Friday, June 28, 2013

Comparing "Old Search" and "New Search" Results on Ancestry.com

There has been quite an outcry on the Interweb about Ancestry.com's decision to retire the "Old Search" feature  from their website at an unknown future date.  You can read a number of posts concerning this announcement on Heather Rojo's Nutfield Genealogy blog - see "Flash Blog Mob."

One of the claims is that Ancestry's "New Search" algorithms provide results that are not as good as the "Old Search" algorithms.  I decided to do a study of both search algorithms using someone in my database (but someone whom I have not extensively searched).  At random, I picked Floyd Leonard Seaver, born 1892.  It's a relatively unique name.  I knew, from an Ancestry Member Tree and earlier research in census records, who his parents were, and his spouse's name (Verna Daniels).  I also considered who a person new to genealogy might search for - typically, a grandparent or great-grandparent.

1)  In "New Search," here is the search form I used on the Home Page - I entered first names, last name, and a birth year:


The Search results (using the "Records" list of results) have 52,872 listed.  The first 20 are shown below (in 3 screens, no overlap):





The results in order are:

*  Public Member Photo & Scanned Documents - Floyd Leonard Seaver (1892-1939)
*  Public Member Photo & Scanned Documents - Floyd Leonard Seaver (1892-1939)
*  1920 U.S. Federal Census - Floyd L. Seaver (born 1893), Solway, Bertrami, Minnesota
*  1930 U.S. Federal Census - Floyd Seaver (born 1890), Emporium, Cameron, Pennsylvania
*  1920 U.S. Federal Census - Floyd Seaver (born 1890), Zeeland, Ottawa, Michigan

*  1900 U.S. Federal Census - Floyd Seaver (born 1894), Bertha, Todd, Minnesota
*  1910 U.S. Federal Census - Floyd Seaver (born 1893), Woodside, Otter Tail, Minnesota
*  Minnesota Territorial; and State Census, 1849-1905 - 1905, Floyd Seaver (born 1894), Bertha, Todd, Minnesota
*  U.S. World War II Draft Registrations, 1942 - David Floyd Seaver (born 1889), Cameron, Pennsylvania
*  U.S. WWI Civilian Draft Registrations, 1917-1918 - David Floyd Seaver (born 1889), Emporium, Penn.

*  U.S. WWI Draft Registration Cards, 1917-`1918 - David Floyd Seaver (born 1889), Emporium, Cameron, Penn.
*  1900 U.S. Federal Census - Floyd E. Seaver (born 1887), Prairie, Keokuk, Iowa
*  1900 U.S. Federal Census - Leonard Seaver (born 1895), Windsor, Windsor, Vermont
*  U.S. WWI Draft Registration Cards, 1917-1918 - Elfred Floyd Seaver (born 1888), Ottawa, Michigan
*  1930 U.S. Federal Census - Leonard Seaver (born 1896), Claremont, Sullivan, N.H.

*  Public Member Photos and Scanned Documents - Lena Siegfried Joe Seaver and son Floyd at Thomber Cemetery
*  U.s. Social Security Death Index, 1935-Current - Leonard Seaver (born 1895), died Claremont, Sullivan, N.H.
*  Vermont Vital Records, 1720-1908 - Leonard Roy Seaver (born 1895), Windsor, Vermont
*  1920 U.S. Federal Census - Leonard R. Seaver (born 1896), Claremont, Sullivan, N.H.
*  U.S. World War I Draft Registrations, 1917-1918 - Leonard Roy Seaver (born 1895) , Sullivan, N.H.

After reviewing those 20, I concluded that Results #1, 2, 3, 6, 7, and 8 pertained to Floyd Leonard Seaver (1892-1939), who married Verna E. Daniels in about 1912, and resided in Minnesota.

2)  The "Old Search" form looks like this (with Floyd Leonard, Seaver, and 1892 in the search fields):


The Results list does not give an umber of matches.  Here are the first 20 matches (3 screens, no overlap):




I won't go through the match list, because they are almost the same.  Missing from the "Old Search" results are the Public Member Photos and Scanned Documents items.

Records pertaining to Floyd Leonard Seaver, born 1892, are # 1, 4, 5 and 6.  They are identical to #3, 6, 7 and 8 on the "New Search" list.

