Ancestry's treatment of "both sides" matches.

[geebee-1015]

A pet peeve of mine is how Ancestry treats "both sides" matches.  Such matches do not represent some sort of third side, neither maternal nor paternal. They are matches which are maternal and paternal.

But look at how Ancestry talks about matches which are on both sides.  With one match that I labeled as maternal, Ancestry wrote this: "We think this match is on your Both side, but you labeled them Maternal."  Well, yeah.  I labeled this match as maternal because it clearly is maternal.  The fact that it might also be paternal does not make it not maternal.

What Ancestry should do is this.  As soon as it becomes clear which side a match is definitely on, label it as such.  If a match is also on the other side, add that label as well.  Don't use "both", but instead use "maternal" + "paternal", and include these matches when a customer asks to see a list for either.  This is basically what they did even before "SideView" with respect to my daughter's matches.  They did not use "M", "P", and "B", but "M", "P", and "M/P".  This more accurately reflects what it means to be related on both sides.

Also, Ancestry should offer evidence that a match is on both sides -- not just their say-so.  As it is, I've instances in which neither parent seems to share DNA with the match, yet Ancestry assigns it to both sides.  This may very well be because Ancestry can see that there is shared DNA, but because of Timber (or for whatever reason) they don't show it.

In cases in which the parents have tested, Ancestry should allow the connecting DNA to be shown even if Timber has adjusted it to below 8.0 cM.  I mean, that's actually a problem in itself -- because it represents differential treatment by Timber of the same DNA.  But I've seen this happen again and again.  A match shows up for our daughter, but not for either of us -- even when each parent is directly compared to that match.

In fact, out of my daughter's 41 matches that are labeled by Ancestry as "both sides", 14 of them -- or 35% -- do not seem to share DNA with either parent, which in mine view means they actually should be labeled as "neither side".  To be fair, these folks may very well share DNA with either parent or both, but if they do it would appear that Timber has adjusted the reported amount to below 8.0 cM.

One example is PL.  Ancestry says that PL shares 17 cM in 2 segments with my daughter (19 cM unweighted), with a longest shared segment of 12 cM.  Yet if I compare PL to each parent, Ancestry says: "[PL] is either not a DNA match or has not taken a DNA test."  Obviously at least one parent should share more than 8.0 cM with PL, so it would seem that this is the work of Timber.

I also think that when there are two tested parents, it should not even require that the customer share any of the DNA that a parent shares with the match in order to be considered on that parent's side.  By definition, you are related to 100% of the people each parent is related to -- even if you didn't inherit any of the shared DNA.

WV is a match I labeled as "paternal".  I did this because 13 of our 14 shared matches are on my father's side.  And not just on my father's side, but specifically on his paternal grandfather's paternal grandfather's side.  One of these shared matches in a 1st cousin once removed to me -- the daughter of one of my father's uncles.  Nearly all of the remaining paternal matches are descendants of the parents of my father's paternal grandparents -- as is WV.  So on top of everything else, WV's and my respective trees show shared ancestry on my father's side, and we are in fact 3rd cousins twice removed.

The problem seems to be that -- according to Ancestry -- the single 12 cM segment that he and I share is not from this side, but from my maternal side.  And it is possible that our 14th shared match -- which is supposedly on my maternal side -- is the only one show shares the same DNA I share with WV.  (Or at least, overlapping DNA.)

This might be so, but then I am actually not related to WV on my maternal side instead of my paternal side, but on both sides.  Unfortunately, unlike my daughter I do not have tested parents at Ancestry.  So I can't do a correct comparison between WV and a parent.

[geebee-1015]

Here are some matches that Ancestry puts into the "unassigned" category for my daughter:

JS.  JS is the son of one of my known paternal 1st cousins, DS.  I share 1,048 cM in 34 segments with DS, who is the daughter of one of my father's full brothers.  Our longest shared segment is 111 cM -- so that single segment would make this make "Timber safe" for me.  I share 379 cM in 23 segments with JS, with a longest segment of 70 cM.  Both JS and his mother are correctly identified by Ancestry as paternal side matches for me.  DS is also correctly shown as a paternal side match for my daughter, with 489 cM shared in 19 segments, and a longest shared segment of 91 cM.

Despite this, JS is shown as "unassigned" for my daughter!  The question is why, when Ancestry can't help but know that JS is my cousin DS's son, and my daughter is ... my daughter.  The two of them are not distant relatives, but 2nd cousins.  In fact, their sharing is 152 cM in 8 segments, with a longest shared segment of 70 cM.

The only possibility I can think of is that Ancestry is unable to rule out that JS might also be on my daughter's maternal side.  However, here's what Ancestry says when I attempt to do a direct DNA comparison between JS and my wife:  "[JS] is either not a DNA match or has not taken a DNA test."

This is stupid, and is at least partly due to Ancestry's convoluted way of looking at what it means to be related on a given side.  Being paternally related means that both father and offspring share DNA with the match; being maternally related means both mother and offspring share DNA with the match.  Although Ancestry doesn't tell us, they must know when the shared DNA is overlapping.

They do not have to wait until they can say that 90% of the shared segments are on the same side.  This would only be needful if they're wanting to be sure that no relationship also exists on the other side.  That's irrelevant, because being paternally related does not preclude also being maternally related (or vice versa.

There are not three sides -- "paternal", "maternal", and "both" -- but only two: "paternal" and "maternal".  But again, Ancestry treats "both" as a side.  If a person is related on both sides, then apparently -- in Ancestry's thinking -- it is wrong to identify the match as paternal or as maternal.  This is wrong.  It's actually correct to use either or both, though "both" would be a more complete answer.

JS is not the only close relative Ancestry finds itself unable to label.  He is merely the closest one.  Ancestry is also unable to identify the side for my maternal 2nd cousin BK.  They correctly identify the side for me, just not for my daughter.  They are, however, able to correctly identify that BK's daughter SK -- but not his son CK -- is related to my daughter on my daughter's paternal side.