Thursday, August 26, 2010

RE: Elementary errors (Was: Rule of three--gone??)

Posting to Autocat

Bryan Campbell wrote:

Given, as you mentioned, that "the rule of three is beaten into our heads so mercilessly," I would think more comments would see this "error" as a brief and welcome break from current tradition. We have complained for years in this forum about the rule of three; in this case, there should be more cheering, less jeering. Sure, it's an error under the current rules, but honestly, it's not one that I would worry much about because no real harm results from retaining the additional names. If it relieves one's anxiety, switch the 100s to 700s, validate the names, and move on.

As I see it, the real error in practice is that we even allowed the rule of three to persist for so many years. Assuming RDA is implemented, I fear that because "the rule of three is beaten into our heads so mercilessly," too many catalogers will gravitate toward the RDA option that allows them to continue it in practice.

Concerning the rule of three and the problems of maintaining high standards, I think eliminating the rule in three may actually make things more complicated. What about dictionaries? What about wikis? What about serials? What about materials put out by international agencies with 20 or so corporate names on them? I'm sure others out there could come up with other materials. As I wrote in an earlier post, "The number of exceptions to the rule would doubtless skyrocket and the practical cataloger would probably rely on the rule of 'do as many as I feel like doing today, based on the amount of other work waiting on my desk and, let's face it, I had a hard night last night.'" As a result, I believe that something that seems as simple as eliminating the rule of three actually winds up *eliminating* a standard.

Still, none of this means that we can't simplify matters tremendously, increase access *to a point* and not hurt our productivity too much, which I think should be of the highest importance today.

But getting back to the topic, I would also like to believe that current standards are not too high, but reality may be showing us something different. If standards in any field are to work, they must be practical to implement. For example, there must be trained people. The standards must be readily available. You should also be able to discuss problems. There should be retraining made available at regular intervals. All of this requires funding (except for the standards being readily available, which today *can* be achieved for free over the internet, if something like the cooperative cataloging rules would be implemented, perhaps even the need for discussion of problems if the Web2.0 tools were used).

If there is no funding available for following and maintaining standards, or only a small, restricted part of the field can afford it (i.e. research libraries and other special institutions), the standards *cannot be implemented* and it doesn't matter if these are standards are in food production, health care, building construction, or anything else. We must ask: what is achievable practically, and especially, in today's increasingly austere financial climate? For example, it would be great to enact a standard that says that all automobiles should get at least 200 miles to the gallon, and maybe one group of people out there can achieve it. But if it can't be implemented generally in a practical, and financially viable way, that standard will get ignored.

I believe that quality of records is lower (although I would be happy to be proven wrong), and does anybody out there really think that RDA will be simpler?


  1. "Move the 100s" -- wait, were they putting more than one 100 in the record? That's definitely no good just for machine processing and general linking -- part of the point of the 100 is to make a 'citation' form. You definitely can't have more than one 100. Oh wait, I see, more the single 100 to a 700 instead, righto.

    But as far as how many authors to list... can't it be left to catalogers discretion? If more than 3, you can list em in 700 if you judge it useful. And ideally, maybe it'd start out with 3, but then some later cataloger would enhance it, sure, why not. Does having more harm anything?

    [Of course, a very real problem is that you can't TELL if a 700 is actually an author, or just some random 'related person' who may not be a contributor at all. If people used 700 relator codes that might be better. But that's there whether there's ]

    I haven't been around for long enough to say if cataloging quality is getting worse, but I can say that in my experience quality of our shared cataloging records is indeed not very good. You can't really predict if the marc values will mean what they're 'supposed' to. (Of course, even figuring out what they're 'supposed' to mean can require a complicated back and forth reading of fairly difficult language in AACR2, MARC standards, LCRI's, and just talking to catalogers who can't say where it's written exactly, but swear that X is 'right'.)

    We also catalog a lot MORE than we used to, as the quantity of published material goes up drastically. Perhaps we simply can not afford to put the per-record person-hours into cataloging we used to? But I'm not sure I see a higher 'quality' of cataloging records when I look at older records, honestly.

    One of the goals of RDA was to make cataloging easier and cheaper and more flexible, by taking a "principles-based" approach, and using clearer simpler language. I'm not sure anyone actually thinks it's achieved that goal, which is a shame. I swear I had that goal presented to me in an RDA presentation some years ago.

    Incidentally, your RSS/Atom feed doesn't display properly on, which makes it harder to keep up with your blog, would be awesome if you could fix it somehow.

  2. Thanks for the comment.

    1) Concerning citation form, which I agree is important for users, I have never seen any citation guide demand a single author name 100/main entry type of form for a citation. They always want more, some wanting all of them, some limiting to 4 or (I believe) 7. I have written against maintaining a *single* main entry, which is--so far as I am concerned--a relic from the card catalog that does no good, and only brings problems, much as people talk about our appendixes. Still, I maintain that it is necessary to separate main authors from secondary authors. This would demand major changes in MARC and some cataloging rules, however, primarily in the realm of analytics and subjects.

    2) While I have nothing against making additional access points, actually getting rid of the rule of three does not do this. If there were a rule of "at minimum, do the first three" or first five or whatever, this would be setting and maintaining a minimum standard. But leaving everything to how the cataloger feels that day, or to the amount of work on his or her desk, or to this airy-fairy "cataloger's judgment" (which I have never understood what this means) is actually the equivalent to eliminating standards completely since anything is allowed.
    In practice, it would probably mean doing the easy headings.

    Finally, thanks for mentioning the RSS/Atom feed. I've noticed it and tried to fix it but to no avail. Don't know what is wrong, but I'll keep on it!

  3. Ah, in terms of citation form, I didn't actually mean a HUMAN citation form. I meant a 'system' citation form, which is actually part of our legacy pre-machine practices.

    Consider the way we 'cite' a specific manifestation in a 7xx (and some other fields too), in a way that theoretically unambiguously specifies the target of the 'link', without using an accession number. We do it with main entry followed by title (unless the main entry was the title). If there was not a single 'main entry', how would you 'cite' a particular manifestation or work in a 7xx, such that the destination of your link could be unambiguously retrieved from the corpus?

    I think the 'citations' AACR2 is talking about are NOT actually human-language citations, they are instead a more formal and systems-centric notion of "strings that unambiguously specify a target entity."

    Of course, there are problems with these systems-centric citations, they don't always actually WORK for a variety of reasons, and it can be difficult to tell if it's a manifestation or a work being 'cited' (ie referenced, ie linked to). But I'm not sure we'd improve things by eliminating the possibility of creating such a citation-reference altogether, or making it more difficult to do.

    The BEST solution would certainly be using actual modern _identifiers_ for these citation-reference-links. But until that happens.

  4. For the system citation form you mention, I would like to point out that the normal library ideas of an "expression" and "manifestation" are not followed by many others out there. Publishers include many more manifestations as separate (e.g. paperback vs. hardback are the same manifestation in the library world, but cannot be for publishers), while antiquarian book dealers delve far more deeply into differences both in the text and in the physical carriers than does a normal cataloger. This is because the difference of a single letter on a single page can make the difference between a first and second edition (from the antiquarian point of view) and therefore may make a difference in thousands of dollars.

    The relevant library rules are LCRI 1.0 at the Cooperative cataloging rules, the ALA rules at Changes between, differences within.

    Another possibility that libraries could use, and at any rate, should consider is the International Standard Text Code, which is an ISO standard and may do what libraries want. See: