The National Federation of Advanced Information Services (NFAIS) released a draft code of practice for discovery services in January. “Discovery service” here is used in the modern sense of the phrase, to refer to any of several platforms that empower users to discover library resources across all formats – materials in the stacks, electronic resources, and potentially other local databases.
The draft makes great progress toward documenting the challenges facing the creation and use of a discovery service as well as those facing “discovery” in the larger sense – connecting a user and her information needs with one or more resources. As written, though, it doesn’t grapple with some of the business realities that underlie why some of these challenges exist, and why they plausibly will continue to exist regardless of what may be best for a user and her discovery process.
Challenges for Patrons
Discovery services typically connect a user to an electronic resource from an entirely new environment (deep-linking to the resource from the search results – following a post-Ranganathan virtue of providing the shortest path from user to resource). This can mean that the user bypasses a context where any contractual issues surrounding use would have been present (e.g., terms and conditions available on the “front page” of a database but not available or obvious once the user is deep-linked to a resource). This can mean that a user can access a resource without knowing what limits on use might be (and thus may use the resource inappropriately with respect to the library’s contract with the provider), but it also can mean that sometimes a user can see resources in search results to which she may not actually have access (e.g., if authentication is required to “click through” and the user is not someone who would have access).
Users will also typically evaluate their search results without considering any or all of the following myriad complications:
- Not all library resources may be included in the discovery service (e.g., due to conflicting contractual agreements). Most services currently do not disclose to the user what is and is not being searched, and it is not in the business interests of the service vendor to be transparent about resources that they are unable to include in search results, as such disclosures will tend to reflect poorly on the quality of the service.
- The discovery service may present “stale” search results, depending on the schedule for refreshing the holdings of electronic databases. A user searching for an article published yesterday, or even several weeks ago, may not be able to find it using the service. Again, it is not in the business interests of the service vendor to be transparent when their coverage may not be up to date.
- The discovery service’s ranking algorithms may or may not reflect the user’s needs – and in cases where there is a proprietary algorithm, it is not in the business interests of the service vendor to disclose what factors have what ranking in the algorithm, except perhaps in the most general sense.
- The discovery service may not be able to link directly to actual resources, depending on whether the library’s actual coverage data is known to the service. In other words, the source of the indexing/abstracting may not be directly related to the provider of the resource for that library, and so an open URL (with all of its potential for failure) may be used to try to build a connection.
Challenges for libraries
Obviously, all problems for users are also problems for libraries, as users are the “customers” of the library.
Not including the full range of library resources (described above) is also a problem for the library in that considerable intellectual and financial resources are spent on collection development, including licensing of databases, and so a service’s non-inclusion of some resources has great potential to steer users away from some of the library’s investments in information access. Ironically, as a service’s coverage got more complete, more users would naturally assume that they had searched “everything” and so the resources that were not included by the service would become increasingly hidden.
Libraries also may not be able to evaluate the impact of the discovery service without detailed data about use and, perhaps more importantly, the relative value of various licensed databases. Adding to this, in some cases the service vendor may also be a provider of resources, in which case there will be an automatic and unavoidable conflict of interest in ranking resources for search results as well as a business interest to be less transparent about whether source/provider is a factor in ranking
Challenges for everyone else
The perceived value added by abstracting and indexing services may also be somewhat lost both because their brand may be somewhat obscured (as users may connect to resources directly from a discovery service) and because the ranking algorithms used by services may either obscure how much they depend on abstracting and indexing and/or fail to leverage the value of abstracting and indexing to its fullest. This may seem fairly removed from the concerns of libraries until one stops to consider that abstracting and indexing services are what have powered the creation of the metadata that is powering searching and finding (“discovery”) in the first place.
None of this is to say that these problems are insurmountable. We certainly have seen strategic business partnerships emerge between some of these vendors that defy years of actual or perceived competition. Where there is a desire to address the above issues, there is hope that a way forward can be found. As always, though, the first step will be admitting that we have a problem.
(NFAIS is inviting all member of the information community to review the draft code and submit questions and comments by March 16.)
Rice Majors is an assistant professor and the director of libraries IT at the University of Colorado, Boulder. Submissions for Backtalk should be 850 to 900 words and sent to Michael Kelley at firstname.lastname@example.org