Entretien avec Tom Truscott sur Usenet
As Duke University graduate students, Tom Truscott and Jim Ellis conceived of Usenet as a replacement for announcement software at the university. Steve Bellovin was interested in the project and wrote the first UNIX-based news software, a precursor to "A News". The software was installed at Duke and the University of North Carolina, creating the original Usenet network.
Tom now works for SAS, an analytics software firm.
Interview (4/18/2007) with Tom Truscott
1. Where are you currently employed? What is your current role? What are you currently working on?
I am a software developer at SAS Institute, a "Business Intelligence" software vendor particularly strong on data analysis. I'm currently working on tools to analyze our software for performance and correctness. That may sound boring, but we have so much software that it actually pretty interesting.
2. Besides Jim Ellis and Steve Bellovin, who else did you work with on early Usenet related projects? What were their roles?
Stephen Daniel wrote "A news" which was the first distributed implementation of Usenet. He also created the dotted newsgroup naming structure.
Dennis Rockwell was a Unix enthusiast at Duke who ported uucp (Unix to Unix Copy Program) to memory-starved machines so they too could participate in Usenet. His "phs" machine was the third member of Usenet, after "duke and "unc".
4. Were the early Usenet projects part of your graduate work or personal projects?
Usenet was an unofficial project done in our spare time. Spare time (highly recommended!) is limited so we looked for the simplest thing that was good enough. Simpler things are also easier to fix, and if one is lucky they might even seem "elegant".
5. In a usenet.hist mailing list discussion, you mention the "technological matrix" that made Usenet possible. Could you elaborate on that concept?
At each point in time, society has a collection of knowledge and technology. Pieces of that can be assembled to make a new discovery or invention. It would have been "impossible" for us to land a man on the moon in the 1950s. But then the time comes.
George Walford calls this "steam engine time": when it is time to invent steam engines, someone will invent them. This seems implausible (especially to the inventor!), but a web search on "independent and nearly simultaneous invention" finds references to Calculus (Leibniz and Newton), tool grinding and polishing (anonymous, circa 3000 BC), the electrolytic extraction of aluminum, and far more.
Sometimes the pieces are hard to assemble. I'm amazed we were able to land on the moon in the 1960s. Other times the pieces are just lying there waiting. In 1993 the Internet, desktop computers, bit-mapped displays, and the idea of hypertext had been around for years. Tim Berners-Lee assembled the pieces, the good folks at NCSA fashioned a Mosaic, and the WWW changed the world. (I remember when I first saw a web browser, I felt like an idiot!)
The pieces for Usenet in late 1979 were call-in modems (300 baud in those days), dial-out modems (quite expensive, but fortunately we had already built a couple ourselves), software to copy data between computers (uucp), the desire to stay in touch with colleagues at other schools, and the "idea" of a local bulletin-board program. We used to have a local bulletin-board program but it wasn't very good and had stopped working due to a recent operating system upgrade. We only had to notice that the pieces could be made to fit together, and then be willing to do the necessary work.
6. What form of communication did you use to collaborate on the development of A News?
Initially we would all meet in a room (Duke and UNC are only 10 miles apart), but after issues had been settled we mostly used uucp's email software to keep in touch.
7. What problem were you trying to solve by developing A News / Usenet?
We felt alone and isolated from other computer science departments. Personal contact was no problem, via postal letter or an expensive phone call. But for general announcements or queries we were limited to journal articles, annual conferences, and newsletters. Newsletters tended to arrive unpredictably and were a burden on the newsletter editor. Usenet was largely modeled as a distributed newsletter without a single point of failure. A typical newsletter had ten or twenty items per month, and we naively modeled that too.
When a system joined to Usenet we could also then send email, which was another plus.
8. What were your main technical challenges when writing A News?
Since communication was via long-distance telephone calls (expensive back then), we needed to minimize phone time. Each article was given a unique name and we (mostly) avoided redundant transmission. Still, we worried about accidentally (or maliciously) racking up large phone bills, since that would abruptly cool interest in Usenet.
We didn't know what topics people would discuss, so we needed to permit arbitrary topic naming. That led to the dotted newsgroup notation. The two initial newsgroups seemed plenty at the time:
general - General local discussions
NET.general - General netwide discussions
9. When you were developing A News did you expect Usenet would grow to a global scale?
We hoped Usenet would eventually connect most computer science departments in the US and Canada, but knew that transcontinental telephone costs would preclude wider access. We were surprised when the University of Sydney, Australia, begin airmailing magtapes to and from UC San Diego. We were even more surprised that companies such as DEC (now part of Hewlett-Packard) would absorb US$10,000.00 monthly phone bills for the greater good. (I should mention, this was back when $10,000.00 was a lot of money.)
10. What benefits were you expecting Usenet to bring to the internet? Did it meet your expectations?
I wasn't expecting any particular benefits to the Internet (called the ARPANET back then). Most benefits flowed the other way. Some high-quality mailing list digests such as "human nets" were gatewayed onto Usenet. As the Internet spread the telephone bills plummeted, which was a huge benefit.
11. What benefits were you expecting Usenet to provide to your professional or academic life?
I think the only benefit we were expecting was that, by announcing this interesting thing and giving away the software to get it going, we would attain some modest level of fame among our Unix colleagues. It far exceeded our expectations, thanks to the hard work of many.
12. How did you feel Usenet would evolve with the advent of the World Wide Web?
I predicted its imminent death, to be replaced by a better service that exploits the high-speed interactive capabilities of the Web. But it appears instead to have evolved in that direction.
13. Do you actively participate in any newsgroups? Which ones?
Each month I post one or two articles, which seems plenty to me. A few years ago I got carried away in the comp.protocols.tcp-ip newsgroup battling defenders of the clever-but-obsolete Nagle Algorithm. Strange but true.