Austin, Texas
May 18-20, 1994
David Stam, Convener, Syracuse University
James O'Donnell, University of Pennsylvania
Prentiss Riddle, Rice University
MR. STAM: Good morning, everyone. I appreciate your interest in this session. I am convener of this session, which came about because of a question asked on the Directors' listserv about three months ago. My question apparently got the biggest response on that list, including a request for a breakfast meeting to discuss it. We have two people with us who have extensive experience in this area. Jim O'Donnell from the University of Pennsylvania and Prentiss Riddle from Rice University. They will each tell us a little bit about their experience and then we will have some discussion.
MR. O'DONNELL: I want to relate a little bit about our experience, mostly to describe some problems and the fruitful state of uncertainty that we have found ourselves in at the University of Pennsylvania. Briefly put, the situation is this: I have just been appointed co-chair for next year of the University Council Committee on Communications, which has been charged to look at the way we, the campus, enter into providing information on campus. The particular impetus is to give some academic direction to information in an academic institution.
One feature of the CWIS, we realized after running it for several years, is that it didn't actually contain, at any one time, very much academic information to show that it was a university. The quality of the academic information tended to be quite low. This seemed to represent at best a phenomenal picture of what a university is, particularly because our system did repute to be campus-wide in several kinds of ways and open to all kinds of information, including a wide variety of features. So the absence of academic information and the low quality of the academic information that was there seemed particularly remarkable. We got into this position because of some important system decisions made between 1986 and 1988. As I look around the world of academic computing and its applications, it seems to me there are a number of questionable and problematic enterprises in operation, some of which were funded and designed at just about that time.
During that period we had a good idea that computers were going to be important parts of the campus communications system. We also had an inkling that there would be a network that could distribute information to computers across the campus. At this time, however, not many of us had ever actually used one of these network-linked computers. This was before it dawned on us, first of all, that information can go in two different directions or, indeed, many different directions, and second, that the resulting community of people using these machines was going to be a bit larger, more chaotic, and more difficult to manage than a traditional university structure.
To tackle this, we bought a good, solid, user-friendly, off-the-shelf package that comes out of MIT, called TechInfo. Now, there are several kinds of people in the computer world. I'm one who has a particular affection for user-hostile systems. User-hostile systems, I find, let you grapple with them and change them around and do things. User-friendly systems are ones that I regard as embodying the spirit of fascism. They are friendly to you for as long as you are willing to do the things that the people who thought up the system expect, wanted, and think is right and proper for you to do. But the moment you try to change something, even such things as irregular paper size in word processing, life suddenly gets very complicated. You find yourself sinking down to a level of a machine that you never expected.
The software package we acquired comes with a slick interface and was, at that time (1988-1989), a very powerful way of putting a lot of information into the community network. I don't know the nuts and bolts of that system well enough to know how much of our problems arise out of the, for want of a better word, "fascist" system and how much they arise out of the university's administrative structure, which I am more accustomed to thinking of in feudal rather than fascist terms, though there are some resemblance between those systems of government.
Our local system is called PennInfo. In order to put information on PennInfo, you need to be an official provider of information on the campus. You need to have been trained how to do this. You need to be fire-proofed by the people who control the system as a whole. When you are putting information on the system, you are entering somebody else's computer. You are putting information in particular kinds of directories. It needs to be formatted in particular kinds of ways, and it is being absorbed into a structure of menus that is centrally, but not authoritatively, controlled by technicians in the computer department of the institution. This means there is very little editing of the menus, in fact, very little editing of the structure that has been put together. It also means that those who supervise the CWIS are very nervous about the quality of the information, not from a point of view of substantive quality of information (such as, "is it true, is it timely, is it still relevant?"), but rather the material formatting quality of the information. The most important thing is that the file is readable in the correct way to software and will display directly on the screen and thus must come from a recognized provider of information.
