Tuesday, August 5, 2008

Making Sense via Motives

Last time I pointed out that Ancestry.com's motive is financial: shareholder value, revenues and profits. Owned by The Church of Jesus Christ of Latter-day Saints, FamilySearch's motive is religious: eternal families and temple ordinances. Today let's look at the terms of the deal in terms of the motives.

If you need to refer back to the text of the announcement, it can be found on FamilySearch.org and the clarification can be found here at the Ancestry Insider. As usual, this is my opinion, not confidential information from either organization.

Images

Ancestry.com's old images will not be as good as FamilySearch's new scans. Ancestry.com created its images back when modem's were much more prevalent than high-speed connections, so it scanned the images at lower resolutions to keep download times smaller. Plus, FamilySearch is scanning first-generation ("master") microfilms while Ancestry used second-generation (copies of master) films. (See "Census Image Quality.") Presumably, if FamilySearch is scanning first-generation microfilms, then NARA must be a partner also, providing access to the master microfilms. It appears that Ancestry.com will bear the cost of hosting the images and paying for the bandwidth to serve the images.

Ancestry gets to offer the better images via subscription. The Church gets more temple ordinances by offering the images for free at family history centers, which are frequented by Church members and genealogically-active members of the public. NARA gets the better images for free public use at their facilities. Ancestry.com and FamilySearch (and NARA) all get what they need for their objectives.

Indexes

While in these deals we speak of indexes, which are document finding aids, it may be more descriptive to call them transcriptions and abstracts, the information extracted from a document. (See Indexes are not transcripts for the difference between transcriptions and abstracts.) And instead of keying we should be talking about indexing. But I digress...

The work of both organizations will be used to form enhanced indexes that will be given back to both organizations. For census years that FamilySearch has already indexed, the two organizations' indexes will be combined. It was not clear from the announcement which organization would bear the cost of this surprisingly expensive operation.

Ancestry.com's indexes have good and bad points. The law of diminishing returns would predict that as the accuracy of a census abstractions approach 100%, it becomes more and more expensive to achieve greater accuracy. So the bad point of Ancestry.com's census indexes is the limit that profitability places on accuracy. The good point is that Ancestry has allowed customer corrections for quite some time. I don't care whether your indexer is a Chinese professional or a random American volunteer, a user with intimate knowledge of a family's names is going to produce the most accurate names. I understand that Ancestry is supplying these user corrections as part of the deal.

For census years that FamilySearch has not yet indexed, FamilySearch can use the Ancestry indexes as the A keying and its own volunteers for the B keying. This cuts its A/B keying effort in half, allowing it to focus its volunteers elsewhere. FamilySearch could also use statistics from its A/B arbitration system to compare the quality of Ancestry's indexes to its own. If the quality is high enough and FamilySearch is not keying additional fields, they could skip the B keying altogether.

Again, Ancestry gets money from the indexes by charging for subscriptions (after offering them free for an unspecified length of time). The Church gets more temple ordinances because the indexes are available for free on FamilySearch.org, for free in NARA reading rooms and for free for a limited time on Ancestry.com.

If the Church could offer the images for free at home to Church members and the genealogically-active members of the public, it would. And it will. Next time I'll look at a new concept announced in the 28-July clarification: the notion of a FamilySearch Member. In the mean time, if you want to help produce a free index of the 1920 U.S. Census, visit www.familysearchindexing.org.

2 comments:

  1. My dear friendly Ancestry Insider.

    To call the FamilySearch/Ancestry census indexes "transcriptions" is perhaps not as accurate as calling them "abstracts". My understanding is that:

    Transcriptions are word-for-word from an original document.

    Abstracts are a few words from an original document.

    My thinking also differs from you when considering that the basis for some of the Ancestry census abstracts are the old AIS census indexes which were fraught with errors because individuals typed into a computer after looking at the microfilm on a reader. Much can be lost as one turns one’s head.

    I knew one such paid AIS employee -- the mother of 2 kids in diapers, with a husband working as a 1st year physician at Walter Reed Army Medical Center. Given all the constraints of that woman’s lifestyle, I can only imagine it would be humanly impossible to be as accurate as a double, blind data entry abstract such as offered by FamilySearch Indexing.

    Affordability of the workload is paramount in any business endeavor, and AIS certainly performed well given the technological constraints of the time period.

    YES, I give you the point that individuals have been able to make corrective updates to Ancestry's census abstracts but realistically, that is probably a small percentage of names compared to the millions enumerated each census year.

    If there was a method for extracting only THOSE census corrections, and combine it with the mighty darn good abstracts provided by FamilySearch Indexing (FSI) might really be a good match.

    In fact, as an arbitrator with FSI, I am not looking forward to working through any such combination of Ancestry indexes & FSI first run abstracts.

    In my view, Ancestry got the better end of the bargain, unless you consider that the scanned census images will again be freely available through the 4,500+ LDS Family History Centers throughout the world.

    THANKS for all you do to keep the genealogy community THINKING!

    Just sign me one of your fans!

    ReplyDelete
  2. My dear friend Myrt,

    Thanks for the correction. I'm more of a genealogical technologist than a technical genealogist, so I'm glad you're checking up on me.

    It was a mistake for me to try and address the ambiguous and sometimes mistaken use of the terms "index" and "indexing" at the same time that I was addressing the Ancestry/FamilySearch deal. The terminology error just sealed my failure.

    Not one to learn very quickly, I have fixed the terminology error but not given up my attempt to make the tangential point. Perhaps by sealing the rat hole, the real point of the tangent will be visible.

    That is fascinating information regarding the history of the AIS census indexes. While these indexes are still available on Ancestry.com (click here), Ancestry.com completely re-indexed the U.S Federal Census. The AIS indexes are no longer the basis for census search results.

    While I have a high regard for the FamilySearch indexes for census pages that Ol' Myrt' arbitrated, my thinking differs from you when considering the experience level of some volunteers.

    I know one stake that made (and achieved!!) a goal of abstracting one million names in one six month period. I chaperoned many nights at a family history center stuffed with young people who could no more read cursive than they could read Chinese. I also fought a continual battle against those who wanted to jump right in without going through any training lessons. I'm sure you've arbitrated many a batch that illustrates these problems.

    Fortunately for all, the best of both indexes will be combined for the enhanced indexes, which will be free to all.

    -- The Ancestry Fan-of-Myrtle Insider

    ReplyDelete