Wednesday, January 7, 2015

FamilySearch Considers Alternatives to Double-Blind Indexing

FamilySearch IndexingI don’t know how I missed this, but FamilySearch recently talked about alternatives it is considering to the double-blind indexing currently employed by FamilySearch Indexing.1 Currently, two indexers independently key the contents of a record. If there are differences, the record is sent to a third person—an arbitrator—who indexes the record after viewing the information provided by the two indexers.

Katie Gale, FamilySearch spokesperson, gave four alternatives:

  • For easy to index records, such as typewritten records, FamilySearch might employ single keying. Results would be audited and if the quality dropped below some threshold, peer review or arbitration would be added. “Research shows that using this process for easy projects produces an index of comparable quality to the traditional double-key-plus-arbitration process with less than half the work,” wrote Gale.
  • FamilySearch might use single keying with review. After indexing, a second, more experienced indexer will review the results. If he disagrees, he specifies what he thinks the information should be. Then, instead of sending the two values to an arbitrator, both values are saved. When published, the record could be found by searching for either value.
  • For easy to index fields, FamilySearch might single key particular fields. For example, indexers rarely disagree when indexing a sex field. Even when applied to just a few fields, when millions of records are involved, single keying yields substantial savings.
  • FamilySearch might allow exceptional indexers to index without second keying and without arbitration. I don’t wish to pat myself on the back, but I used to spend considerable time researching hard to read names. I’d look at the neighbors, find the neighborhood in another record set, and identify the illegible name. Then it hit me: the arbitrator wasn’t going to put in the same level of effort. My hard work was as likely to be discarded, as accepted by the arbitrator. “As volunteers gain significant experience and expertise, they sometimes reach a point where review or arbitration of their work actually increases the likelihood of introducing errors,” wrote Gale. (Italics in the original.) Audits would be employed to ensure quality.

Look for these advanced methods when FamilySearch releases its new indexing system, sometime next year.


Sources

     1.  Katie Gale, “Magnifying Volunteers’ Gifts: A Progress Report,” FamilySearch [Blog], 4 August 2014 (https://familysearch.org/blog/en/magnifying-volunteers-gifts-progress-report/ : accessed 29 November 2014).
     Image credit: “Media Library,” The Church of Jesus Christ of Latter-day Saints, website (https://www.lds.org/media-library/images/young-woman-indexing-records-889380 : accessed 29 November 2014), search for “young woman indexing.” Copyright © by Intellectual Research, Inc. Used with permission: “You may post material from this site to another website or on a computer network for personal, church-related, noncommercial use unless otherwise indicated.” See “About the Media Library,” The Church of Jesus Christ of Latter-day Saints, website (https://www.lds.org/media-library/about : accessed 29 November 2014). Also, “We encourage members to use the images on the Media Library on their blogs and personal websites.” (https://www.lds.org/media-library/frequently-asked-questions).

7 comments:

  1. Unfortunately, there is no way to correct a transcription error after the indexed record has been published. Currently, other interpretations and identification of conflicts can only be recorded when attaching a record to a person in FamilyTree, but that not always practical, and certainly not visible to other people searching the records.

    ReplyDelete
  2. I have expertise in French Canadian and Hawaiian names. I volunteered to index the 1940 census in New Hampshire, Maine, Massachusetts and Hawaii. I was sad to see that some of the names I knew from the neighborhoods were not accepted by the arbitrator. It made me sad to know that many errors were being made. The fourth scenario described above would work here, except for the fact that I was a new indexer and not at all "exceptional". There would have to be exceptions for volunteers with special talents, not just a long record of indexing.

    ReplyDelete
  3. I've recently been indexing the 1890 NYC Police Census. I'm finding many of my careful indexing efforts have been overridden in arbitration, to an inaccurate result. It's disappointing and somewhat disheartening.

    ReplyDelete
  4. I too volunteered to index the 1940 census in Montana, Idaho, and Oregon. None of the Norwegian, German, or Swedish names I indexed were accepted by the arbitrator even though many times they were my own family members. It was an extremely frustrating experience for me.

    ReplyDelete
  5. The comments about arbitrators changing well-indexed records are justified. I've done both indexing and arbitration and this can be a very frustrating work. Many times, it is obvious that either the indexer or the arbitrator have not read the instructions for the particular project, thinking that if they've read one set of directions, they've read them all. I'm not sure how to overcome this human tendency to think we are experts because we've done one or two projects. It does help to keep in mind that these are indexes meant to help the researcher find the original record to look at. But limiting how many people see a particular record may not improve the quality of the index. At what point does speed of posting batches become more important than accuracy. And at what point does being able to claim huge numbers of people indexing records overcome accuracy. When volunteers are doing the work, the end result is frequently worth just what is being paid for the project.
    I believe that the majority of the problems could be avoided if people would read the directions for each project, and even each field being indexed.

    ReplyDelete
  6. Hi,

    I want to let you know that two of your blog posts are listed in today's Fab Finds post at http://janasgenealogyandfamilyhistory.blogspot.com/2015/01/follow-friday-fab-finds-for-january-9.html

    Have a great weekend!

    ReplyDelete
  7. That is why I stopped indexing. I carefully read and followed the project instructions, and an arbitrator changed the entries to be incorrect. In essence, my time was wasted. AND...when you send for review, you never get any feedback.

    ReplyDelete