Those of you who are responsible for collection management in a regional depository know that such an activity is an oxymoron. To give the full flavor of the collection management issues involved I need to begin by providing some background information about the libraries I am with, and outline a few of the issues surrounding implementing a local area network in a library network our size, and our decision to include depository CD-ROMs.

Currently, all our CD-ROM databases are accessed on stand-alone stations. There are approximately 35 workstations distributed among 7 service sites. The automated reference collection is comprised of commercial subscriptions on CD-ROM and floppy diskette, government depository databases, and selected online services. The total number of databases exceeds 75 titles and more than 200 discs. Hence, our dilemma.

At the University of Maryland at College Park Libraries, we are fortunate to have renovation funds from the recent expansion of our main library to earmark for a LAN. What cannot be paid for from this money are labor and subscription costs associated with the network. At present, we are working with a consultant to develop the RFP for the network hardware and software. It looks like the network will comprise 100+ stations distributed as staff terminals, workstations in two reference units, three instruction rooms, and a microcomputer lab with 50 workstations. In addition, seven branch libraries will be able to access the network over the campus backbone, Internet. Remote access is under consideration, but no decisions have been made.

To aid in collection management decisions for machine readable databases, I chaired a committee which developed a draft evaluation/selection form. The criteria include search interface, subject coverage, software installation, documentation, accompanying materials, and duplication within the system. All products must be tested prior to purchase. This, generally, occurs as a 30 day trial period as opposed to a vendor demonstration. We will no longer order databases from vendors who do not permit an in-house testing prior to purchasing. This form is used to determine what will be renewed each year, also.

There were several factors involved in our decision to network selected databases. What has yet to be determined is which databases will offer concurrent access to multiple users. This is, almost exclusively, a cost issue. We would love to have concurrent access for all of our databases, but cannot afford such a luxury.

Generally, the factors were:

Clearly, before any work is begun on investigating the wheres and whyfors of a LAN for a library, primary clientele must be identified. Who is using the network will dictate which databases goes on it and when. Are your users faculty, undergraduates, the public, or all of the above and more? Do your users need to have access from home or office? Will library staff or patrons need access during hours the library is not open? Will there be files which only staff need to have access to? Do you already have databases available, but not in a networked environment? Who is using these databases and when? How many files would you like to see in the network? Do you also want access to application software for those titles which do not have individual search and retrieval software? How many concurrent users will there be? How many can you afford?

What should go on the network? The University of Maryland at College Park Libraries has undergone several internal studies of what databases would go in a networked environment, either through the integrated library system or a separate LAN. Three years ago the first list was developed for the purposes of identifying databases to be accessed via Victor, UMS' integrated library system. That list remains virtually unchanged, although many more databases exist. This is because our collection is curriculum based. The most recent evaluation, conducted in March 1993, yielded a list of 32 databases to go on the network. It has not been decided which we will negotiate concurrent multiple use for. The prospect of negotiating 32 individual subscription agreements, in addition to the usual renewal negotiations, is more than I can bear to think about just now.

Of the depository databases currently in the collection, some 15 were identified for the network. This decision was made based on subject coverage, quality of interface, and use. The ARS evaluation form helps librarians with such determinations. The SIGCAT CD-CINC Guideline is a valuable resource for those just beginning CD-ROM selection or cancellation efforts. Regarding depository CDs, what we don't have adequate information on is degradation of search speed with current use. Can we, legally, copy the contents of the CD-ROM and its search and retrieval software to a hard disk for better search performance? We will need to see about that, as well.

Compatibility of search and retrieval software is imperative. Software incompatibility may swiftly reduce your list of network contenders. UMI and Bowker software won't operate on the same server. So, if you are only planning on one server you won't be able to run a product like Inspec or Dissertation Abstracts with British Library Catalog or Books in Print. Extract and Bowker have problems, also.

Learn from the mistakes of others. There are several articles written on software compatibility in a LAN environment. The LAN software and CD-ROM drives has a great deal to do with compatibility, and may influence which network software you choose.

What about multi-media databases? What about communication or application software packages? What about the many depository discs which are raw data files with no search and retrieval software? What about full image databases, such as clipart or Patents? Do you just network the index without the full image discs? Do you jump off the bridge now or later? Do you take all your CD-ROMs with you?

4. What possessed me to leave Kansas?

No matter how you get there, neither Archie, Veronica, Ken or Barbie can get you home if you don't know where you are. Add Internet-accessible databases to the interfaces you'll have on the network and just imagine the mess for patrons and staff. Gopher and Archie interfaces are improving daily, and it's hard to evaluate a fluid interface. However, you need to make choices to keep the users of the network as sane as possible. Gradual additions of Internet-accessible files is probably a good idea. The problems arise when the only place to get the information is via Internet. I would argue that the evaluation/selection criteria that you developed for acquiring other databases would serve you well here. Remember that end-user documentation may become a local problem (i.e., you want it, you write it).

5. All major credit cards accepted?

Three major issues come into play at the budgeting level. The first is hardware cost, both for the LAN and the workstations needed for users; the second is labor costs; the third is subscription costs.

Our expenses for hardware and network software are being handled by the renovation funds. However, we anticipate costs to exceed $200,000 for the optical servers and software. Workstation costs add another $400,000 to the pot. Miscellaneous costs for additional Internet connections, modem pools, routers, etc. are not yet known (and we're only at phase one of the project).

Regarding labor costs, the concerns are what costs can you anticipate having with a network? Are you prepared to allocate the resources of a full-time individual (or more) to the operation and maintenance of the network? Are there labor costs which will develop after the network is installed? What impact does labor have on the operation and maintenance of the network? And, what are the trickle-down effects of an inadequately managed network?

Regarding subscriptions, commercially produced databases are of great concern to me. If you network the discs, do you negotiate licensing for concurrent use or do you only offer single-user access? The costs associated with concurrent use can be staggering. Let's use Government Publications Catalog on CD-ROM as an example. GPO from Information Access Company is networkable on CD-ROM and magnetic tape. My library classes at level A in their pricing schedule, meaning we'd pay about $3,500 for 2 standalone stations or about $8,000 for CD-ROM networking. We could load the tape and provide unlimited access for the same subscription cost of CD-ROM networking but with much higher labor costs. Through SilverPlatter, use by 2-8 concurrent users would cost $945, 3 times the cost of single-user access.

In my case, our CD-ROM and online database costs were about $156,000 last year and we anticipate a 6% increase. Of this, we want to network, with concurrent access, 70% of what we subscribe to on CD-ROM. The cost to do this is 3 times more, about $306,000. Needless to say, we cannot afford to do this. So, what are our alternatives:

Where does this leave me? Scratching my head wondering where the money will come from to continue to fund such efforts. Wondering when the cost of accessing information will come down to a manageable level. Wondering if a CD-ROM network is the right course for us to take. Asking many more questions than there could possibly be answers for, many of the same things I did before CD-ROM was developed.