Usenix was a relatively small and informal organization in the eighties. Its members wanted to launch Usenet in a Usenix meeting in 1980. Those were the days when the organization held its meetings at universities instead of big conference halls and meeting rooms of plush hotels. The meeting was held in Boulder. While Steven Bellovin was not present in the meeting, it was attended by Jim Ellis and Tom Truscott.
Bellovin felt that apart from announcing about Usenet, they required non-experimental code while his prototype was not adequate to succeed. However, he does not exactly remember the exact deficiencies present in his C version. At the same time, he thinks that one of them could be the code’s inability to configure, which neighboring websites would get what newsgroups.
Issues while declaring Usenet
Incidentally, Stephen Daniel came up with the code, which was referred to as “A-news”. A crucial alteration was the feature to exist with multiple hierarchies instead of simply having the original “NET.” or “NET”. The production version also had the possibility to configure which hierarchies or groups a website would get.
The members believed that the configuration should be in a file instead of being included in an array within the code. This particular point was not always considered. For instance, UUCP had an array for listing the commands that remote websites were allowed to execute.
To execute the rnews command, a system administrator had to alter the source code before recompiling. While looking back, this procedure looked like a wrong decision. However, in those days, some argued that it was justifiable to do so. After all, there are very few commands at that time.
Sorting the problems
The group decided to sort out this problem and came up with a mail-to-rnews program. It means a sending website would be able to email articles instead of directly trying to execute rnews. A clock-driven system was designed for retrieving the email messages and then transferring them to rnews.
The system had to be clock-driven as there was no other procedure available in those days to get the email directly delivered to a file or a program. As such, the A-news’ remote site configuration file is required to have a command for the execution.
The group wanted the first articles to cover issues such as trouble reports, general requests for help, and bug fixes. At that time, there was a strong positive culture to offer mutual assistance among developers not only in companies like Usenix but also in the world of the IBM mainframe.
Another proposal was to locate source code that appeared interesting without flooding them to the network. The reason for that was the chances of software being bulky and phone calls at that time were costly. In those times, the phone rates at nighttime were around 0.50 USD for 3 minutes.
It was also a time when at 30 bytes/per second or 300 bps, one could transmit a maximum of 5400 bytes. If protocol overhead had to be taken into account, the conservative estimate was about 3000 bytes or one kilobyte every minute.
So, the source to UUCP is approximately 120KB at 1 KB per second, which is 20 USD or 2 hours. If inflation is to be adjusted, that is more than 60 USD in today’s time. Also, most people do not wish to have most packages.
Lack of adequate bandwidth
Another issue was Duke had just a couple of autodialers. So, they did not have the necessary bandwidth to transfer large files to several places. If they tried to do so, it would stop all news transfers to other websites. Rather, the suggestion was that perhaps Duke could act as a central repository. The software might be retrieved on demand. UUNET adopted this model later on.
However, the most interesting point is that the announcement did not discuss any possibility of non-technical use. There were no hobby discussions, social discussions, or political discussions. They did not think people wanted to discuss issues with someone they did not meet ever.