of the events will reveal that much of it is quite 
accidental.  And the inflection points sometimes hinge on very small 
seemingly unimportant events.  What you see is more the result of a 
random walk than the deux ex machina of history.

Take care,

More information about the end2end-interest mailing list