PCC-Standing Committee on Standards Response to CPSO Proposal on "Additions of Dates to Existing Personal Name Headings"

Dates for persons

1. Existing headings with open spans
Users
Many thanks to Alva Stone for so articulately raising on PCCLIST what should be the first aspect of any potential change that we consider: impact on users. As date spans for persons are commonly used beyond the library catalog, it is natural that users carry their existing knowledge and expectations of how date spans are used with them when they use a library catalog. In nearly all contexts, an open span means that that person is alive, to the best knowledge of those providing the span. Therefore it is to be expected that for most users open spans for persons known to be dead are inaccurate, plain and simple. It is condescending to our users to refer to this as a “cosmetic” issue. If we deem flaunting standard English language conventions as merely “cosmetic”, then we should not worry about capitalization of names, ignoring initial articles when sorting, placing surname before forename, etc. Users come away with the impression that those of us who build the catalog are lazy, stupid, careless, and/or not interested in actually serving them. And even when the reason for not closing such a span is explained, they do not find our practice helpful or rational. One PCCLIST member wrote that this state of affairs is embarrassing; it should embarrass us all. (_Any_ cataloging practice that needs to be frequently explained should be reconsidered. There are some that we may need to keep for now; we still need to communicate better about them.) We need to capitalize on what users bring to the catalog rather than contravene it.

This one issue will of course not bring ruin to libraries tomorrow; after all, we've had open spans for years. But unchecked, the attitude of many in the profession that what our users think is unimportant will bring the undoing of the profession. Enough poor user experiences (in particular that searching in the library catalog takes longer and is more difficult than searching in Google) are already leading toward a worst-case situation for libraries: users (and funders) perceiving us as irrelevant. This brings real negative impacts: problems in recruiting into our field, lack of trust in us, erosion of political and financial support, etc.

It was mentioned on PCCLIST that the number of complaints from users about inappropriate open spans has been very few. Studies in multiple contexts have shown that the number of people who don’t like something is many times larger than the number who report it. Furthermore, it is our professional responsibility to improve service not only reactively when users complain, but also proactively, using whatever we know about users, technology, external trends, etc.

One PCCLIST member wrote that “the potential loss of information and access for the user in the case of an unrecognized heading split is surely greater” than leaving closable spans open. On the contrary, the harm done to users’ perceptions of us when we display inaccurate data is likely greater than that done by having two headings (with the death date and without) that will just about always sort adjacently in search results, and will appear to just be a typo.

Heading stability
Much of the discussion on PCCLIST has focused on the need for stability of headings. One early comment was: “The universe of shared cataloging and shared authority work is built on the assumption that headings will we as stable as possible.” Some library folk may indeed have this assumption, but it is not a basic, intrinsic principle of shared cataloging. Stability of headings is not a goal in and of itself. If it were, then we should just use a randomly assigned identifying number for each heading, rather than the name, dates, etc. We don't do this because we want the heading itself to be meaningful to our users. We try to reduce headings changes because of our workload, not because changes are per se against the best interest of our users. There is a clear parallel here with bibliographic records for serials: some CONSER records change frequently, yet remain successful in a shared context, even though a change in one record sometimes requires changes to many other records. Rather, a principle would be that we all need to agree on a common set of expectations (i.e., approved standards) for records we want to be shareable, including an expectation that an indication be provided of which standards are followed in a record overall and in its various elements: AACR2 vs. earlier rules, LCSH vs. MeSH, full vs. core level, etc.

Workload
Philosophically, providing accurate, useful data to users needs to trump minimizing our workload. Period. That is not to say that we can or will immediately implement every change that will bring about better service to users. What it does mean is that our perspective must always include the user. This perspective needs to permeate our planning, prioritizing, communication, etc. For example, if a user asks one of us today why so-and-so's dates aren't closed, rather than saying that we don't do that because the heading is already unique, we might say:
Yes, we know that's a problem;
we're working on getting funding for additional staff to maintain these headings.
[or] the software that runs our online system find-and-replace yet, but that's coming in the next release due out in 8 months.
[or] we're working on a contract with a vendor to do that work, which should start by next year.
Thanks for your patience!
If providing certain data accurately is deemed not feasible for the time being, then we need to consider not providing it at all. It is sometimes better to provide no data (for a particular element) than to provide inaccurate or misleading data. For example, LC and NACO policy on 665 corporate body history references is to delete them when the record is updated for other reasons, because we haven't had the resources to maintain them, and the data is becoming less and less up-to-date over time.

Technology
Another theme in the PCCLIST discussion has been the use of technology to manage headings changes. Clearly the know-how to do this exists, and the functionality has existed in some systems for several years. Restricting policy not just to what is achievable, but, more narrowly, to what can be achieved by all libraries at this very minute, is short-sighted. It is inappropriate to hold back the large number of libraries that have chosen to spend resources on database management capabilities, whether via global change in an ILS, outsourced authority control services, or staff time to change records manually. Those that have not are likely to already have sufficient inconsistency in their catalogs that not updating existing bibliographic records with death dates will not be a significant decrease in quality. It is often through developing standards (such as the NISO and MARC holdings standards) that are not fully implementable by everyone on the day they are approved that gets us the technology we want and need. ILS vendors are understandably more likely to develop their systems to conform to national or international standards than to satisfy individual customers’ desires. We wholeheartedly agree with the statement on PCCLIST that “If we wait until everyone is in a position to change we will probably never adopt anything new since that time is unlikely ever to arrive.”

