Ancestry matches by parent update Aug 2023

[Andrew Lloyd]

they can make use of the relationships that members have manually entered, and the Common Ancestors hints

I have about 5 instances of Common Ancestors where my sisters sideview assessment is Paternal Side and My sideview assessment of the same common ancestor is Maternal side. A similar number of disagreements between me and my brothers assessments but different common ancestors. I like to think that in these instances the phasing algorithm hasn't quite worked correctly.
I also like to think that they are not using the common ancestors hints to determine the sideview results, as this introduces a degree of inaccuracy as common ancestor hints depend on members trees and as we all know members trees are often inaccurate.
I have taken a note of all the instances where there is a discrepancy with my siblings results and will recheck when there is a new update of Sideview results

 

[peter]

I have about 5 instances of Common Ancestors where my sisters sideview assessment is Paternal Side and My sideview assessment of the same common ancestor is Maternal side. A similar number of disagreements between me and my brothers assessments but different common ancestors. I like to think that in these instances the phasing algorithm hasn't quite worked correctly.

That's likely (though it could be that, unbeknown to you, you are related to them on both sides of your tree). Are they recent matches, or do they predate the revised system?

I also like to think that they are not using the common ancestors hints to determine the sideview results, as this introduces a degree of inaccuracy as common ancestor hints depend on members trees and as we all know members trees are often inaccurate.

The chance of Ancestry coming up with a Common Ancestors hint which links you to a match on the wrong side of your tree must be very small - it's hard to see how it could happen at all, unless some of your parents' ancestors had the same surnames and came from the same area (in which case you might really be related to your match on both sides).

 

[Andrew Lloyd]

Are they recent matches, or do they predate the revised system?

They all predate the revised system.
So for example I have a common ancestor RK whom I have validated as a 4th cousin 1x removed on my mothers side. For me RK is classified by sideview as maternal side, whereas for my sister RK is classified by sideview as paternal side. Being a fairly close match we both have shared matches for RK. I have 8 shared matches for RK all classified as maternal side. On the other hand my sister has 5 shared matches (all within my 8) where 2 are classified as maternal side and 3 are classified as paternal side.

 

[peter]

They all predate the revised system.

Let's hope that the new system proves more consistent and more reliable - not that I've noticed any major problems with the original system (though I only look at how Ancestry assigns a match if it is relevant to my researches).

 

[Andrew Lloyd]

So we now have another update to Matches by Parent (Updated Dec 2023).

I cant see that any my conflicting matches (as in previous post) have been resolved. However the number of Unassigned matches has reduced by about 10% (to 44%) which is encouraging, but this is still nowhere near the proportion of unassigned new matches (<5%).

Lets hope the next update whenever it comes continues the trend of reducing unassigned matches.

 

[Stuart]

My December 2023 update has now arrived, and the "by parent" page has a sticker saying "2688 labelled". I'm not sure since when, as it's not quite the number I get. But it's still more than half of them - reducing the unassigneds from over 23% last week to under 9%.

As to whether the process that assigned these is as good as the old process that couldn't - I can't see any way to find out. All I do know is that the six unassigneds from the last ten weeks are, unsurprisingly, still there. And we still have no explanation of the big difference between the fraction in the full list (8.8%) and in the last ten weeks (1.8%).