Monday, March 16, 2015

Oldest Registered .COM Domains and where they are now

Today (for a few more minutes), is the 30th anniversary of the creation of the first Internet domain. A number of years ago, while I was at Sun Microsystems, I wrote a related blog. Unfortunately, Oracle shut down all of the Sun blogs (they would be fun reading now).  But thanks to, I've managed to recover my original posting and am reposting it here.

A number of people (Out of the Woodwork, Sun tied with IBM for 11th oldest .com domain name, and The 100 Oldest Registered .COMs) have already noted this web site with the 100 oldest currently registered .COM domains and noted that Sun was one of the first registered web sites, registered at the same time as IBM and before Intel and years before Cisco and Microsoft. What got my attention was the names the top of the list. The oldest registered .com domain, one full year before Sun and IBM was Symbolics, followed by BBN and Think.
Those names really take me back. BBN, of course, was one of the early contractors on the ARPANet. As a student I remember the excitement when the first "network interface" were rolled into MIT. That's one in the picture! Technology has really improved over the years. So it is not at all surprising that they were one of the very first to register a domain.
But its the and domains that brought back the most memories. Symbolics was the Lisp Machine company that spun off from MIT. They built $80K workstations that were heavily used by the Artificial Intelligence community, but also by researchers in vision and in the animation industry at the time. (They were about the same size and price as that BBN IMP we used as our Arpanet interface.) I remember being awed by the "render farm" of Symbolics machines at Pixar on a customer visit to Pixar, which was one of Symbolics' larger customers. Symbolics did their own animated film, Stanley and Stella in Breaking the Ice, by using the rendering jobs to burn in machines coming off the manufacturing line.
Symbolics' (Lisp implementation) of TCP/IP was the reference implementation when it came out. It was the first implemented and due to the debugging and development tools on the Lisp Machine, people had the most confidence in it in those early days. I remember discussions at Symbolics about Sun; how they didn't have any interesting technology, how buggy their networking code was, and how they could never survive. I believe Symbolics had a market cap about twice Sun's when it went public. At that time Symbolics was taping out their first custom microprocessor (Ivory) and was in development of of their second version (Sapphire), while Sun was just using commodity Motorola 68K's. And of course, Symbolics had the very advanced operating system developed for the Lisp Machine, while Sun was using Unix, which was just an inferior rehash of Multics. How wrong we were!
Somewhere in there, a grad student at MIT and I were involved in an evaluation of Symbolics' Lisp machine implementation and TI's (which also had a lot custom silicon). The two of us had a great time eating barbecue at the County Line in Austin and fantasizing about the super computers we could build by combining hundreds of these machines.
Well, that never happened, but the grad student was pretty successful. He did design some super computers (using Symbolics machines) at Thinking Machines ( before they also went out of business and he (Greg Papadopoulos, you probably guessed) came to Sun along with his friends Dave Douglas, Steve Heller and others.
A year ago I also joined Sun (and work for Greg). I think we all still believe in those same fantasies we shared overlooking the river Austin. You see glimpses of it in Greg's discussions about global computers and my ramblings on utility computing. The next few years are really going to be fun!
It's amazing how the circle closes.