Interestingly, changes to headings appears _not_ to be a significant concern for administrators, at least in large academic libraries. When Jennifer Bowen (ALA's representative on JSC) discussed AACR3/RDA with the Technical Services Directors of Large Research Libraries Discussion Group (“Big Heads”) at ALA last month, one message she got was that they were quite open to improvements in the rules that might lead to changes in headings, in particular because technology has come so far since the introduction of AACR2.

The suggestion over the years to have a single “mega-union” catalog with highly useful database management tools certainly does have its appeal. It may not be achievable for years, but it does seem worthwhile to begin exploring its desirability and feasibility. The PCC Standing Committee on Automation would be one group that could work on this.

Change
It is puzzling why there is so much consternation about these potential changes to personal name headings. There have been many headings changes over the years that have required significant database management, which may have felt overwhelming at the time, but which we have weathered, and come out with better user service. Recall some of the other substantive changes to headings we have had in recent years: qualifiers for British place names, split LCSH headings (Labor and laboring classes, Nurses and nursing, etc.), changes to MeSH subheading practice, pinyin conversion. Were they overwhelming to contemplate? Yes. Did they take a lot of effort? Absolutely, but less effort than will be needed to close date spans. Did we survive? Yes, and relatively unscathed. Are our users better off than they were before? We certainly hope so.

We are currently undergoing major changes to headings for U.S. Indian entities and government-designated parks and forests, and NLM is considering distributing records with faceted MeSH headings (which is how it stores them), which may lead some medical libraries to convert their existing records. We will survive these efforts and many others in the name of better user service.

Many environmental factors that impact the library world have changed and continue to change. Many of our users utilize Google and the web to the exclusion of our catalog and resources. Budgets remain flat or shrink. Users’ time to devote to information gathering is decreasing. This is a very dangerous time to not be making our catalogs more user-friendly. In the end, it matters not what we are trying to achieve, but rather what our impacts on our users are. Libraries will either evolve or die.

Recommendations
The PCC Standing Committee on Standards (SCS) strongly supports allowing the closing of open spans in personal name headings by adding death dates. We feel it is appropriate and worthwhile to make such changes to headings as:

current: Warhol, Andy, 1928-
updated: Warhol, Andy, 1928-1987

As has been pointed out on PCCLIST, trying to restrict this to “famous” or “important” or whatever people is a futile exercise; workloads of NACO contributors will serve as natural limits to the number of headings changed. Another option that should be allowed judiciously is to remove the open span (whether the death date is not known or not) when the removal does not create a conflict.

The superseded open-span form will be helpful for database management, but not as a displayed reference to the closed-span heading. We would like to see the open-span form in the record, but coded not to display:

400 1 $w nnea $a Warhol, Andy, $d 1928-
rather than:
400 1 $w nne $a Warhol, Andy, $d 1928-

2. Conflict between existing headings and new names being established

Uniqueness in headings is a core principle of library authority control. When a new heading would otherwise be the same as an existing heading, AACR2/LCRI practice has been to add information to the existing and/or new name to differentiate the headings. Given the need to close open spans when a death becomes known, and the fact that dates are often _not_ very helpful to users as distinguishing data (for example, authors born 3 years apart, one of whom writes poetry, the other writes on Korean history), it may be time to revisit the preference of dates over all other ways to distinguish headings. When dates are used, as others have noted, a birth date could be identified by “b. ” before it rather than by a hyphen after it, which would obviate the need for closing open spans. However, that introduces language of the catalog into the heading, which impedes international sharing of headings, the convention may not be understood by users, and it causes complications for sorting (although it would seem wise to pursue the possible use of non-sort characters around “b. “, etc.).

3. Existing headings without date spans, and new headings unique without additions

Discussion
The current practice of using dates in new headings when not needed for differentiation (as a conflict “prophylactic”) has its advantages and disadvantages. It has been pointed out on PCCLIST that the time spent to close open spans in authority and bibliographic records when those people die may be more than the time spent to add dates (and/or another qualifier) to an existing date-less heading when it conflicts with a new name. Adding dates to an existing heading that doesn't conflict with a new heading requires more database management, in bibliographic records that have used the heading. An additional factor to consider for date-less headings is that each time a NACO cataloger runs across date (and other potentially qualifying) information in a resource they are cataloging, that cataloger will need to skim through 670 fields to determine if the information has already been added to the authority record.

The date information clearly does have some value beyond heading differentiation. Library staff promote bibliographic records as providing authoritative and relatively full data about our bibliographic resources, and we strive to live up to that. It is natural that users would come to expect the same for any data they see in our catalog; we need to know what data they do indeed find useful. On the other hand, the library catalog cannot be all things to all people.

Recommendations
At the most basic level, PCC SCS supports the existing practice of supplying potentially distinguishing or otherwise helpful date (or other) information in cataloger-oriented 670 fields, when a cataloger comes across the data in the normal course of cataloging. As there is a lack of consensus at this time on what all the purposes of authority data should be, it would seem to be best to move forward with the narrower purpose until and unless consensus does coalesce on a broader one. That is, we should continue the practice of not changing an existing heading just to add date information, unless it matches a new non-distinguishable name being established. And we should change existing practice, so that we no longer add date information to a new name, unless it matches an existing heading. At the same time, it seems reasonable to allow NACO contributors to add information deemed useful to users to a note field for display in the catalog. Given that the 678 field does not seem to meet the criteria for a distinct note field (see The MARC 21 Formats: Background and Principles, section 6.6, at http://www.loc.gov/marc/96principl.html), it would seem that it should become obsolete, and the 680 be used instead.