forwarded message from David Green (Marc Andreessen)
Date: Fri, 9 Jul 93 03:31:19 -0500
From: (Marc Andreessen)
Message-id: <>
Subject: forwarded message from David Green
X-Md4-Signature: 39037fb92eb6e2e08281e08303ae8491
Status: RO
------- Start of forwarded message -------
From: (David Green)
Subject: Organizing gophers
Date: Fri, 9 Jul 93 17:27:31 EST

I wrote the following comments in response to a query on 
another list. However I thought that subscribers to  NIR 
may also find this of some interest, as it reflects some 
of my experiences in the "front lines" of network information. 

It's unfortunate that discussions of network information 
tend to focus almost exclusively on hardware and software. 
The issues involved in actually organizing the information 
are just as great, and are rapdily becoming more critical 
as the user base of Gopher, WWW etc expands. 

Cheers .. David Green 

- --------------------------------------------

Dora writes ... 

> I think it would be interesting to begin a discussion on how
> to organize this mass of information using a gopher network ...

A  number of us who run gopher services have been thinking about 
this problem for a long time now. While the number of servers, and 
the amount of information, is relatively small the problem is not 
too bad. Most managers have adopted much the same approach as I 
have, namely create a list of "other sites", so that people can 
explore for themselves. 

However this approach soon bogs down because most sites provide 
information on a variety of topics, so it can take users a long 
time to sift the "wheat from the chaff". The following are the 
steps that I have taken in order to help users: 

1. I've tried to place the infromation that users really want as 
   close as possible to the top level menu. Hence I provide a 
   series of thematic menu options on subjects that I try to 
   support, e.g. biodiversity, landscape ecology, molecular
   biology, ...

2. I've tried to build some redundancy into the system by 
   providing alternative paths to the same information. 
   e.g. the path ~/biodiversity/other_sites leads to the  
        same information as ~/other_sites/biodiversity. 

3. Instead of just providing pointers to "other sites", I 
   also try to "pick the eyes" out of each site by providing 
   direct pointers to useful information at those sites. 
   An extreme case of this is my medical folder - I maintain 
   almost no relevant data locally at all (except virus taxa
   database), but have collated a series of pointers to major 
   This may sound parasitic to some of you, but it serves some 
   very basic, but important functions: 
   (a) information is maintained at its source; 
   (b) collating live pointers to services, not just sites, 
       is an EXTREMELY helpful indexing service for users. 
   (c) it helps draw attention to sites that may be overlooked 
       otherwise - users tend to go back to the "big" sites again 
       and again. 
   (d) There's mountains of material out there. No single site 
       can hold everything, as many are now discovering. 
   In my view is that there is at present far too little of this 
   sort of sifting network information. Obviously the sifting 
   requires a lot of effort by the system manager. However users
   should realize that an important contribution they can make is 
   to compile "live" lists of links around specialist themes. All 
   you have to do is compile a host of bookmarks. There are lots 
   of newsgroup FAQs about, but very few of us seem to be compiling 
   this sort of information into live links. Many would see that 
   this sort of activity is irrelevant, given the power of Veronica 
   searches, but as with "other sites", veronica searches take a 
   long time to sift through. 

   I would be very happy to hear from users who have compiled 
   live links information and would be happy to create some 
   specialist folders out of such material. 

   The main difficulties with this approach are that servers often 
   reorganize, or delete material without notice, so links sometimes 
   become dead without warning. Also, as network traffic increases 
   the need to "mirror" rather than merely provide pointers becomes 
   more important. For example, I pull down weather images for various 
   parts of the world each day, so that (Australian) users won't be 
   continually transferring satellite images across the Pacific. 
   Here automation is the key - a few cron scripts go a long way!
4. The next stage beyond (3) is to develop coordinating projects for 
   specialist themes. My idea is that sites willing to provide 
   support for a particular topic become "nodes" in a thematic 
   information network. Here "support" means (a) maintain on-line 
   information; (b) accept contributions from users; (c) coordinate 
   with other nodes. My first experiment with this approach is 
   "FireNet", which we hope will grow into a world-wide information
   network for everyone interested in landscape fires. I am 
   currently developing material for other such themes, including 
   palynology, viruses, and complex systems. 
   The principle is that the information stored at all nodes in a 
   thematic network becomes a truly distributed information system. 
   Each node would provide pointers, or mirrors, for information 
   on other nodes. Moreover the directory/menu structure should be 
   consistent across all nodes in the system. Other sites can, of 
   course, provide access to the system simply by pointing at one 
   of the nodes. 

5. The above issues become even more critical as we move beyond 
   gopher to hypermedia information stored under World Wide Web. 
   Here I have adopted the principle of providing "home pages" 
   for each topic that I support. Thus I have home pages for 
   biodiversity, FireNet, pollen, weather and global change, 
   Landscapes, etc to parallel the gopher menus. These include 
   gopher links, of course, but also pointers to other hypertext 
   documents. Again the key is to provide lots of thematic trails 
   through all the documents you provide. 

Finally, I should say that I see myself very much in the role of 
network publisher. Indeed I am now literally obtaining ISBN and ISSN 
codes for some of the material that I place on the system. It's 
important that the user community come to recognize this and act 
accordingly. i.e. instead of being passive users, they should look 
at what they can contribute. Those of us who are trying to maintain 
systems cannot do the job on our own - it's intensely annoying to 
receive complaints of the kind "your info. is out of date" or 
"why aren't you providing ...". As the scale of the system becomes 
some user's need to play a far bigger role than they have to date. 

I hope that your readers will find the above comments of interest. 
I attach below contact details for my Bioinformatics information 
service for anyone interested. 

Cheers ... David Green 
name 		Dr David G. Green  
address 	Research School of Biological Sciences & 
			Centre for Information Science Research
		Australian National University   
		GPO Box 475  Canberra 2601 AUSTRALIA
phone  		61-6-249-2490(or 5031)
fax  		61-6-249-4437

Access to to ANU Bioinformatics ....

FTP      : 

Gopher   :         
Name=ANU Bioinformatics

World Wide Web

------- End of forwarded message -------