This paper is the intellectual property of the author(s). It was presented at CAUSE98, an EDUCAUSE conference, and is part of that conference's online proceedings. See http://www.educause.edu/copyright.html for additional copyright information.
Since 1995, the Library of Virginia, working with VTLS, Inc., has implemented an extensive Digital Library Project now providing web-based access to more than 2.2 million digital images of original documents, maps, photographs, etc. and more than 60 new databases and finding aids. Most of the documents that were digitized were previously inaccessible to remote users, and were costly to provide to onsite patrons.
Although the initial cost of creating digital images and associated indexes, databases, and finding aids can be high, presenting them online offers greater opportunities for distributed uses, with a subsequent lower unit cost per use. We will present a model of the differential in cost to provide the original or hard copy document to a patron, both remotely or on-site vs. the initial and ongoing costs of developing and mounting the digital version of the document on the Internet.
The Library of Virginia's collections encompass some eight hundred thousand volumes and eighty-three million manuscript items, ranging from portraits, for example, to the original fire insurance policy on Mount Vernon, from Confederate pension lists to seventeenth-century colonial documents, and from election ballots to numerous donated family Bible records. In order to serve its international audience more effectively, the Library in 1995 decided to provide universal Internet access to this vast treasure of historical documents, records, finding aids, and pictorial materials through the process of digitization.
The project team prepared a detailed analysis of workflow design, and developed a detailed project plan. The initial primary tasks were to analyze the various collections; address long-term preservation issues; understand and document project requirements; estimate cost and time involved; perform document analysis; propose the indexing strategy; and design a custom, turnkey solution for organizing the collections, digitizing more than two million documents, and providing powerful search engines (for more than sixty databases and indexes) and multimedia retrieval tools to access gigabytes of distributed data over the Internet. The Library contracted with VTLS, Inc., to implement a complete solution. The URL for the Digital Library Project is http://image.vtls.com.
The new finding aids and databases of images use technologies in digital imaging and document preservation that result in the more efficient and more comprehensive access to Library materials than currently available as well as the protection of fragile and aging materials.
Established by the Library of Virginia, the Colonial Williamsburg Foundation, Inc., the Virginia Historical Society, and the University of Virginia, the VCRP seeks to reconstruct the archive of Virginia's colonial history--a documentary record decimated by war and fire. Between 1955 and 1985, project agents visited more than 100 archives and libraries in Great Britain, Ireland, and France to survey their collections for Virginia-related documents. The findings of the VCRP agents were recorded in 14,704 Survey Reports that describe relevant collections in European repositories and 964 reels of microfilmed original documents were also acquired. While the VCRP had long been recognized as a basic resource for the study of Virginia colonial history, indexes and finding aids were sorely needed because of the number of reports and reels of film.
The Virginia legislature in 1985 gave the LVA a one-time appropriation to support computerization of the half-finished VCRP index. The project used MARC bibliographic formats to create the index database. Using a standard format rather than proprietary database software enabled us to migrate the data to newer versions of the VTLS software and utilize advances created for this standard, particularly those associated with hyper-media links. Finally, the rigidity of MARC formats facilitated gaining editorial control over the disorganized information in the existing incomplete index.
The VCRP database now contains bibliographic records for all 14,704 Survey Reports and provides access to the 500,000 personal and ship names appearing in the reports. The records also list the microfilm reel number for the original documents. The Survey Reports themselves were digitized and the 28,000 images linked to the MARC bibliographic records via the 856 field for immediate access by clicking on a multi-media icon. Prior to this project the Survey Reports were only available on microfilm at the participating institutions and were not indexed. Now historians, archivists and other researchers can locate and examine any of the Survey Reports from their own home, office, or library.
The images were scanned at 400 DPI resolution and stored in TIFF format with CCITT Group IV compression. These images were stored on tape and lower resolution 300 DPI surrogates were created for WWW access.
Since the availability of the VCRP's 28,000 pages of Survey Reports, detailing the contents of each manuscript item examined, via the Library's Digital Library, the use of the materials has increased exponentially. The Colonial Records Project is now consulted constantly whereas, before, public-service staff members had no way to search its contents effectively.
The existence of the online VCRP Survey Reports has made it possible for on-site and online users to identify relevant research materials and to request microfilm of the manuscripts via Interlibrary Loan, or to purchase photocopies.
Indeed, once the online database was first available, colleagues were so certain of the potential demand for VCRP documents that the Digital Library Project funded the production of two additional copies of the more than nine hundred reels of microfilm for Interlibrary Loan access.
The creation of Bibliographic Records cost an estimated $3.63 per record. The creation of Digital Images cost approximately $.41 per image.
