Ancestry.com and FamilySearch.org are both great online family history services. The main difference, of course, is that FamilySearch.org is free with (largely) free access to records (records from Fold3.com being the notable exception). Ancestry.com is a paid membership service – although it provides a good level of free access to information to get budding family historians and genealogists going. There, I got that distinction between the two out of the way.
I’ve found another, and more subtle, difference between the two which I’m about to share. It all about performance. But one boring bit first before I get to that. Understanding this first bit will enable you to get the overall performance point I’m making about these two services.
The power and the value of Ancestry.com and FamilySearch.org aren’t solely based on the sheer volume of records each possesses. The same records and digitized archives can pretty much be found on both.
It’s the behind the scenes stuff that seems, in my regular experience of using both, to be the difference. What behind the scenes stuff? Algorithms and databases. The websites of both services are driven by databases – think of these as ginormous warehouses that contain all of the records and data you access when you do a search on either Ancestry.com or FamilySearch.org. The databases have to be exponentially huge to hold all of that data.
OK, so I know what an algorithm is. But I was finding it challenging to explain this in an engaging and meaningful way. You know, the kind of way that anyone would be able to understand. So I did my usual Google search to see if there was a far simpler explanation. Blimey, the one’s I read reminded me of every boring and dry math and statistics class I ever took. So I’m going to simplify things and boil it all down to its essence. Purists, forgive me! Calling an algorithm ‘computer code’ or thinking of an algorithm as just ‘some sort of computer language’ would be simplifying things far too much. Think of an algorithm as the lovechild produced by a mathematical equation and a written language. Think of it as looking something like: x+y=a-b+Italian. This lovechild acts like your own person courier. Basically, you’re telling an algorithm to go and fetch something on your behalf. In this case, you’re asking it to fetch you data and records about your ancestors.Each service has its own unique algorithm. Just like Google has its own search algorithm – which is unique to Goolge and completely different from the algorithm used by Yahoo or Bing.
When you type in the name of an ancestor in either service’s online search form, the different algorithms used by each service go off to their respective, huge data warehouses. Each has a look around in its own warehouse, determines what data best fits what you’re looking for, and trots back to you with that data in tow. You know, the data and records (census records, birth records, marriages, etc) it thinks is best suited to your search. An algorithm tries its best to determine what records are the most relevant to your search.
Ancestry.com and familySearch.org pretty much have the same kind of warehouses that hold all those records and data. Their algorithms, however, are very different. Looking at it in another way…
I’m going to use the horrors of high school algebra and/or trigonometry to illustrate this concept. You’ll find some illustrative examples of what I mean below to better visualise what I mean:
Think of Ancestry.com’s algorithm as something like: 2+3, [ 0 ]=0, [ 1 ] =m, [ 2 ] =n
Think of FamilySearch.com’s algorithm as something like: 2+2, [ 0 ]=0, [ 1 ] =m, [ 3 ] =n
On the surface, at first glance, they look pretty similar. And they are. But those subtle differences determine what records turn up after you click the ‘search’ button on either service. The quality of the search results is largely due to the algorithm each company uses and the language and coding used to produce that algorithm.
The more I research my non-European ancestors and relations, the more I find that Familysearch.org produces far more accurate and better results. And it’s all down to the whatever algorithm it uses to fetch records back from its data warehouse.
I’ll show you what I mean below. I’ll start with Ancestry.com and then move on to FamilySearch.org.
So….I want to find records for Johann Peter Mattil, born on 16 Mar 1725 in Höheinöd, Sudwestpfalz, Rheinland-Pfalz, Germany and died on 20 Jun 1787 in Thaleischweiler, Sudwestpfalz, Rheinland-Pfalz, Germany. He married Anna Elisabetha Scheffe.
As you can see from the above, I tend to use the ‘restrict to exact matches’ option. I tend to do this with all of the variables where this option is available. And last, but not least…
I selected his gender. I also applied the country filter – in this case Germany since I really only want to see German records.
And these are the results I receive from Ancestry.com’s algorithm:
On the positive side, I did get German records (This hasn’t always been the case. I received US-centric results for a number of other 17th and 18th Century German-domiciled Mattils I was researching). However, none of the nine records Ancestry.com suggested were relevant to my search. All nine were 19th Century records. There are no records suggested for a man who clearly lived and died in Germany in the 18th Century.
In my experience, Ancestry tends to work best within national search parameters. Ancestry.com is robust and accurate for American records. Ancestry.co.uk is brilliant for British records. Do an international search…and the results become less accurate.
And now for the same search on FamilySearch.
As you’ll see from the above, there are fewer search options and filters on Familysearch.org. However, the results its algorithm produces looks like:
Not only did I get results for the Johann Peter Mattil I was seeking…I also received a string of results for other 18th Century Mattils. There wasn’t a single 19th century record suggestion.
The result of all this? Well, for the time being, I’ll be using Familysearch a LOT more for my international record searches. For whatever reason, its algorithm is better suited for the job I need it to do researching non-American ancestors.
Has anyone else noticed any subtle –or not so subtle – performance differences between these two services? Feel free to share via a comment below.