This is, in its way, a noble vision of how information can be distributed at universities. It reflects a model whereby we are attempting to take the old administrative structure of the university and replicate it in the CWIS environment. One problem we found was that the CWIS tended to reflect the hierarchical structure already in place. Take for example, the Office of the Executive Vice President of the university. The executive vice president of the university is a very powerful person. Few faculty will have met her personally, but all will be aware that she is essential to the institution's well-being. She does many important things. I am not actually sure what they all are, and I am especially not sure what offices report to her. So, if I am placed in a menu that says, "Office of the Executive Vice President of the University," I, as a user, have no clue why I might want to click on that item or what I might expect to find under that heading; but the hierarchical structure of the institution is securely in place.
One insight we can derive from this experience is that, in fact, editing menus and the structure of the menu itself is a substantive editorial job quite separate from the job of gathering the information. It needs to be done with a view to reflecting the interest of a community, rather than the hierarchical structure of the institution.
This authoritative way of bringing information together from providers of information means that, in fact, you will have much less academic information on the system than you would expect otherwise on an academic computer. A powerful entity within the institution can probably find a secretary to become the provider of information, but many of the people producing dynamic academic information are people who do not have that kind of support staff. For example, in our English department, the undergraduate chairman in charge of advising all the majors literally cannot spare his secretary for that kind of time. It is not regarded as a useful expenditure of time to put the information out there. So, our CWIS, about five years after we started it up for active use, contains a whole lot of information that is not very well organized and does not accurately reflect the structure of the institution itself.
Another problem with TechInfo that we did not envision when we first installed it is that it is not especially compatible with or transparent to other forms of information provision on campus. There are gateways to and from PennInfo to Gopher and now to World Wide Web. However, they tend to be relatively narrow gateways, like fire exits in a not-very-safe kind of structure--you need to know that you are inside one to get to the other, and you have to work your way through a particular set of menu choices to get from one place to the other. Some of the worst problems arise from thoughtless transplantation of information from one medium to another. From my point of view, one of the handiest things on the CWIS system is the current roster of courses that is being offered and the listing of rooms in which they are being offered. However, the files were put together in the Registrar's Office according to the printed course catalog. This means, for example, that in my Department of Classical Studies, our courses in ancient history are alphabetized under A. Our courses in classical studies are alphabetized under C. Our courses in Greek are alphabetized under G, etc. Furthermore, each department has, for reasons of mere bureaucratic convenience for the Registrar, three different ways to list all courses: as seminars, as courses offered through the extension branch of the university, and as courses offered by the university's College of General Studies. So if you click on Latin, you find Latin courses, Latin seminars, and Latin CGS courses. If you back up from that, you find yourself in the sublist of course headings from, roughly, H through M. If you back up from that you find yourself in the sublist of all departments. To navigate from courses in Greek to courses in Latin, then, means navigating back up several clicks then back down several clicks. Most such trajectories take five or six clicks on the menu to get from one set of courses offered in a single department to another.
This is a logical way to create information for a published book, with one kind of browsing in mind, but not necessarily compatible for use in a CWIS. When the files are dumped into the CWIS system the structures suddenly appear wild and inappropriate for that environment and program.
These are some of our problems. What shall we do? An easy and obvious solution that presents itself is Gopher or World Wide Web. These are environments where information can be provided by people who do not have to know many details about the formal structure of providing that information. It can be provided locally all around the campus. You can put information under a particular directory on a server, if that is the right place on the CWIS menu, and that is automatically included in the campus-wide system.
The moment you begin thinking about that, the next thing you do is begin managing a grand structure. I just asked Prentiss before we started the meeting, "Are lectures offered by the Department of Chemistry included in your CWIS?" If I were anywhere hanging around the chemistry department, I would like to be able to look at the lectures on a list.
The problem is finding the person in the system who is going to take responsibility for putting chemistry department records on the system in a timely way and--here is the tricky part--finding the people to do that for biology, botany, classics, history, English, and so forth, so that in the end the user has a reasonably reliable idea that you can go to the CWIS and find lectures of interest on this campus without necessarily having to redefine yourself in chemistry and still get a reliable body of information. You simply have a "forthcoming lectures" directory, and people dump stuff in that. Unless you have gone around and made sure that every single department of the university has a responsible person doing it, you very quickly get a directory full of garbage. You will have some departments' lectures poorly represented, or in some cases inaccurate or outdated information. You need outdated information included in the back of your system.