Primary ongoing costs include the following components:
Through time and motion studies, and tracking of fixed and variable costs, it is possible to determine the exact daily cost to the Library of maintaining the Digital Library project in its entirety, and the proportional costs of the VCRP.
The cost for Library of Virginia staff to serve the originals to users, whether those actually present in our Reading Rooms or those that request copies remotely, can be measured through sampling and periodic tests. The costs include the salaries (and benefits) of the public-service staff, interlibrary loan staff, and stack pages needed to assist patrons in identifying needed documents, request documents from the stacks, locate the items in the stacks, deliver the materials to the patrons, retrieve the collection from the patrons, and, finally, reshelve the items in the stacks.
Those requesting copies remotely require interlibrary loan staff to process requests and direct them to the Photographic Services Department, stack pages to retrieve documents and deliver them to the photoduplication laboratory, and lab staff to prepare the copies, then ensure that the copies are eventually mailed to the patron, which then involves the Mail Room staff.
Finally, staff in the Accounting Department become involved during the billing process. The Library of Virginia does charge cost-recovery fees in the latter case, but it is doubtful that these fees actually cover the true cost of the service.
It is also possible to prepare a model, again based on sampling, of the differential in cost to provide the original or hard copy document to a patron, both remotely or on-site vs. the initial cost of developing and mounting the digital version of the document, plus an estimate of the cost of server storage and Internet transmission. All of the Digital Library resources are available via Internet-connected PC's throughout the Library's reading rooms.
An associated model would include an analysis of the turnaround time involved to deliver the requested materials to the on-site or remote patron using both the hardcopy and digital iterations of the documents. Obviously, the transmittal time for the former is much greater than for the latter.
The different means by which a patron might use the Virginia Colonial Records Project Project, and how the Library provides access to it, offer a useful comparison of costs to both user and service-provider.
The model poses three distinct researchers: one who is physically present at the Library, one who is requesting information and copies of documents from the VCRP via standard first-class surface mail, and a third who accesses the collection online via the Digital Library Project. The on-site visitor travelled to the Library by air from California, the correspondent lives in Tennessee, and the online user lives abroad. (The Library, in fact, receives hundreds of written requests from Tennessee each year; and to date the Digital Library has attracted users from more than ninety foreign nations.)
Using actual hourly wage data (including employee benefits), the Library of Virginia has estimated a cost-per-minute for all staff members that would be involved throughout the transaction. The Library has also estimated the number of minutes each stage of the transaction would consume to determine a possible total cost to the institution for each transaction. The Library also estimated the total cost to each user for the transaction.
This model can be used to propose the true costs to the user and the Library.
The constant elements throughout each sample research experience are:
The on-site researcher will require the assistance of a staff librarian or archivist to interpret the information request and then target the VCRP as the logical, primary source. The staff member then introduces the bibliographic database for the reports to the researcher, and provides instruction in its use. The researcher spends approximately one hour perusing the database, and locates one promising Survey Report. The researcher contacts a staff member and requests to view the Report on microfilm.
The staff member retrieves the reel from the microfilm cabinet, and assists the patron in loading the reel on the microfilm reader/printer. The patron advances the reel until he/she locates the desired Report. At this time, the patron uses the Reader/Printer to make a copy of all four pages of the Report.
The staff costs to the Library of Virginia for interpreting the request, assisting the patron in the use of the database, retrieving the microfilm, and assisting the patron with the reader/printer total approximately $12.17 to fulfill the request. The costs to the On-Site Researcher may total as much as $1,120.00 and would include, for example, long-distance telephone calls placed in advance of any trip, babysitters or housesitters, costs associated with automobile travel or public transportation, lodging, meals, and of course charges associated with document copying and other on-site research work.
There are hidden costs for such research as well. For instance, what is the value to the researcher of valuable vacation leave from a place of employment in order to conduct research? In addition, what is the cost of lost research hours, as the Library is not open in the evenings or on Sunday? And, what is the cost of time spent searching indexes, finding aids, and library catalogs while on-site before even viewing, evaluating, and copying the original documents actually relevant to the individual's research?
The surface-mail user will require the assistance of a staff librarian or archivist to interpret the information request and then target the VCRP as the logical, primary source.
Because of the thousands of research requests received by mail, the surface-mail user ordinarily must wait an average of six weeks for delivery of the VCRP materials.
The Survey Reports are provided to the surface-mail user via prints from microfilm.
The staff and other costs to the Library of Virginia for processing the written request, identifying the appropriate resource, searching the database, retrieving the microfilm, duplicating the VCRP records, and processing the reply and payment total approximately $18.15 to fulfill the request. The costs charged to the Off-Site Correspondent--for standard research, photoduplication, postage, etc.--will total at least $28.00.