3)  Frankly, I don't see any significant difference from the "New Search" results.  The only differences are the Public Member Photos & Scanned Documents items.  If this was your grandfather, wouldn't you like to have the two photos?  And the link to the Ancestry Member Tree that has more information about him?  I would. [Note:  In this study, I'm not concerned with accuracy of an AMT, only the results obtained.]

4)  Thinking that maybe I just picked someone and lucked out, I did three more studies with essentially the same results.  These results match studies I've done in the past year - there is not much difference between the "Old Search" and "New Search"  algorithms.  There used to be some differences, but now there are not.

What about common name persons?  Yes, those are harder - there will be hundreds or thousands of matches just on the name.  The solution there is to narrow your search - use what you know about birth date, birth place, spouse's name, parents names, etc. to narrow your search in a specific collection.  Using wild cards to account for spelling variations works well also, and sometimes you have to search for no surname, or even no name at all (but using narrowing parameters).  Every serious researcher should know how to do this - and needs to practice it regularly.

5)  This is why I present "Searching Ancestry.com Effectively" all over Southern California.  My next presentation is at the San Diego Genealogical Society meeting on Saturday, 13 July 2013.  At the end of each presentation on this topic, people come up to me and say something like "I didn't know that there were all of those options and features."  I tell them that the Ancestry.com "New Search" algorithm is the most complex and most powerful search engine for genealogy records, and needs to be practiced and learned on a regular basis.  If you are paying for an Ancestry.com subscription, you should be willing to invest enough time to learn how to use it efficiently and effectively.

6) For those readers who think that "Old Search" is superior, I encourage you to do a similar study for several of your ancestral persons.  Please tell me about it, or write your own blog or Facebook post, even if it's just a summary of your effort.

I hope that other geneabloggers will perform similar studies as the one above (you don't have to do everything I did), and demonstrate that the "New Search" capabilities are similar to the "Old Search" capabilities.

The URL for this post is:  http://www.geneamusings.com/2013/06/comparing-old-search-and-new-search.html

Copyright (c) 2013, Randall J. Seaver


9 comments:

Denise Fischer said...

The reason you don't see much of a difference is that you do not have the Exact Matches box checked on the Old Search form.

When you do, the list is much more compact, much easier to navigate and I find what I am looking for much faster.

Geolover said...

There also has been some discussion (some not nice) and an Ancestry staffer's input on an Ancestry.com message board, here:

http://boards.rootsweb.com/topics.ancestry.ancsite/12871/mb.ashx

Andrew Hatchett said...

Denise, the "Exact Match" boxes weren't checked in the NewSearch example so it was a fair test.

Checking them in one test and not is the other is comparing apples to oranges -i.e no test at all.

Geolover said...

Another series of discussions pro and con are here:

http://boards.ancestry.com/2010-Search-Beta/mb.ashx

Heather Wilkinson Rojo said...

Thanks for the mention and link, Randy. I added this post to the list, too.

Randy Seaver said...

Denise,

I had "Exact Matches" unchecked for both the "New Search" and "Old Search" tests. The results are similar with "Exact Matches" checked as one would expect in 2013.

I can do a test with "Exact Matches" checked if you want!

Randy

Randy Seaver said...

Geolover,

I think "New Search" has changed considerably since 2010, so the 2010 comparison may not apply. I think I did a comparison several years ago also.

The Ancestry staffers seem to have the patience of Job...

Cheers -- Randy

Denise Fischer said...

Checking what seems to be the equivalent of "exact matches" on New Search gives results that have the same appearance.

My issue is the extreme amount of empty space presented on New Search, requiring much scrolling to read them all. Therefore it takes longer to scan through and find possible real matches.

When I use Old Search with exact matches checked I am presented with about 5 possible census collection matches, 5 birth, death etc collection matches, 5 military collection matches. This can be clicked through much more quickly.

Randy Seaver said...

Denise,

Are you saying that you are getting a list of matches for your search in "New Search" that are from different databases? If so, you are seeing results on the "Records" tab. If you switch to the "Categories" tab (top of the Results page, on the right) you will see the orderly list of databases within the categories - the 5 census databases, the 5 vital records databases, etc. Does that work for you?

Maybe i'll post about that feature again - I last wrote about it in http://www.geneamusings.com/2013/05/ancestrycom-changes-search-results-page.html

Regards -- Randy