When I looked at the CWIS systems around the country, the one that impressed me the most was the University of Washington--UWIN (University of Washington Information Network). This system provides a lot of high quality information centrally provided from the institution without getting down to the degree of granularity of individual departmental lecturers. It is a very useful thing. It is the place you go to find the Oxford English Dictionary, Books In Print On-line (i.e., resources licensed for that university's community to use). You find current information centrally provided by the institution.
For the future, I try to imagine that the information provided on the campus-wide information system would be substantially less ambitious than has been the case with TechInfo. The CWIS should contain information that can be provided with a high degree of reliability by centrally, or nearly centrally, funded and organized information servers. I envision it producing a lot of information with a good transparent interface that may then be linked to all the other kinds of information that other parties on the campus provide. I would rather have the option of having one campus-wide information service whose quality I know is very high, whose range of information is more, but at least I know what I am getting when I go there. It would be transparent so that other departments, units, and even individuals on the campus can have it as a subset of their provision of information, and they in turn can become a recognized, but less official and less predictable, subset of the CWIS.
If I have a Gopher or World Wide Web server for a Department of Classical Studies or a Department of English, I would have every possibility of organizing that information in the clearest possible way reflecting my discipline and my department's sense of what is important and useful. Then I could incorporate into each departmental menu the campus-wide server as a module of reliable, high quality information. By accessing the campus-wide information server via a campus department, the user can navigate through potentially dozens to hundreds of different paths that have come to that information server from all around the campus. I have then accessed this information from my department's home page. This way the chemistry department has the responsibility for making information about special lectures available to both students and faculty in that department, but it is also available to anyone using the CWIS.
These things can be revolved in circles. The CWIS can contain a list of information providers including those that are not necessarily affiliated with an academic department. One of the beauties of this kind of information is that it can be placed in many different places in the hierarchical structure, like a book that you place in a variety of places in the library so that a number of people can access the same book in different parts of the library although they may have different information needs.
If we can imagine a structure like this, then the quality of information on the CWIS becomes more reliable. There is more incentive to individual department program centers to provide accurate information. A secretary does not need to be trained to be a provider of information on somebody else's system. Control over the kind of information provided by the classics department returns to the classics department: what we want, when we want it, and the quality that we are used to. We get the complaints that we did badly, and the information that is provided on campus, higher quality, gets to be more alive.
I believe strongly that campus-wide information should be integrated closely with information that is provided by the library. But something else that I have learned at this ARL meeting is that librarians are beginning to hear that users do not care about the bureaucratic structure that provides the information as long as they get the information; a CWIS is a good vehicle for that. The University of Washington system that provides the OED through the CWIS seems to be a perfectly rational system to follow. There is a risk that the library won't get "credit" for its contributions to the common stock of information, but I believe that such issues of attribution can indeed be worked out in a way that still maximizes the accessibility of the information.
One advantage of following the model that I am proposing is that the information server also becomes an off-campus information server that provides high quality, central information. I like the University of Washington system because of the high quality information that comes from there. With a flexible structure, there can be one "front end" to the information on the CWIS for those on campus (perhaps even a slightly different front end for faculty, students, and staff) and another very different one for outsiders. For the campus community, it may make sense to have the first item that strikes your eye be, as it was for us recently, "Crime Alert!", but the Office of Public Information might suggest that we give that item less emphasis in presenting our "front page" to the world outside.
MR. RIDDLE: I am a systems programmer in our central computing facility at Rice University; I am not a librarian. Our project is a joint project between our Information Systems Department and Fondren Library at Rice, and that has been extremely valuable. I have a lot of ideas about how to organize information, but I find that when I talk to librarians, they have tremendous insights that I need in order to do what I do.
The interesting thing about CWIS development is that at our campus, as at most campuses, it began as an inward-looking project to get information about the university online on the campus network so the university community could be informed about itself. But the tools that do a good job of that--Gopher and World Wide Web, for example--are also excellent technologies for exploring the Internet as a whole. And that raises the possibility of looking outward. That means that the client software that you train users in and deploy all over campus is not only the chief way of looking inward at information about the university but of looking outward to the Internet as a whole. A good CWIS should be designed with that in mind from the beginning, which again cries out for library involvement.
