Now that the IPO market and Venture Capital are back from the dead, it’s always interesting to read the box score updates for your favorite teams in the uplifting exits many have enjoyed of late. My colleagues here at NEA have certainly posted our share of points on the board and have rightfully earned some very nice ink in the trades. Note that all of these scores are measured in "Dollars." There is absolutely zero wrong with that; it's why our LPs participate and every VC wants to post The Big Score.
But there are other meaningful ways to measure a VC firm’s impact. Just measuring dollar returns fails to value contributions made to The Innovation Business with longer, deeper impact than a spectacular return. Although essentially unknowable at the time of the investment, if you are very lucky, some deals change future history dramatically. It is worth reflecting upon them—we can better understand where we are and where we’re going by better understanding how we got here.
To this end, I'd like to offer my nominee for "The Most Consequential Tech Investment" ever made by NEA. “Consequential" attempts to capture the notion of “Big Score” downstream impact not readily denominated in Dollars.
My nominee is the funding of Bob Metcalfe and 3Com to commercialize Ethernet technology.
This is the greatest jail-break story since Steve McQueen rode his motorcycle over the fence in "The Great Escape." Given the unfortunate history of Xerox PARC, it is pretty easy to imagine a dramatized mini-series version of the scene:
Metcalfe and Ethernet, arm in arm like Butch and Sundance, scrambling over the wall at Xerox PARC, scuttling away into the darkness and going feral.
They spent some tough time in the wilderness making friends, surviving (and starting) a few epic bar brawls, all while getting bruised, toughened and buff. Eventually, with the combination of elegant simplicity and low cost, Ethernet started devouring its more complex challengers. In the process, Metcalfe and Ethernet completely rewrote the future history of computing and enabled The Global Internet as we know it today. Without this jail-break, the world would be very different indeed. Let's take a look at how.
Back when my hair was still its original color, even small computers were big and expensive. In the world model that underpinned early computer networking efforts, computers came in largely two flavors: "host" computers which provided resources and services to users and "user" computers to provide access to those resources. That distinction was much more subtle back then. Today we use the terms “Web Server” and “iPhone.”
The first computer networks interconnecting host and user computers were quite expensive to build because they were constructed of yet more computers! The approach of using dedicated packet switching computers was an inspired engineering decision when it was first made given all the other challenges of building the first computer networks. The cost of this approach made it clear pretty quickly that networks based on the ARPAnet model would be very exclusive clubs.
As a result, Computer Networks of the early days were not something just any Fortune 500 company could own. They were considered "strategic assets," the province of governmental authorities like DARPA and the European PTTs. Computer networks were scarce enough that the very concept of connecting them together was novel. Nothing reflected this assumed scarcity of networks more than the fact the original Internet Protocol address format provided for only 250 networks – about the number of countries in the world at the time. Each one of those networks could contain 16 million host computers, but the built-in assumption was that the world would have small number of networks, each with a lot of computers.
At Xerox PARC, Ethernet was born as part of a grand experiment which assumed "computing plenty" instead of "computing scarcity" and centered on groups of users – “workgroups.” This was clearly different from the prevailing hardware-centric models of Timesharing and Batch Mainframes. A fundamental consequence of this new model was that a lot more computers would need network access, and that in turn demanded much less expensive computer networking technology. The "use a computer to network another computer" model of ARPAnet just would not fly. No, the network needed to be very inexpensive, if not downright cheap. The technology developed came to be known as Ethernet. As for the origin of the name, it is a pitch-perfect expression of die Gestalt von Ethernet, a reference to the “luminiferous Æther” long thought to be the pervasive medium which carried electromagnetic waves. (The famous Michelson-Morley experiment disproved the existence of Æther, but Ethernet remains a pervasive medium for data packets.)
With Ethernet, the network medium itself degenerated into a simple piece of coaxial cable borrowed wholesale from the extremely cost-sensitive cable TV world. All the interface electronics required to attach a computer to that cable got squeezed onto a single, (eventually) inexpensive circuit board that found its way into desktop PCs and even onto the PC motherboard. And the result of this technological upheaval?
Suddenly EVERYBODY has a network and they are EVERYWHERE!
Instead of the "Central Services" model with a few million computers on each of a few hundred networks, each operated by a duly-certified franchisee of The Sanctified Brethren, that model of the Internet was flipped on its head to become: a few tens of computers on each of a few million networks, with each network owned by the person who spent his own money to build his network. The Global Internet is the aggregation of all these networks – it is not just a “thing,” it is a “meta-thing,” a description of a collective like “an exultation of larks” or “a truss of architects.”
The willful aggregation of its constituent threads makes The Global Internet the largest collaborative project in the history of the world. Think about that for a moment: the largest collaborative project in the history of the world. Not a bad bit of work for a bunch of rabid visionaries willing to take on industrial behemoths armed with "yellow snake" Ethernet cable and a protocol stack that escaped over the wall in the dark of night.
Ethernet created an explosion of networks, with each one wanting connectivity to its siblings. The necessary transmutation of the original Internet Address to deal with this explosion is precisely why “ownership” of the Internet can be so incredibly diffuse. The most fundamental point to understand about The Global Internet is that it works entirely because people want it to, pure and simple. The Global Internet exists because all the people around the world willed it into existence by building their piece and then attaching it to the whole.
Mercifully, there is no Global Department of Internet Impediment, not at the ITU, the UN, or even at the Department of Helping Too Much. Let us hope this continues unimproved.
MILNET was created as an operational "Son-of-ARPAnet" and was to be the first network to speak only IPv4. Not long before it was due to start-up, the explosive implications of Ethernet in the marketplace were recognized by the good folks at the US Defense Communications Agency (DCA) and the brakes were applied so something could be done to deal with it. A decision was taken that the IP address encoding be revised to provide for a small number of big networks and a big number of small networks. Ethernet had just demonstrated the depth and breadth of its influence - DCA decided to make such an important change at a very late date because the Ethernet-driven model of "computing richness" would profoundly change the rules of the game.
Today, one is hard-pressed to find a computer that does not support Ethernet connectivity. Along the way, Ethernet adopted new physical media types, upped the performance dramatically, learned to talk and listen at the same time, and with the advent of WiFi, abandoned “wires” altogether. Myriad details have changed in the way the packets traverse the various underlying physical media, but at its very core, the essential semantic promises made by Ethernet and the Ethernet packet format have remained the same.
In retrospect, Ethernet was the catalyst for a transformation of computing, from connectivity as a centrally-managed scarcity to a pervasive given. That, in turn, was a major force which helped propel The Internet into its role as the universal growth medium for Innovation on Biological Scale.
But that's another post.