Pages

Thursday, November 5, 2015

How do the search engines stack up?

It has been a considerable time since I last wrote about search engines. The idea is to compare the various search engines with a partially random search and compare their individual results. There has been a lot of water under the bridge, genealogically speaking, since my last post on this subject. According to Google Trends, "genealogy" as a search term has continued its decline.


Although the implications of this dramatic decline may be controversial, there is an overall indication that searching generally for genealogical subjects is not as common as it once was. In previous search tests, I have used the name of my Great-grandfather, Henry Martin Tanner. But in the last test, I changed to another ancestor. I decided to use the name of my paternal grandfather, Leroy Parkinson Tanner (b. 1895, d. 1944). There should be far fewer responses for Leroy than for Henry Tanner than for his father, Henry Martin Tanner.

Most genealogists find themselves using the Internet frequently in their research. Although from the response I receive in my classes during the week, I suspect that few of them use more than whatever default search engine they have selected.

There are several websites that highlight the top 10 or the top 15 most popular search engines. The variety is interesting. For example some lists include Baidu, the popular Chinese search engine, others include DuckDuckGo. Whatever the makeup of the list, the top spot is always occupied by Google. Included in the top 4 are Yahoo, Bing, and Ask. The number five position seems to be up for grabs. By Alexa.com rank, number 5 would be DuckDuckGo but AOL appears in several of the lists. I am going to rule out Baidu simply because I do not read Chinese. Here is my list for comparison. If you want to see how any of the other popular search engines stack up, you are more than welcome to run your own tests.


The difference in the number of visitors per month between these various search engines varies by a factor of 100. For example, Google has over 100 times the number of visitors than Dogpile.com.

I always suggest that researchers check out the name of any ancestor they are researching using a Google search before spending any more time looking online. A quick search using a variety of terms will give you an immediate idea of how much information there may be online about your ancestor and whether or not anyone else has posted any pertinent information.

Here is my initial search term. The quotation marks are included, by the way.

"Leroy Parkinson Tanner" Arizona

Here are the rough results of a search in each of the above search engines:

  • Google.com -- 519 results and it appears that all of them are my grandfather.
  • Bing.com -- 8 results.
  • Yahoo.com -- 29,800 results, but it appears that the search results include references to Parkinson's disease. By the second page of results, the responses are certainly not my Grandfather and it also appears that the search has ignored the quotation marks altogether. I counted 19 pertinent responses until the list got more and more off topic.
  • AOL.com -- 10 results. It looks like to me that people using the alternative search engines are more interested in news and sports than searching.
  • Ask.com -- 9 results mixed in with random other responses on two pages.
  • Excite.com -- Something over 20, but the responses were so random it was difficult to tell how many were pertinent and how many were not. 
  • DuckDuckGo.com -- 36 results, but they were all on target. This site also did not have a huge number of ads and extraneous information such as links to Parkinson's disease. 
  • Lycos.com -- Something over 30 mingled in with non-pertinent results. But we are back to obvious, saturated advertising. 
  • Dogpile.com -- Almost identical to Lycos.com, probably just over 30 pertinent results mixed in with Parkinson's disease and other, mostly advertising, links. 
Actually, the results have been utterly consistent over the years I have been running this simple test. There is a lot of extraneous noise from most of the websites. 

To give some idea of the actual number of online records for Leroy Parkinson Tanner, I did a search for his name in the FamilySearch.org Historical Records Collections and found nine records. But by "matching exactly" his name, the number dropped to two. I might mention that I see 41 sources attached to the FamilySearch.org Family Tree for Leroy.

In my search on Ancestry.com, I found about 30 records, including the ones I had already attached as sources. Most of these were photos.  

I guess there might be some arguments about the validity of this kind of test. I realize that there are some extremely sophisticated techniques that can increase specific accuracy including, but not limited, to using Boolean Operators and other techniques. But once again, I am doing all my searches with Google. I see no reason to bother with the others. 



2 comments:

  1. Tried your search on StartPage.com, got one page with a half-dozen results -- from your blogs and one item from Wikitree.

    ReplyDelete
  2. Hi James - thanks for posting this thought provoking article. Of course I immediately beetled off to Google Trends and started posting all sorts of search terms like family history (not much better although it did seem to show a tiny upwards movement) - family tree seemed marginally better and ancestry was, perhaps predictably, going through the roof. Then I started putting in possibly irrelevant search terms such as knitting and sewing as a point of comparison. It really is fascinating studying these trends. I will continue to Google with renewed confidence.

    ReplyDelete