Wednesday, January 12, 2011

The World Has Had Enough of Silly Presentisms

In the 22 November 2010 edition of Mailbox Monday I challenged readers to check Ancestry.com’s attempts to correct silly place names in their indexes. Readers sent many examples that show that Ancestry.com’s automated abbreviation expansion still produces silly results because context is not considered.

As I read through reader comments, I wondered if English speaking pairs of FamilySearch indexers would ever misindex obvious abbreviations. I decided that a good test case would be a newborn living in Indiana. The younger the baby, the more likely it was born in Indiana, not Iowa.

I searched the 1850 U.S. Census on Ancestry.com for newborns Ancestry.com thought were in Iowa and enumerated in Indiana. I found one and verified from the image that the birthplace was “Ia.” Then I checked the FamilySearch index.

As I expected, FamilySearch Indexers had correctly indexed “Indiana.”

I searched again and found John Houts, 2 months old, born in Ia, and enumerated in Shawnee, Fountain, Indiana. I was surprised that both on Ancestry.com and on FamilySearch.org this newborn Indiana resident’s birthplace was rendered “Iowa.”

Both FamilySearch and Ancestry.com likely misindexed the birthplace of a newborn in Indiana

Presentism

This may be an example of presentism. “Presentism is … historical analysis in which present-day ideas and perspectives are [used for] interpretations of the past. … Historians seek to avoid presentism in their work because they believe it creates a distorted understanding of their subject matter.”1

Present day indexers know that IA is the present postal code for Iowa. They may be applying present practices to a past that was different.

As Kathy Gunter Sullivan put it in a recent edition of NGS Magazine, “Despite being handicapped by twenty-first century experience, we attempt to time-travel [to] meet these… folks who by their choices and proclivities affect our lives as well as the shape of our noses.”2

Vendors Listen Up

Ancestry.com and FamilySearch would do well to heed the warning of Elizabeth Shown Mills. “When we record our findings, we should not alter the language of the records. If we do, odds are, we will change the original intent and distort our portrayals.”3

With this in mind, here’s how the search result for John Houts, the example above, should be rendered:

Record display should avoid presentism

Best Practices:

  • Indexers key exactly what is seen; any interpretation is secondary.
  • Vendors display exactly what is keyed; any interpretation is secondary.
  • The vendor displays any interpretations (such as birth date calculations and abbreviation expansion) inside square brackets. That is the English language punctuation standard for text inserted into an original.
  • The vendor displays the image at the top; it is the primary result of the search.
  • Keyed information is derivative; the vendor displays it second. To minimize vendor interpretation, it is arranged like the original.
  • The vendor uses the same column headings as the original. Otherwise, vendors provide mouse-over popups that display original headings.
  • The vendor provides links to the enumerator instructions.
  • To avoid interpretation of relationships within a household, the vendor displays the result for a single individual within the context of a household.

As Paul McCartney never put it:

The world has had enough of silly presentisms.
I look around me and I see it’s really so.

Quality conclusions don’t come in a minute.
Sometimes they never come at all.
I only know that when I see ‘em,
Quality conclusions are not silly at all.


Sources

     1. Various contributors, “Presentism (Literary and Historical Analysis),” Wikipedia, The Free Encyclopedia (http://en.wikipedia.org : 6 September 2010, 14:35 UTC).
     2. “Late-night Ruminations of a Tired Genealogist,” October-December 2009, 34.
     3. Elizabeth Shown Mills, Evidence Explained: Citing History Sources from Artifacts to Cyberspace, 2nd ed. (Baltimore, Maryland: Genealogical Publishing Company, 2009), 20.

10 comments:

  1. On a somewhat related topic, in the 1960s/70s LDS Four-Generation Project, vowels were to be left out of county and state name submissions. Maine was thus MN. When the IGI was released to FHCs on fiche, the Zip Code had made Minnesota MN. I found several Maine ancestors now mistakenly listed in Minnesota by computer people entering info from these old family group sheets. I wonder if this has ever been corrected.

    ReplyDelete
  2. # Indexers key exactly what is seen; any interpretation is secondary.
    # Vendors display exactly what is keyed; any interpretation is secondary.

    I'm not so sure this is always a best practice for indexing. FamilySearch indexers have indexed La Crosse, Wisconsin at least 13 different ways in the 1875 census of Wisconsin, including Laleropes, Saleiope, Lalewpe, and more ridiculous ways. So how is one who is doing a place-related search supposed to find all those ridiculous spellings? When it is known what the spelling is supposed to be (e.g., from the film title, the manuscript cataloging, etc.), why not correct the nonsensical eyeballing of the indexers?

    ReplyDelete
  3. Anonymous is somewhat correct. The Indexers are supposed to index what they see and not interpret. On the other hand if it is a variation that is unknown or not clear they can index as blank for the searcher to interpret on their own.

    ReplyDelete
  4. I have discovered ancestors born in Indiana indexed as being born in INDIA. Just a not for those who are searching for people in Indiana...check India as well.tably

    ReplyDelete
  5. Dear Not So Sure,

    I should have been more clear. The vendor can display both a keyed value and an interpretation of the keyed value.

    As shown in the example, the interpreted value is shown in editorial brackets.

    Laleropes [La Crosse]
    Saleiope [La Crosse]
    Lalewpe [La Crosse]

    Had the vendor done what you did, return to the source and re-keys with more context, then the better keyed value replaces the keyed value:

    La Crosse
    La Crosse
    La Crosse

    Note these were vendor corrections. User corrections are a different animal.

    -- The Insider

    ReplyDelete
  6. In regards to the Iowa and Indiana 1850 census records State abreviations were the first intial and the last so both Iowa and Indiana had the same intials. And I believe that Iowa may not have been a State at that time, only a territory. It is important to keep in mind the history of States and how the also affect our owm research. I found this out the hard way when looking for my Indiana ancestry with the Ia place of birth.\in 1850.

    ReplyDelete
  7. AI, your presentation is very well put.

    Now if Ancestry.com could only be convinced to drop the made-up intra-household relationships that are not actually given, in extractions and encoding for US Federal Census enumerations 1880+. Invented parent for grandchild of head-of-household and invented spousal relationships are among the errors of this type. The provisions for user-suggestions do not include these items.

    ReplyDelete
  8. I see Ancestry indexing mistakes very often One of my favorites was found recently in the obituary collection in which Jesus Christ was named as the deceased. In reality, it was for one of the "other persons mentioned" who "went home to be with her Lord, Jesus Christ."

    ReplyDelete
  9. I'm fascinated to learn that there is a word - "presentism" - to describe "historical analysis in which present-day ideas and perspectives are [used for] interpretations of the past".

    Some years ago I coined my own term - "chronologism" - to embrace both this phenomenon and the accompanying mindset, which invariably views the past as unsophisticated, backward and inferior compared to the present.

    ReplyDelete
  10. Dear AI;
    The vendor in this case is FamilySearch. I had a lengthy correspondence with them about their silly indexing of La Crosse. They failed to understand the problem. As far as they were concerned, if two indexers agreed that the place name was Saleiope, then Saleiope it was. The fact that one could search for a surname of Smith or a first name of John in the 1875 Wisconsin state census in La Crosse and get NO results (because they were indexed in Saleiope, Lalewpe, etc.) didn't seem to bother FamilySearch in the least. Now that's a silly presentism! --Not So Sure

    ReplyDelete