Software Designers~The People Behind the Code~(英語)

#30Elizabeth “Jake” Feinler: Information Scientist; Internet Pioneer

In 1960, a young woman from West Virginia took a job at the Stanford Research Institute, and took a front seat in the history of the Internet. Elizabeth ⁠Jake⁠⁠ Feinler’s background was in biochemistry: she had worked as an editor for Chemical Abstracts Service, which was then trying to put the world’s chemistry literature online. While Feinler wasn’t a part of that automation project, she was fascinated with the idea of computer science as a resource, not just for number crunching, but for information. When she heard about a comparable job at SRI, she applied.


At SRI, Feinler found herself trying to organize a mountain of information using three-by-five inch cards. Figuring that some computer power would help, she wandered up to Doug Engelbart’s Augmentation Research Center. What she found was a group of people sitting around low tables, looking at what seemed to be television sets, and maneuvering a little device they called a mouse. Engelbart told her: ⁠You’re all going to be doing this someday,⁠⁠ and, with a bit of persuading, he asked Feinler to join his group. Her first mission: compile directories of the sites and people who were on the Defense Department’s packet-switched computer network--the ARPANET--the direct predecessor of the Internet. She worked on Engelbart’s NLS/Augment system that, in 1970, already had many of the features now common on today’s machines.

Feinler describes her work as creating a ⁠prehistoric Google,⁠⁠ an online hub where people could come to exchange information about their sites. She ultimately became the director of SRI’s Network Information Systems Center, which operated the Network Information Center, or NIC, for the ARPANET and the Defense Data Network (DDN)--administering the network’s domain names and IP addresses.

You had moved to the Silicon Valley at a famously turbulent time. What was it like?

For me, it was quite a shock. This was California in the 1970s, and this was a very far out group. San Francisco was in its flower child age, and everybody was changing their way of doing things, from the way they looked to the way they thought. The rest of the Institute hadn’t quite caught up with Engelbart’s group yet, so they stood out. Most of the men had beards and wore Birkenstocks, and the women had thrown away their bras. Sometimes they sat on the floor. I had come from a corporate culture of back-combed hair, high heels and business suits, so it was indeed a culture shock for me.

Were you in cubicles?

The age of cubicles hadn’t yet arrived. Engelbart’s group had a wing, with real offices with windows and doors, some of which we shared. But equipment was very expensive: a workstation in those days was well over $100,000. So the workstations were out in an open area, called the ⁠bear pen,⁠⁠ surrounded by the offices. At first we took turns, but eventually, everybody got their own workstation, which was usually in their office.

How was the work structured?

Almost everything on Internet projects was done by teams. DARPA would define a project, then they would look for groups that could carry out that project. Sometimes these groups worked together, sometimes there was a friendly competition. A working group might have people from many different sites around the net--and anybody’s idea was up for grabs. The work was described in something called the RFCs--requests for comments. RFCs were then, and still are, the technical notes of the Internet, and serve as a way of building consensus--though the process is not quite as democratic as it used to be.

I started out working on the NIC project, then I became the principal investigator. In 1977, Engelbart’s group left SRI to become a commercial enterprise. I wanted to keep working on the Internet and wanted to stay at SRI, so we went our separate ways. By this time the NIC project was a separate project anyway--and it stayed at SRI.

You’ve been credited with coming up with the now familiar top-level domain names like .com and .edu.

Our team did that. The concept of generic names was essentially mine to begin with, but we always discussed these things as a group before we took them to the client. The NIC was the naming authority for the network from the beginning--probably because nobody else wanted the job. It was a painstaking but important function: if you didn’t have the right address on the list, somebody’s email was going to bomb.

What happened was that the list of names kept getting bigger and bigger until we finally saw that having a centralized list was not that great an idea. There wasn’t a lot of bandwidth then, or a lot of space on computers. The list got so big that the small computers would pick and choose who they ⁠thought⁠⁠ a sender wanted to reach--which was not a very good idea. Then the naming got too big for the space they had allotted in the headers. That’s when they decided to go with the domain name system, which was decentralized and hierarchical.

To test that out, the Network Working Group, DARPA, and the Defense Communication Agency, chose .arpa as a test case--appending ⁠.arpa⁠⁠ to every host name on the Internet. It was just a space holder, a test, but it still caused a bit of fuss because many people using the net had nothing to do with DARPA. So the Defense Communications Agency considered adding a broader term of ⁠.mil⁠⁠ to everybody’s name, because all the host computers were on a military network. But that would not have been very popular either, because a lot of users were at universities or non-military facilities and did not consider themselves military. Meanwhile, we were getting called by big companies who insisted on being assigned a top level domain. We told them we hadn’t come up with a system yet.

