Let me be precise about what this page argues, because precision is the whole point. Robert Kahn, Vint Cerf, and the generation of architects around them built the internet’s protocols. TCP/IP is a monumental achievement and I have never claimed otherwise – my entire thesis is that I did not invent the features; I delivered the collective benefits. This page argues the mirror image of that sentence: the people who invented the features did not deliver the benefits. Specifically, they did not invent eCommerce, and they did not make it. I choose those verbs deliberately. “Enabled” is a word anyone upstream can claim – the man who paved the road can say he enabled your restaurant. Invented and made are verbs with receipts attached. eCommerce was never their project, it was never their network’s purpose, and their own written history confirms it.

Exhibit: Their Own History

In 1997 the internet’s founding generation wrote its official autobiography: “A Brief History of the Internet”– nine authors: Barry Leiner, Robert Kahn, Jon Postel, David Clark, Daniel Lynch, Stephen Wolff, Vinton Cerf, Leonard Kleinrock, and Larry Roberts. The inventors, writing their own story, choosing their own emphasis.

Search the document yourself. The internet’s architects wrote the history of their own achievement, and in it, commerce – the thing the internet is now most famous for enabling – does not appear. That is not an oversight. It is an honest reflection of what they built and what they didn’t.

The Internet Has Lanes

Here is the root of the confusion, and it deserves its own section. When the public hears “father of the internet,” they picture the internet they actually use: click here to buy, type a card number, stream the movie, book the flight. But that internet is not TCP/IP. TCP/IP is a protocol: one lane of a many-lane industry. The internet people know required the protocol lane, yes, and also the telecom lane (circuits, carriers, international private lines), the engineering lane (ATM, CBR, QoS), the productization lane (turning capability into something a customer signs for), the finance and venture capital lane (the checks that build physical networks), the customer acquisition lane (Cisco does not sign because a protocol exists), the provisioning lane (someone orders, installs, and lights the hardware), and the commerce lane (the storefronts and transactions that made any of it matter to ordinary people). Every one of those lanes had to be worked, by people, with capital at risk, before “the internet” meant what it means today.

Bob Kahn and Vint Cerf worked the protocol lane, and worked it brilliantly. They did not work the others, and their own history, quoted above, does not claim they did. The trouble is the title. “Father of the internet” quietly annexes every lane on the highway on behalf of one. The co-author of TCP/IP is the father of TCP/IP, and TCP/IP deserves its father. But the internet the world means when it says the word, the one with a checkout button, had many parents in many lanes, and I am from several of the others: eCommerce, telecom, ATM CBR engineering, productization, finance and venture capital, customer acquisition, provisioning, QoS. I have said this directly to Vint and to the ISOC and Internet Hall of Fame communities I correspond with, and I will say it here: honor the protocol lane fully, and stop letting one lane’s honorific write the other lanes out of history. This is not a diminishment. It is a map.

Why They Didn’t: It Was Never the Mission

There are three structural reasons the protocol generation neither invented nor made eCommerce, and none of them are failures of intelligence. They are facts about what the project was.

First, commerce was against the rules of their network. The internet the architects ran was a government-funded research and education network, and the NSFNET Acceptable Use Policy barred commercial traffic from its backbone until April 30, 1995. For the entire era in which these men ran the internet, running a store on it was against the rules of the network – rules their own institutions administered. You do not make what your own policy prohibits – you cannot even claim to have enabled it. eCommerce could not begin until their network was decommissioned and the backbone spun off to commercial carriers like MCI and Sprint. The starting gun for internet commerce was, literally, the end of their stewardship.

Second, their design goal was best-effort delivery, and commerce requires determinism.TCP/IP is a triumph of resilience: packets find a way, eventually. “Eventually” is a fine standard for email between universities. It is a disqualifying standard for a payment. A merchant needs to know that a transaction will complete, complete quickly, and complete the same way every time – across an ocean, at peak load, with money and liability on the line. Best-effort routing over congested public peering points cannot promise that, and nothing in the protocol stack ever did. Determinism doesn’t come from a specification. It comes from infrastructure: private circuits, controlled routes, guaranteed capacity. That had to be built, physically, by someone else.

Third, a protocol cannot carry trust. In 1995 I ran an eCommerce site – perfectwheels.com, and here is the exhibit – and I can tell you exactly what internet commerce looked like under the architects’ internet: the website was a catalog, and the actual transactions arrived by fax with full credit card numbers on the paper, or by phone, read aloud, punched into a terminal on my desk. Card-not-present, unencrypted, unverifiable – fraud dreamland. SSL existed as a feature in the browser. But a feature is not a benefit: nothing carried that encrypted transaction across borders with the speed and certainty a real merchant and a real bank required. The trust layer of global commerce was not a protocol problem. It was a delivery problem.

Who Made It

Making eCommerce took a different kind of work, done after April 1995, by operators, not authors. It took merchant transport: putting the transaction itself in the browser under SSL, and carrying it over international private leased circuits at constant bit rate – deterministic, sub-300-millisecond, border-crossing delivery that a merchant could bet a business on. That is what I invented – a concept forced on me by running a storefront on the 1995 internet – and then made real at global scale as a co-founder of Digital Island, whose Tier-0 network (a class I coined, and a network I helped architect and build) carried exactly those transactions for Cisco, Visa, MasterCard, E*TRADE, Schwab and the first generation of global eCommerce. The evidence for that story is documented across this site, with dates and contracts.

The division of history is clean, and both sides of it are documented in the participants’ own words. The architects built a network that could move packets anywhere, and wrote a history that ends in April 1995 with the defunding of their backbone. The globalization of commerce over that network – the benefits – had to be delivered by the people who came next.

Read the internet’s official history. Search it for “eCommerce,” “merchant,” “SSL,” “transaction.” They aren’t there. Their history ends in April 1995 with the defunding of the NSFNET backbone.

Mine begins there.

Source: Leiner, Kahn, Postel, Clark, Lynch, Wolff, Cerf, Kleinrock, Roberts, “A Brief History of the Internet”. See also: Perfect Wheels, 1995 – The Genesis of Merchant Transport.