Wednesday, September 29, 2010

Relative Finder, Ancestry.com Style

What kind of product manager designed my alarm clock?

The alarm sounds like a quarter-ton cricket. CHIRP! CHIRP!

Right next to your head, it’s effective, I assure you. There’s nothing like the sound of a gigantic arthropod in bed with you to simultaneously thrust you from bed and chase all remnants of sleepiness away. CHIRP! CHIRP!

And my alarm is just as insistent as an entire summer evening’s orchestra. CHIRP! CHIRP! CHIRP! CHIRP!

Last night I discovered a whole new reason to despise my alarm clock’s designer. The cold, advance scouts of autumn snuck through town the other night, chilling any cricket foolish enough to stand its ground. When the scouts fell back to prepare for summer’s fall, but one cricket remained. Chirp, chirp.

It sounded like the neighbor’s alarm clock was ringing… Chirp, chirp. Chirp, chirp.

All night long… Chirp, chirp. Chirp, chirp.

Incessantly… Chirp, chirp. Chirp, chirp.

But I digress… (chirp).

Last week I looked at a preview of a developing FamilySearch technology that calculates the relationship between any two people in the new FamilySearch (NFS) Tree. I tried Orville Wright and Philo T. Farnsworth, unsuccessfully.

Click on More options and Find famous relativesI chose Philo T. Farnsworth and Orville Wright so I could also try Ancestry.com’s Famous Relatives Finder.

While FamilySearch Relationship Calculator is designed to calculate the relationship between any two people in the NFS tree, Famous Relatives Finder is designed to find a relationship between any person in One World Tree and a predefined list of famous people.

To find famous relatives of someone in an Ancestry.com Member Tree, click on “More Options” followed by “Find famous relatives.”

According to Famous Relatives Finder, Orville Wright and Philo T Farnsworth are 4th cousins, 4 times removed. Gosh, I hadn’t considered a closer connection than their White ancestor, but it was certainly possible.

Famous Relatives Finder displayed the connection:

Common Ancestors Rebecca Sykes (1678-1760)
Siblings [not specified] Benoni Wright (1719-1761)
1st Cousins Benjamin Wright (1660-1743) Dan Wright (1757-1832)
2nd Cousins Mindwell Wright (1694-1712) Dan Wright (1790-1861)
3rd Cousins Mary Belding (1722-1766) Milton Wright (1828-1917)
4th Cousins Anna Kellogg (1755-1838) Orville Wright (1871-1948)
  Reuben Farnsworth (1787-1847)   —once removed—
  Philo Taylor Farnsworth (1826-1887) —twice removed—
  Lewis Edwin Farnsworth (1865-1924) —3 times removed—
  Philo T. Farnsworth (1906-1971) —4 times removed—

 

Oh boy.

Here we GIGO again. Benjamin Wright (b. 1660) can not be the grandson of Rebecca Sykes (b. 1678). Nor is it likely that Mary Belding (b. 1722) was the daughter of Mindwell Wright (d. 1712).

The back-to-back failures of relationship calculators on FamilySearch and Ancestry.com, both because of garbage in their trees, was too much to handle.

I cracked.

All I could hear was that incessant cricket. Chirp, chirp. Product managers come and go, incessantly repeating the mistakes of their predecessors. Chirp, chirp.

Ancestral File. FamilySearch attempts to use automated matching to stitch together thousands of source-less, mistake-riddled pedigrees. As with the examples today and last week, many of the trees fail pedigree analyses so basic that even PAF would complain. Chirp chirp.

The result: Garbage in, garbage out. Chirp, chirp.

One World Tree. Ancestry.com attempts to use automated matching to stitch together thousands of source-less, mistake-ridden pedigrees. Chirp, chirp.

The result? Again, garbage in produces garbage out. Chirp, chirp.

New FamilySearch. FamilySearch tries stuffing garbage into a human-assisted stitcher. Chirp, chirp.

Result: Garbage in, garbage out. Chirp, chirp.

Ancestry.com Member Trees. Ancestry.com tries throwing garbage at human stitchers. Chirp, chirp.

Result: Chirp, chirp. Chirp, chirp. Chirp, chirp. Chirp, chirp…

8 comments:

  1. The first time I saw that another member on Ancestry.com had added my line to theirs, I was confused and thought they had merged or changed my own line. By the next few notices I learned to look carefully at the member's information and sources. I only merged those that matched my own information which I had been collecting over fifty years. I also wrote to new cousins which was a real treat.
    When I saw that some of the information was absurd, I politely wrote and pointed out that a person born in the late 1500s could not have been born in the New England colonies as they did not exist prior to 1620 and later. I was thanked by some of the members, and am under the impression that some were adding an entire line without checking each and every bit of information.
    Having once been a genealogy registrar, there were guidelines we used - a child should have parents who were alive (fathers at least nine months earlier) and a minimum of twelve to fourteen years old (early Quebec). A mother should have to be less than fifty years old and her children should fit into a conventional birth order.
    One printed genealogy merged an ancestor with her own mother. I wrote that it was unlikely that a woman would be having children in her seventies or live one hundred and twenty five years despite the cold Quebec winters. My sarcasm was ignored. ETM

    ReplyDelete
  2. AI, well said.

    Happily, Ancestry.com's OneWorldTree has been abandoned. The compatible Tree software was cut loose a few years ago and OWT is not "updated" to incorporate the new-format "Member Trees" additions.

    However, Ancestry.com is still encoding what its software thinks are "same individuals" in trees and in selected databases (including the junk ones based on family group sheets, IGI, and assorted unnamed genealogies).

    So the potential remains for re-creation of an even larger OWT-type swamp. In the past few years I have noticed addition of ridiculous items in Member Trees to some of my ancestral lines. So if some corner of Ancestry.com is working on OWT-type compiling software compatible with the Member Tree format, there is potential for re-creation of the same sort of folly, with even more GIGO.

    The only reasons I can see for doing such a horrorshow is either competition with newFamilySearch, or some sort of integration with it. Race to the bottom, anyone?

    ReplyDelete
  3. Sorry you got out of the wrong side of the bed. But you're absolutely right. As wonderful as the new connecting technology is, I believe it is incumbent on the primary institutions to identify all of these linkages as hypothetical constructs at best, and particularly label all un-sourced compilations as of no value. Each should come with a "WARNING".

    ReplyDelete
  4. Bravo!

    Why Ancestry even continues to allow access to that mis-begotten collection of junkology they call OneWorldTree is completely beyond me!

    ReplyDelete
  5. I don't merge families even if I know the person that posted the info. I use the Ancestry Connect as a research tool, I go find my own sources and if the person who posted the info didn't also post the source I contact that person and ask what was their source. If no reply, I delete them as a contact. I keep all my info and sources on a software program and only use Ancestry as a tool.
    E. R

    ReplyDelete
  6. I am NOT a Praying MantisSeptember 29, 2010 at 10:41 PM

    A.I. ..... perhaps your best chirp ... er, blog ... ever!

    ReplyDelete
  7. Reading genealogy work without sources is like trying to cash an unsigned check. My research is just a major hobby, but a friend in the LDS church was horrified when she signed into New Family Search to discover "merged" families that were filled with errors.

    ReplyDelete
  8. It doesn't sound like there's much optimism here. I think the One World Tree was flawed, but the theory is sound. I've encountered people in my own family that have discounted sources and evidence for their own gut feeling. Some people take it very personnel when you try and change a family record, even if an undisputed source shows a discrepancy. However for the most part I think a universal tree can benefit everyone. Error's in many family tree's could be corrected if multiple people were scrutinizing and adding sources to a shared tree. There's always going to be people that won't agree regardless. But over time I think reason and debate will weed out much of the bad information. The question is; who's going to create the best database to enable that kind of shared family tree? WeRelate.org seems to be the model that family Search is looking at copying. The problem with WeRelate.org is the lengthy process it takes to upload a Gedcom. It discourages people from contributing because of all the steps one has to take. But I do think it's on the right track. Ancestry might consider a shared tree based on the their "Public Member Tree" model. Allowing members to keep their own trees while contributing to a separate shared tree. They may have to compromise on how to make money on such a tree though. Especially if FamilySearch.org ever gets their act together.

    ReplyDelete