At Rice we have promoted the idea of a central infrastructure, which is not departmentally organized. It is organized with a user-centered point of view, and departments are free, of course, to put up their own servers. If they put up servers with information on them that is of value, we will happily point to it, but we define the CWIS as the centrally administered portion of what is online. Those of us at the center are not responsible for the content, however. We are responsible for setting up the servers and keeping them going. We do not simply put a server on a Macintosh on a secretary's desk and then suffer when the secretary goes on vacation. Instead, we run a central server which we promise to keep up and do so professionally.
We decentralize the provision of the information, and we train people all over campus to put the information in. The means of putting information into the system that we offer is not as difficult as TechInfo. Gopher is fairly forgiving, in that it accepts straight ASCII text. World Wide Web, while more powerful, is less forgiving because it requires the use of HTML. We train users to produce ASCII text using word processors, and we train them to produce HTML. Any provider of information, regardless of which department they represent, is responsible for the information that they make available on the server.
Another portion of our responsibility is to work out, with these information providers, where in the organizational structure of information their material should go. To use the library analogy, we determine on which shelf it is going to go, and we figure out how it is going to be cataloged, but the information provider tells us what books belong in the library.
We have a CWIS committee that was formed to design and oversee the system. We formed mailing lists and other informal, ongoing means of communication. This means that after a few months the committee basically ceased to meet face to face, but nobody has called me on the carpet for not having called a CWIS committee meeting in all these months. The communication lines are strong enough now that we do not seem to need to take time out of everybody's schedule to meet face to face. We also have a person on whose desk the buck stops--that's me. Some people call me the RiceInfo Czar, which recalls the "fascism" we just heard about. I hope that a joking reference does not mean that I am arbitrary in the way I make decisions. In fact, it is frequently the case that someone will come to me and say, "I think something should be different," and I try to be responsive and change it. I see my czar-like qualities as relating to nagging people to update information and to make sure that all information put into the system is clearly stamped with who is responsible for it and when it was updated, so that users can be aware of issues of quality and timeliness of the information and can alert the original data provider if something is out of date.
Like many institutions, we started our CWIS when Gopher was the best thing available and the clear choice. Then World Wide Web became more and more important, and we now operate our Gopher server with a layer of Web on top of it. When people come to us with information they want to provide, we encourage them, unless there is a compelling reason to use World Wide Web, to put it on the Gopher server. Gopher remains our lowest common denominator. Users can get to it from any platform on campus, not just the high-end Unix, PC, and Mac systems necessary to use Mosaic. However, more and more creative ideas will come up for resources that lend themselves particularly well to Mosaic and World Wide Web, and we encourage people to use the Web when it is more appropriate.
There are technologies appearing on the horizon that make it easier to combine Gopher and WWW in a single server. If we were starting over we would probably do that. For the foreseeable future, which in my field is no longer than about 12 months, we are probably going to continue to use old-fashioned Gopher servers for the bulk of our information.
I will briefly mention some innovations at Rice that have been successful, and then I am really interested in questions and discussion. One is the "Information by Subject Area" in RiceInfo. Early in the development of Gopher, many people wanted to start organizing pointers to Internet resources and came up with a few subject menus in which they started collecting links. These were people without a deep knowledge of how to organize information, and the way they did it was often short-sighted, but interestingly enough, a number of these systems were compatible with one another. Our solution at Rice was to write some software that merges compatible collections of links to Internet resources into what one might call a "union catalog" of subject menus. It is one of the better subject trees out there, and is certainly by far the most popular that we offer. This approach is not going to work much longer, however, because using a couple of dozen categories to sort all the information out there on the net is simply inadequate. We are trying to think now of ways to build a proper index of online resources, using librarians to categorize information on the net. Also, we have something called the Electronic Studio Project, which dovetails with our CWIS and is related to curriculum content. I want to see us go further in this direction. One of the big issues about campus-wide information systems is determining where the CWIS stops and electronic publishing of other kinds begins. In some ways that's a false dichotomy. On the other hand, if you want to say, "I have responsibility for this and you have responsibility for that," you have to draw the line somewhere, and sometimes these false dichotomies are necessary.