A Segment-ology TIDBIT
To understand how it works, we need to think like the ThruLines algorithm does. I’m pretty sure the algorithm has three parts: 1) start with two DNA Matches (you and someone else); 2) find a Common Ancestor in their Trees*; 3) find and report the path from the CA down to you and the path down to your Match.
The TL algorithm has no way to check to see if the ancestral lines back to the CA are correct or not (witness the TL reports that report incorrect genealogy). It doesn’t know if the lines are even biologically correct or not – it relies on what you and your Match have entered into your respective Trees. TL also does not check spelling of the names of the descendants from the CA – you could list your grandfather as Rumplestiltskin JONES, or Bio Father, and that is what TL will report.
Also note that even if all the information in a TL report is correct, there is no guarantee that the identified CA is the Ancestor who passed down the shared DNA segment. As reported here, I found at least 10% had a shared DNA segment that did not come from the CA.
Edit 20190801 *sometimes other “3rd-party” Trees are used to complete the linkage from a Common Ancestor down to you and/or your Match.
[22AO] Segment-ology: How Does ThruLines Work TIDBIT by Jim Bartlett 20190801
Pingback: AncestryDNA and Other Goodies (a test covering this will be given at the DNA Sig) | Monterey County Genealogy Society
I’m pretty sure step 1 is the trees, not DNA matches. I’m from the Netherlands, where few people test with Ancestry. Most of my Thrulines don’t have any DNA matches in them but just show the ancestors, mostly the same information that I already have in my tree, sometimes replaced by another person’s incorrect tree
Yvette, Interesting… All 1,820 of my ThruLine Matches are DNA Matches – every one of them shares some DNA with me. When I’m reviewing my AncestryDNA, I don’t ever see someone who is not a Match. However, I haven’t fully checked all of the folks involved in a ThurLine link… Have you taken an AncestryDNA test? Do you have ThruLine Matches who are not DNA Matches? Jim
Yes, I took a DNA test, I don’t think there’s a way to get ThruLines otherwise. It sounds like you’re looking at your DNA match list. I’m talking about the ThruLines page, accessed from the third block on the main DNA page at http://dna.ancestry.com.
When I go to the ThruLines page, I see all my ancestors there up to 5th great grandparents. When I click on them, I see how I descend from them. For a few, I see DNA matches, but for most I don’t see any DNA matches.
Good, I think we are in 100% agreement. I, too, see all the Ancestors I have in my Tree out to 5th Great Grandparents. In a few cases I don’t know my Ancestor at that level, and in some of those cases Ancestry list a potential Ancestor that they have figured out. In a very few cases, there are no DNA Matches listed. In one case there was only one child, so no cousins. In all of these cases, when I click on an Ancestor, if there are Matches listed, they are all DNA Matches. My maternal grandmother’s parents were 1850s immigrants from countries with few DNA test takers – so those are the lines with no DNA Matches listed. ALL of my lines with Colonial VA ancestry, have many ThruLines DNA Matches listed. This is not an issue (or any problem) with the AncestryDNA test or ThruLines – it’s an issue cause with my ancestry, and the dearth of cousins on some lines, and/or in the cases with good cousins who did the DNA test, but have not linked to any kind of Tree. ThruLines depends on Tree links, not DNA links, to determine the DNA Match lists. All we can do is encourage our closest cousins to build even a small Tree – I’ve gotten ThruLines Matches with people who only list their parents…. Jim
At least in my experience, a majority of the thru-line links do not have a common ancestor nor any overlap in the trees. Just as hints on sources and “possible parent” on trees are given, they seem to be using a hidden merged tree. Very often my links are via others trees to that match. Sometimes correct and sometimes not.
My biggest gripe is that you cannot tell it to ignore the bad link in a tree like you can a hint. I sometimes have research threads in my ancestry tree that i later disprove. Even though my tree is hidden, it gets picked up as “possible parent” for others who then copy into their tree without any reference. I subsequently delete it from my tree (and add a comment as to why disproven). But it then keeps coming back as suggested parents and thru-line connection. Arggh!
I’m a little confused because I thought all ThruLines were formed on a Common Ancestor. And each one has some rationale to back it up. I acknowledge that sometimes the rationale is incorrect, but it’s not just a wild stab in the dark. I’m fortunate in that I have over 45 years of research done – so I have some good references as to what is right and what is wrong. My experience, so far, is that it’s mostly correct.
I agree with you that we need a way to shut down links that are clearly wrong. Jim
Hi Jim … I am not sure you have step 2 correct. It seems to be more complex than that.
I have a number of Thrulines where the tree of the DNA match being compared doesn’t go back as far as far as the common ancestor being identified. Nor do we have people in common in our trees. However, they do have sufficient in their tree to link them to other trees, some of which do go back to the common ancestor.
So in step 2, the algorithm seems first to gather up trees with the ancestor in (regardless of whether there is a DNA match to the tree owner or not) and then looks for commonality between those trees and the tree of the DNA match being compared.
Thanks for all your Tidbits … really helpful and I enjoy reading them.
You are correct. ThruLines sometimes uses several Trees to “fill in the blanks”. And the concept of how TL works is still the same, looking for any linkage between you and Common Ancestor and then back down to a Match. As I’ve reviewed some of these links, and determined through my collection of notes over the years (and sometimes some additional bdm and census research), I’ve added these links to my Tree. Soon the EVALUATE tag goes away and TL just uses my version. But often there is interesting records and other clues in the side bar of other Trees – lots of innovative help from TL (along with a few bloopers). Thanks for your feedback, Jim
Yes … I couldn’t agree more about the usefulness of Thrulines when considered in concert with other sources of info and a good dose of critical inquiry!!
I have even found some of the bloopers useful … sometimes if you look hard enough at why the algorithm got it wrong, it can lead you to see what it missed.
I have been able to add so many collateral lines to my tree as s result of Thrulines … and of course the more you add the better it gets!!