My first experience using the Internet involved using e-mail and within a short time I had signed up for several services which promised to bring me useful information about topics in which I professed interest. The first rule of ³signing up² -- keep the instructions for ³getting off² -- was promptly ignored and there were several mornings when my e-mail in-basket was full of stuff that was of little interest -- at least to me. Since then I have found that a judicious use of e-mail services can provide a useful way of exchanging information. This month I have asked Dr. Stefan Krumm, Institut fuer Geologie, Erlangen, Germany, to write the ANON column. Stefan maintains the Very Low Grade Metamorphism WWW [or alternative].

Facilitating Scientific Exchanges Via The Internet

There are two e-mail based forums for discussion on the internet: newsgroups and mailing lists. The main difference between them is that newsgroups are public and at least allow reading messages for everybody with internet access. Posting messages also may be possible for anyone, although some news groups are moderated, i.e. incoming messages are first passed to an administrator for confirmation of posting. The reader is not informed about a new message appearing in the forum but actively has to keep track with the discussions.

After having subscribed to a list, the user will receive any message distributed by the list. These e-mail mailing list generally have a bit more privacy, as messages can only be read by a defined number of list subscribers. While a newsgroup needs a certain minimum of interested people to be initiated, a mailing list can connect 3 or more people Therefore, mailing list are an ideal forum for information and news exchange about narrow fields of research.

We will report on our experiences running e-mail lists for about one year. We are running several small (c. 30 users) local list and two larger, open lists. One is about geological software and has currently about 350 subscribers, the other connects researchers in the field of very-low-grade metamorphism (100 subscribers). Interestingly, it is the smaller one of the both lists that produces more input and more scientific output.

Listservers are available for any computer platform. However, as many e-mail related tasks are native UNIX, it is a good choice to run the server on an UNIX-machine. The opportunity to write scripts for message filtering, preprocessing, etc. is another argument for UNIX. We are running the list server on a Sparc10 workstation with usually nearly completely filled up 1 GB harddisk and 32 MB memory. The operating system is Solaris 2.4. The software we are using is the Listproc 6.0 a system developed by A. Kotsikonas. We do not know if it is the best system, we are just using it because it was the first we found when searching the net, it is free and we do not yet have any reason to complain about it. It comes as C source code with makefiles which enables recompilation on any machine/platform and what allows easy modification of some program functions. However, you need a C compiler to get the package running. Installation, compilation and setup is easy as most work is performed by a very comfortable setup script that manages that tasks nearly automatically. If something is not working after setup it is most probably the "UNIX error #1": a problem with read/write permission setting. If further assistance is needed, one can subscribe to the unix-listproc e-mail list (see below). This forum is joint with users/listadmin with various skill levels as well as by the author of listproc so there will be a work-around for any problem. Having only very basic UNIX experience at that time, I was able to setup the system within one day.

A list server system has to perform at least the following basic tasks: first, it must be capable for receiving a mail message and to distribute it to the subscribers on the list. It is good if the system allows checking for messages that already have been submitted, for exceedingly long messages and for filtering messages that may not be meant to be for the list. Further, the system must handle request for subscription of new users and signing off of bored list users. The amount of work for maintaining the list depends a bit on the list setup and the skills of the users. If everybody on the list would follow the "5. basic rules for list users" the list would run with about 1 or 2 hours a month needed for watching, archiving and/or deleting the growing log-files. The first problem is to sort out user addresses from the subscriber database that permanently cause delivery problems. In order not to get problems with our computer center, I filter out each address that caused a single delivery problem.

The next problem is caused by user that are not willing or are not able to unsubscribe from the lists by the server requests. They have to be removed manually.The main problem is caused by people that subscribed to the list but were not aware of doing this (this is not a frequent happening!) or that do not remember that they subscribed. These people will complain about receiving e-mails from persons that they do not know, mail back to those person and/or to the list and cause gigantic disasters resulting in chain reactions on the list or list users that complain that their e-mail goes out to everybody and that they receive anybody's e-mail. If that happens, the list administrator can cross the next 2-3 days off his working time-table.

We are mirroring the discussion on our list to the World Wide Web. The purpose is mainly to enable looking up of previous discussions in the case that a message got lost at one location. Another thing is to get the messages into a thread presentation that makes following the discussion easier than if they are mixed up with other messages in a personal mail file. Furthermore, some of the messages are of interest also for a broader readership. We are using "HyperMail" for doing the conversion. This program is currently available only in UNIX flavor. It reads a mailbox and automatically converts each message into a html-document with links to preceding and following messages. An index page allows selecting the way the docs are sorted, either by subject, or by author or as thread. Generally, some manual editing of the mailbox is necessary before using Hypermail. Hypermail uses some information in the e-mail messages that a normal user does not see. This information, however, is necessary to enable the building of threads. Some special characters or line feeds at the wrong place in the mail file can crash hypermail. Or if a user changes the "Subject:" field of a message, the thread also can not be reestablished by Hypermail. Further, frequently there are messages that are not worth to be put into the public: misaddressed unsubscribes, things that are out of date (jobs, conference dates) etc.

   Since January 27, 1997