Demand Balancing in Fee-free Resource Sharing Community Networks with Unequal Resource Distribution

Even the largest libraries struggle to maintain
a comprehensive journal collection. In 2003
Australian universities subscribed to over 1,300,000 journals of which of which 974,000 were in aggregate digital collections. This represented over 273,000 new serial titles and over 150,000 cancellations (Council of Australian University Libraries, 2005). One emerging approach is for libraries to form consortia that take
a joint subscription to digital resource collections. The consolidation of substantial collections with direct
delivery has seen the gradual attrition of subscriptions to the traditional print format (Fox & Marchionini, 1998; Weiderhold, 1995), but this cost saving is offset by the substantial increase in digital resources. The wealth of international research resources presents an even
greater dilemma for small research institutions: how to effectively and economically access such a wide
base of information resources within sometimes
highly constrained budgets. The cost reductions obtained
through aggregate subscriptions and consortia do
not necessarily offset the net growth of fee-for-use
published resources, and may have the consequence
of centralizing subscriptions through a few large
distributors – with the long-term collection risk that this centralization presents.
Small research libraries that cannot afford participation
in national inter-library loan networks have formed
fee-free networks of collaborating libraries that share their journal resources. While a fee-free Inter-Library Loan (ILL) service offers obvious attractions to smaller participating libraries, alternative economic approaches are needed to avoid excessive demand on resource-rich members, and to avoid the phenomenon of “free-riders”. This paper presents the resource distribution approaches
that have been used to balance resource demand in
GratisNet, an Australian network of 250+ health research libraries, where collaboration is fee-free but resource holdings among member are unequal. Dynamic ranking resource-based approaches are used to encourage the equitable distribution of resource load.
2. Economics of demand balancing in a fee-free network
Hooke (1999) highlighted the need for evidence
-based approaches in the management of
information services. In a fee-based environment, the metrics for efficiency may centre of cost versus speed of supply. Fee-free collaboration does not have the same economic driver for equilibrium between demand and resource supply that emerges in the long term in a
fee-based service. Furthermore, resource sharing networks operating in a fee-free environment face
several risks that are common to voluntary online
communities. In Gaming Theory “outcomes” and “payoffs” are differentiated (Shubik, 1975). In the case
of ILL collaboration, the payoffs are the supply of
particular ILL requests in exchange for the provision
of requests raised by other libraries at the risk of
absorbing the costs of supplying requests raised by other libraries. The outcomes include access to a wider base of research resources than would otherwise be available to the library, and the potential for requests to exceed loans and constraints on the limit of demands based on membership of a closed community. One of the risks is the “free-rider” phenomenon, or those who take the benefit of membership of a collaborating community but provide no net contribution of resources. “Free-riders” can be managed in a number of ways: through “closed shops” (the example of unions that limit benefits to those who are members only), or through adjustment of the payoffs(Hamburger, 1979).
3. Demand balancing
Imbalanced distribution of workload in a fee-free environment, if unmanaged, can create imperfect resource management through inequitable distribution of demand over time. These imperfections may be
expressed in terms of a reluctance to declare resources or
supply requests (a form of compliance failure), or
through inequitable distribution of demand resulting in a delay in supply (a queuing problem). This reduces the payoff potential for the larger members of the network. While there is a risk that libraries may reduce their own collections through reliance on wider networks, the
trade-off is the delay in fulfilling requests when they are completed through an ILL rather than directly out of their own collection.
The rational choice in fulfilment of an individual ILL request in a fee-free environment is the selection of
the nearest library that has the highest probability of
fulfilling the request. This provides the best payoff in
probability of fulfilment and timeliness of supply. However,
aggregated over time this choice is likely to place a larger burden on those participating libraries with the largest
collection of resources in a given region. Where staff
represents as much as 80% of document supply cost
(Morris, 2004), this can be a considerable burden on larger participating libraries. In the GratisNet network, search results for resources held by members of the GratisNet network are inversely ranked based on historical workload
contribution. Participating libraries are requested
to select from resources in the top-ranked selections
presented, but compliance is voluntary. Participating
libraries supply ILL requests at no charge to members and with no specific reciprocity. The objective of the
ranking process is to adjust the payoff implied by a ration selection of the largest, nearest library by tempering this
choice through ranking of search results based on
previous workload of participating libraries. Libraries with a higher historical workload are ranked lower in search results. Libraries are encouraged to select from one of the first three listed libraries that have holdings in the journal they are requesting.
Table 1 shows the percentage of libraries that by-passed the computer-recommended ranking when raising ILL requests for the years 2002, 2003 and 2004. Since
compliance is voluntary, this change demonstrates
increasing trust in the workload distribution mechanisms. While participating libraries do exercise a measure of
discretion in selecting outside the recommended
rankings, voluntary compliance to the ranking
recommendations is generally good and has improved over time. Table 1 Compliance in ranking selection
Game-theoretic formulations can provide a useful
approach to the design of co-operative IT systems(Mahajan, Rodrig, Wetherall, & Zahorjan, 2004). To illustrate the contrast between a time-efficient system for ILL delivery and one which distributes workload across the network, these same transactions were reprocessed under to a game scenario which simulated a rational select on the basis of proximity and breadth of holdings matching the
request for the most recent two years. The objective of this
scenario was to contrast the aggregate effect of load-based
ranking with a utility-based approach to request ulfilment (see Table 2 below). In a time-efficient approach, larger
libraries are consistently net providers, reducing
their aggregate payoff from participation. Pure
Egalitarianism takes the approach over time that yields the highest combined utility to participating libraries.
The voluntary element of the ranking yields a
“relative egalitarianism” which balances the result that yields utility achieved overall with the lowest level of frustration.(Moulin, 1988).
Table 2 Contrasting load-based ranking
to utility-based ranking The risk facing groups collaborating on a fee-free basis is that inequity of resource distribution could result in the resignation of members where their level of “frustration” exceeds the benefit they gain from participation.
4. Conclusion
Participating libraries in the GratisNet network commit to supplying ILL requests at no charge and with no specific reciprocity, on the basis that they can be confident that an increase in demand on their library will be balanced progressively with a lower ranking in
search results. Transactions for the period 2003 to
2005 are analysed to illustrate the ways in which a
ranking-based approach to resource discovery improves workload distribution for participating members overall.
Results from the GratisNet network illustrate the
effectiveness of formal approaches to resource
distribution in fee-free collaborative networks. This
analysis also gives an insight into the ways in which
service metrics can help in the management of workload in a fee-free environment.
References
Council of Australian University Libraries. (2005). Caul statistics 2003. Accessed 29-nov-2005 from http://www.Caul.Edu.Au/stats/caul2003-pub.Xls.
Fox, E. A., & Marchionini, G. (1998). Toward a
worldwide digital library. Communications of the ACM, 41(4), 29-32.
Hamburger, H. (1979). Games as models of social
phenomena. San Francisco: Freeman.
Hooke, J. (1999). Evidence-based pracitice and its relevance to library and information services. LASIE(Sept 1999), 23-34.
Mahajan, R., Rodrig, M., Wetherall, D., & Zahorjan, J. (2004). Experiences applying game theory to
system design. Paper presented at the ACM
SIGCOMM ‘04 Workshop, Aug 30-Sep 3, 2004, Portland, Oregan, USA.
Morris, L. R. (2004). How to lower your interlibrary loan and document delivery costs: An editorial. Journal of Interlibrary Loan, Document Delivery & Information Supply, 14(4), 1-3.
Moulin, H. (1988). Axioms of cooperative decision
making. Cambridge: Cambridge University Press.
Shubik, M. (1975). The uses and methods of gaming.NY: Elsevier.
Weiderhold, G. (1995). Digital libraries, value and
productivity. Communications of the ACM, 38(4), 85-97.