Computer Administrator Princeton University Press
It was just over a year ago, shortly before the last ARL/AAUP Symposium, that Nebraska became the first university press to put a searchable catalog of its books online. To be sure, Nebraska did not use Gopher, though it plans to switch to Gopher as soon as it can. The first Gopher-based online press catalog was created, appropriately enough, at Minnesota. Several other university presses soon followed -- taking advantage of their parent institutions' campus-wide information systems, which is a good thing to do if you can.
This past September -- such a long time ago -- Chicago became the first university press to mount its own Gopher server on the Internet. A few days later Princeton followed suit. In the beginning, we weren't ready to put our catalog online, so we built a menu pointing to all of the other academic press online catalogs. (This went through several iterations before we got it right, as can be attested by readers of the AAUP List.) Initially, our menu pointed to just the aforementioned presses plus British Columbia, Illinois, Johns Hopkins, MIT and Rutgers, but it quickly grew. Moreover, a growing number of other Gophers began pointing to our menu as a route to University Press Online Catalogs, or created their own versions of it.
Of the thirteen North American university presses that had catalogs on the Internet as of last month, eleven had chosen to use Gopher. Why? For starters, in the two years since it was developed at the University of Minnesota, Gopher has become the most popular vehicle for campus information systems across the country. As of November 1, more than 2900 public Gopher servers were at work on the Internet Because of its power and ease of operation, more and more people are using Gopher as an "Internet Browser" or tool for navigating their way through and finding information in this still largely uncharted mother of all networks. And as a practical matter, there is a pool of expertise at most universities to draw upon for assistance in setting up a Gopher-based catalog.
What makes Gopher even more useful as a research tool is the availability of Veronica search engines (the name was intended to complement a similar indexing system called Archie -- archive without the "v" -- used for locating files in FTP sites). The Veronica indexers scan the entire Gopher universe twice a month and catalog every directory and file listing. These indexes are transmitted to a half-dozen central Veronica search engines around the globe that can be accessed by any Gopher user on the Internet. A boolean search on Veronica produces a menu pointing to all the "hits" found; the user can then follow the pointers that seem most promising to their source.
Consequently, all a university press has to do is construct its online catalog so that every book is listed in files containing the relevant bibliographic data and organize it so that the files are listed on menus with the authors and titles in the item descriptions. Then when the Veronica indexer makes its scan, it in effect constructs a union catalog of all participating university presses. To take a Princeton example, if you go to any of the Veronica search engines that have indexed university press Gophers and search for "Woodrow Wilson," you will get a list of all 69 volumes of the Woodrow Wilson Papers published by Princeton, along with many other references to the 28th President. (As of this writing, six Veronica sites have picked up university press listings: PSI in California, NYSERNET in New York, University of Manchester in Britain, University of Cologne in Germany, University of Pisa in Italy, and SUNET in Sweden.)
As in the above example, Veronica searches will mix hits on university press books with many other kinds of listings. This is advantageous inasmuch as a scholar using Veronica to look for information about a certain subject can hardly avoid encountering relevant university press titles. But if one is looking only for books, the results of a Veronica search will be somewhat cluttered. This is where another search tool comes into play: it is called Jughead and, as the name implies, it is a kind of limited Veronica. Whereas Veronica searches all of Gopherspace, Jughead can be made to index and search only a predefined portion of it--such as all university press online catalogs--filtering out everything else.
Thanks to the efforts of Bruce Barton at the University of Chicago Press, there is now a Jughead devoted exclusively to the AAUP universe. It can be accessed from the Chicago Press Gopher (go to the "Catalogs from Other Presses" menu) or from the Princeton University Press Gopher (go to the "Online Academic Press Catalogs" menu) or from any other Gopher server that points to either of these. Before long, we hope to have more AAUP Jughead search engines at other locations, which will increase users' chances of getting through when traffic is heavy.
Furthermore, we are very close now to establishing a central AAUP server that will provide an online catalog of all the books in print of all the member presses. The exact configuration remains to be worked out, but the idea is to create a comprehensive "union catalog" that would enable one-stop shopping as well as separate access points so that each press could have its own individual catalog on the central server. This Gopher would supply basic bibliographic data for all AAUP members with optional pointers to their own Gophers, and member presses would be able to add supplementary detail (such as catalog blurbs, tables of contents, abstracts, reviewers' quotes, etc.) and perhaps even to build more elaborate menu structures. The server would also furnish full-text indexing and search capability for such descriptive matter, and some kind of order-form mechanism.