I thought a set of generic categories might work better. The University of Southern California Information Sciences Institute, or USC/ISI, developed the technology for the domain naming system. The NIC used this technology to implement the official name server and came up with the top level domain naming scheme. That’s how it all began. We first chose ⁠.mil⁠, ⁠.edu⁠, ⁠.org⁠, and ⁠.gov⁠. Commercial sites were not allowed on the network unless they were government contractors or approved by the government. So the question arose: should commercial sites be associated with the domain of the agency that was sponsoring or funding them, or should they have their own domain? We finally decided to add one more TLD: .com, for commercial.

So “.com”--probably the most common, sought-after domain--was an afterthought?

It was truly an afterthought, because you were not allowed to be a commercial organization on the network in those days unless you were a government contractor or sponsored by the government. Who knew that naming and addressing would become a multimillion dollar industry? Now, ICANN, the standards body for naming and addressing, has opened up the top level domain names to anybody that wants to spend $185,000 to buy one. I think this will finally take the focus off of what kind of group an organization is.

You were one of the first people to experience the Internet. Could you see its broader potential back in 1968?

I’m an information person. When I saw what was happening on the Net with text-editing, document production, hypertext, email and file transfer, I was in heaven--and quickly became a devoted user. At first, a lot of it didn’t work very well, and that was very frustrating. In the early days, the developers just wanted a proof of concept. They didn’t get too involved in all the details of making things really work. Even so, it was incredible―you can imagine what having email was like back then. To send a message, we used to have a secretary take dictation, then type up a letter and mail it. Now, it was instant. This was a very insular world: the outside world wasn’t doing any of this yet. And if you tried to explain it to somebody who wasn’t involved, they just looked at you like you were nuts.

How did the early Internet start to catch on?

The ARPANET grew rapidly, especially at universities where many of the researchers were located, even though they were designing the network for military clients. At the universities, undergrads, graduate students and professors all got involved. Soon other people began looking over their shoulders, and they were impressed. So the network began to spread. Many host computers had users on them that weren’t part of the military internet project, but they began trying out this new technology anyway. It was the best game in town and every techie wanted to play. Soon similar non-military networks like BITnet and uucpNet sprang up as well.These were the ⁠poor person’s⁠⁠ ARPANET, but they gave people, especially students, a similar way to network with friends and colleagues. Then the National Science Foundation started the NSFNET, which was a government sponsored network for students and educators. That opened the doors even wider. Finally, the administrators allowed the gateways to handle commercial traffic, the Web arrived, and all hell broke loose.

Eventually you left SRI and went to NASA

I left SRI in 1989. I was worn out. The problem with the NIC was that it never stopped. We handled hundreds of emails a day, and some were always falling by the wayside. I’d think: oh dear, did I miss something that was really important? There was lots of travel to meetings and working groups. I was also pretty sure there would be a bidding competition for the project, and it would probably go to another bidder. I didn’t have the heart to be the one to let all my people go. That would have killed me--they were like family. So I decided to take a year off. SRI was being reorganized, the project was solid, and there were good people to take over. So it was a good time to leave.

I was hanging out at home, and had planned to go to South America and be a visiting researcher. Instead, my step-father died of a heart attack, which left my mother alone, and she was blind. I spent most of that year dealing with those problems. At a local meeting, I ran into some people I knew at NASA Ames Research Center in Mountain View, and they asked me to come to work there. So I did. I stayed there for about five years.

You gave your papers to the Computer History Museum.

I now volunteer there, and I gave them 350 boxes worth--NIC papers and others describing Engelbart’s activities and the early network. We collected stuff for years. Part of the reason for so much hardcopy was that when people called in with questions, the network wasn’t fast enough to access the answer online. So we kept much information on paper for quick access. Also, the NIC was a repository for hardcopy reports from all the sites around the Internet.

It was quite an extensive collection. When Engelbart left in 1977, he left a large bank of filing cabinets that went down half the length of the wing--and SRI wanted this space for other things. The janitors came up at night and took all the papers out of the filing cabinet and dumped them in a pile and hauled off the filing cabinets. They were about to take all the papers to the dump when I happened to come in. I knew there were important things there that chronicled the history of the Internet, so I called the president of the Institute in the middle of the night. He wasn’t too happy about that, but said I could have all the papers if I removed them quickly. Everything got stuffed into boxes and went to my garage. After the 1989 earthquake hit, everything got dumped on the floor again. The collection was quite a mess when it got to the museum, but over time we got it organized and boxed. Now we are looking for funds to scan the important documents.

What is your view of the Internet these days?

It’s an amazing tool. The size and impact of it is incredible, as are some of the more unsavory things that are happening on it. The Internet is like electricity. It started with one light bulb--or in our case--two hosts, UCLA and SRI.And now the whole world is lit.