I strongly support Teague Allen’s “do no harm” rule. Don’t create authorities if you are not confident they’ll be correct.
However, let me say that I don’t see that as meaning you have to aim for absolute perfection. Here’s an example of the difference that I think could be helpful. Last weekend I created a personal name authority and added it to a bib record. The bib record is OCLC #14804848, and the authority is for Lenoir, Olivier, ǂd 1867-1922 (no2016069415). You notice, on that authority, that all I put in a 670 is that the heading occurs in OCLC (twice before I used it, to be exact). I did not put anything on the authority from the book, such as its statement that he was the author’s student, because I am not 100% sure it is the same Olivier Lenoir. I think it probably is, because Olivier Lenoir medical books start about the date of this work, in 1899. (To be exact, they start in 1898 with his thesis). I think the creation of the authority is adequately justified by the existence of that heading in OCLC, showing that someone, probably in a French library, had information on the life dates of someone by that name who wrote on medicine. If you’re going to put Olivier Lenoir on the bib record at all, you might as well conjecture that it could’ve been this person. Eventually, a researcher may determine that it wasn’t. That might motivate a cataloger to create an authority for the correct person. But I don’t think the presence of this AAP on the bib record is going to ruin someone’s research. It may lead them down a blind alley at some point, but that’s outweighed by the opportunity it creates for researchers to learn more about the person who lived from 1867 to 1922. For example, it will make them want to find this book and see if it provides more information about him.
Someone might have ideas about things I could’ve added to the authority to indicate my doubts about the identification.
Another thing that I think mitigates the error with the Potomac Canal Company is that a cataloger doesn’t have to use the heading. I can easily imagine a cataloger working on a book and saying to himself, “I see references to the Potomac Company and the C. & O. Canal Company. I notice Potomac Canal Company is supposed to be intermediate, but I don’t see evidence of that in the book,” and deciding not to use it in his bib record. So the consequences of an error may not be catastrophic.
Since the system makes it difficult to change or delete AAP’s, I think Teague’s suggestion that we should be able to mark a relationship as spurious is worth considering. I also think Stephen’s suggestion of the qualifier (Proposed), as well as Chris’s idea that it could only be for subject use, might be good.
First, we have been collecting and recording the names of businesses and business people for 50 years outside of NACO, almost always from primary sources. Years ago we built a data base of all transportation enterprises between New York and Virginia down to the Civil War, compiled from the session laws of the states and official reports, so it was easy to spot a mistake. The existing 670 didn’t really enter into it much, because I had better trust in our data base.
I gave this as a very, very simple example to show what happens when people act in haste outside their area of expertise and why the rule to “do no harm” is a very good one. I was also trying to show how increasing the number and types of linkages under RDA multiplies both the chance for error and the amount of work necessary to correct it. This one was a straight chain of three, and no AAP’s had to be changed. Unfortunately, the NACO data base is full of even more complex mix-ups. I am in no position to do that much. My employers are not going to pay me to do something that has no payoff for them. All I can do is protect my own local catalog.
Regarding Teague Allen’s good remarks, I should add that if you want to have a system that does more than disambiguate books and authors and that is supposed to link up with other expert communities, then things will have to pass at least limited muster with those communities. Unfortunately, I am more pessimistic that this can be done, because such expert knowledge tends to be very dispersed. That is why I seconded Ted’s point that perhaps you are trying to do too much, admirable perhaps, but beyond the means at everyone’s disposal.
Ben, Your observation is spot on.
It sounds like the system is working. Maybe not as infallibly as some would like, but still. Someone created an authority record that was, to the best of their knowledge, accurate. But their knowledge was faulty. So someone else came along, with better knowledge and/or more time to do research, and corrected it. Moreover since the first person kindly left a 670 sourcing their decisions, the latter fellow was able to spot where the error originated.
Acquisitions and Discovery Enhancement