"One man's wage rise is another man's price increase."
-Harold Wilson

In a world of unlimited resources, users could avail themselves of all potentially beneficial services at their maximal levels of quality. However, in any real digital library, resources---including raw computational resources, human attention, access to intellectual property---ultimately have some cost. (Costs of some resources reflect the denial of opportunity to deploy them in alternate activities within the library. Costs of others reflect those incurred outside the library system.) Indeed, in even a moderately scoped digital library, there is potentially unbounded demand for computational resources. For example, any amount of preprocessing of data in the collections---such as indexing, meta-data gathering, or caching---might improve the response of the system to subsequent user requests. Given only finite resources, however, we cannot take advantage of all such opportunities. We also cannot generally try every method for accomplishing a given task, but rather must choose among those available based on resource requirements and prospects for success.

In approaching this resource allocation problem, we treat the alternative information services as competing economic activities. Agents interact in supplier-producer relationships, where each agent produces value-added information products from the input products provided by others. Agents dynamically connect with each other as opportunities arise for mutually beneficial exchanges. The collections, represented by collection interface agents (CIAs), provide some ultimate ``raw materials'' in this process. Library end users, represented byuser interface agents (UIAs) are the ultimate consumers of the ``finished goods''. Mediator agents (``middlemen'') bridge the gap by bringing to bear knowledge, processing, storage, or other computational resources to improve in some way the expected value of the information as it passes along the chain from agent to agent.

A negotiation and exchange facility for information goods and services must potentially support a wide variety of market types. For example, if one were to sell an electronic version of a magazine, it could potentially be sold in many different ways, such as:

• by subscription, individual issue, or individual article,
• to individuals, libraries, or other defined group,
• prior to paper publication, immediate on publication, or with delay,
• with or without license to further distribute.

The magazine might also be bundled with related services such as searching or notification of new articles.

Each of the different good descriptions, negotiation and payment procedures may have advantages and disadvantages for the various types of information goods and services. Therefore, we provide generic facilities for defining these goods and procedures. Our general method for determining which agents provide services to which others, and under what terms, is negotiation through anauction protocol. UMDL auctions operate by collecting offers and determining agreements consistent with those offers. For example, one simple kind of auction collects bids and settles them by finding a single price that clears the market, while others may perform a complicated matching procedure.

The auctions used in UMDL are based on the same technology that runs the Michigan Internet AuctionBot. The AuctionBot web site has further information about the structure of auctions, and available auction types.