There are hidden costs here as well. For out-of-state correspondents, for instance, there is an additional charge: a twenty-dollar service fee assessed each non-Virginian for each query; moreover, to keep pace with the considerable rate of mail, research letters may not include more than two (2) research requests. The potential delays may also affect research or publication deadlines. And there are the costs associated with pursuing a line of inquiry whether or not a relevant document exists; perhaps the requested search will prove fruitless, in which case the Service Fee for time staff expended is nevertheless charged to the patron.
The online user discovers the VCRP resource either through familiarity with the Library of Virginia's Digital Library Project, or through initiating a search within a World-Wide-Web search engine, or through the Library's online marketing and public relations efforts.
The online user spends 60 minutes online after first locating the VCRP Digital Library resource. After searching the database, and locating the desired report, the user prints the four-page document.
The costs to the Library of Virginia for providing material from a pre-existing Digital Library resource are calculated at a rate of $0.03 per minute for the Internet Connection to a particular Server Resource.
Assuming that the activity of this particular user is the only activity that is consuming any network or server resources (which is totally unlikely), and the user employs the online resource for thirty minutes, it will cost the Library $0.90 for this particular transaction. This figure, while it does not include any fraction of the staff costs to maintain the equipment, or the ongoing costs associated with the online storage of the resource, is most likely completely unrepresentative of the actual cost of the transaction to the Library. It is almost certainly considerably lower.
The likely costs to the Online Researcher will include his monthly Internet connection and monthly telephone service, equipment, and consumables such as printer paper. The cost of the transaction to the user, without considering amortized capital costs of equipment, would be approximately $0.13.
The Library has developed three potential scenarios to compare the development cost of the project with the costs associated with serving the materials to the On-Site Patron, the Surface-Mail Patron, and the Online Patron.
Using the preliminary model, the Library estimates that it would cost $12.15 to provide a single copy of one four-page Survey Report to the On-Site Patron.
By comparison, it will cost an estimated $18.56 to supply the Surface-Mail Patron, and $0.90 to supply the Online Patron (again, this figure assumes that there is no other activity occurring on the server, which is a false assumption, so the cost will be exponentially lower.)
Combining the service costs shown above, and the development cost of four images of Survey Report pages and one bibliographic record, one online transaction would cost the Library $6.17.
Over time, it is possible to determine the proportional reduction in cost to the Library to serve the Online Patron, and to compare these reductions to the costs associated with serving the On-Site and Surface-Mail user. The development cost per transaction will be offset over time.
Between January 1, 1997 and September 30, 1998, Internet users executed 86,400 primary bibliographic searches in the VCRP database, and viewed 73,700 Survey Report page images.
The development cost for the digital images ($11,766.00) spread among 73,300 page image transactions would be $0.16 per transaction. The development cost for the bibliographic database ($53,408.00) would be $.62 per bibliographic record viewed. After including the theoretical $ .90 online transaction cost (again, this cost assumes no other activity on the server) to the Library of Virginia to the $.94 development cost, our model online transaction would consume $1.84, with two Survey Report pages viewed and printed.
Over time, the cost per online transaction will continue to decrease.
Over time, the costs to provide the same materials via surface mail ($18.56) and to onsite patrons ($12.15) will likely increase.
It should be taken into account, however, that it would obviously be physically impossible for the Library to effectively serve these numbers of patrons and complete these numbers of transactions either on-site or via surface mail.
It is possible to extrapolate costs/savings to the user with this model.
Without factoring in real and hidden costs to the On-Site Patron, such as travel or vacation time, and only considering the actual charges to the user for reproduction of copies, the same number of images would cost $92,125.00. Surface-mail users would incur $196,528.00 (this total does not include the $20.00 service fee imposed upon out-of-state users). The costs for the online users would be negligible.
The Library of Virginia is seeking substantive answers to three primary questions to accompany the raw cost data. Components of an extensive evaluation will include, for example, analyses of the known as well as perceived advantages and disadvantages, or impediments, to online research within the VCRP as part of the Digital Library Project.
The identification of recognized and perceived advantages and barriers from users includes querying on-site patrons, traditional long-distance patrons (who access the Library via telephone, mail, Interlibrary Loan, etc.), and more recent online patrons.
The collection of data is by survey, focus groups, listservs, on-site interviews, and server log-file analysis. Staff members as well as patrons are included in such analyses. Preliminary work has already resulted in a sampling of user comments on the differences between present on-site and potential online research within the VCRP collection. The primary questions to be posed are as follows:
Factors to be addressed include:
Factors to be addressed include:
Factors to be addressed include:
After all cost analysis is completed, the most satisfying measure of providing access to valuable research materials over the Internet, such as the Virginia Colonial Records Project, is characterized by the frequent communications Library staff receive from remote users. Their anecdotal responses confirm that the monetary and other resources allocated to creating Digital Library holdings are a very wise investment, indeed.