An online catalog should not be viewed as an end in itself, but rather as a beginning: what we have here is a viable means for delivering electronic product in the future (and I use the vague word "product" because we have only the vaguest idea yet what forms of information we will be delivering). This delivery mechanism will become more sophisticated and will offer more features -- including a secure means for billing to credit card numbers. And to judge from the current scramble of cable operators, telephone companies, and media giants, it will not be long before we can deliver efficiently to home as well as office.
For those who are new to this technology, it might be helpful here to elaborate a bit about what is actually going on when one Gopher points to another. The key to Gopher's client/server architecture is that the server does not have to "hold state" for its clients, which minimizes both the amount of computing power and the amount of network bandwidth required. That is, when a user (client) calls up or follows a pointer to a Gopher host (server), it in effect sends a single message (called a selector, a request to have something sent back). The server responds by sending back the requested item (typically a file or a menu of pointers to other items) and closes the connection. The user can then think as long as he wants before sending his next request, for he is not putting any load on any server or any traffic on any cable.
The client software retains each menu as the user picks his way through the tree of directories. Behind each item description that he sees on his display screen is the associated pointer. It consists of a numerical code indicating the type of item listed (0 for file, 1 for directory, etc.), the text describing the item, the path to the drive and directory where the item is stored, the address of the host server, and the port on that machine to which the connection must be made. When the user chooses the item he wants, typically by clicking it with a mouse, a signal is sent to the associated host server, via the designated port, transmitting the desired selector. It's almost that simple.
The elegance lies in the combination of ease of navigation and the economy of demand on network as well as local machine resources. Mark P. McCahill, the Gopher project leader at Minnesota, has dubbed it variously "Internet Duct Tape" and "the Finger protocol on steroids." The beauty derives also from its flexibility: at the server end, one can build any structure of menus pointing to resources and directories of files for reading or downloading. If the information you want to make available in your online catalog already exists in a database, it should be possible to write routines that will automatically generate both the menus and the text files. (We did this at Princeton using FoxPro and we have posted an annotated version of the code in a "Developer's Toolbox" on our Gopher for anyone who is interested.)
UNIX Gopher servers create menus by reading the designated directories and then adding in other pointers and more elaborate item descriptions as indicated in associated "link" and "cap" files. The Macintosh Gopher server makes the process of building menus even more transparent: you just go out and grab bookmarks and drop them into your menu structures (defined by file folders) as desired. This amounts to assembling what McCahill has described as an "Internet Convenience Store" (he also says, "If you build a good organization, they will come.") Looking at it from the other side, you can put together menus of selected Internet resources in whatever pattern is most convenient for your staff to access them.
You do not need a big machine to mount a Gopher server, though you do need to have a direct connection to the Internet. Some of Minnesota's top level servers run on old Mac IIsi (25 Mhz) boxes; in the PC realm, a 386 machine is adequate (in both cases, you want to pop in as much memory as you can afford). If you don't have a computer specialist on your staff, you probably will need some technical help but, as noted above, that should be available at most universities these days. Indeed, you may want to leave all the technical details to your university's computer service department and just specify the menu structure you want and provide the data files. Most university press Gophers are installed on machines being run and maintained by their parent institutions.
While you can get by with a Mac or a PC, if you are at all ambitious about your Gopher, you will want to run it on UNIX, because that is where the primary development effort is being made and that is where the most tools are. You can run Apple's UNIX (known as A/UX) on a Mac, and there are several UNIX options for 486 machines, such as Linux (which is free) or NextStep. Client and server software and documentation for UNIX and native Macintosh are available through boombox.micro.umn.edu.
A second-generation protocol, called GopherPlus, has recently been implemented to varying degrees in different clients and servers for the various platforms (principally UNIX, Mac, PC, and VMS). Among the new features it offers is a facility enabling users to fill out order forms and leave them posted at the host server--an addition of obvious utility for university press online catalogs. Other improvements over the first generation include:
retrieval of more kinds of non-text items, including many graphics formats
alternative representations (or VIEWS) of a document, besides plain text, such as postscript or word-processor-specific formats, or even foreign-language translations
multiple attributes associated with an item (name of administrator, time of last modification, abstract of contents, etc.)
more efficient binary transfer
ability for client to send data to server
server-defined forms (to be filled in by client user)
better error reporting and handling
easier indexing capabilities
a place for authentication techniques (still under development)
The UNIX GopherPlus server, which is now up to version 2.10, supports full-text search (WAIS or NeXT) and an integrated gateway to WAIS and FTP. It comes with a Perl script called go4gw that will automatically connect to Archie, USENET, Finger/Whois, Webster, Netfind, and other services. It is also capable of fanning out parallel searches to more than one server.