Check for new replies
Ancestry wishlist
#16
(07-23-2024, 10:39 AM)Rufus191 Wrote:
(07-09-2024, 11:45 PM)rmstevens2 Wrote: Just noticed a new male match at Ancestry with my surname. Once again I find myself wishing that Ancestry gave Y-DNA haplogroup assignments at least to the extent 23andMe does. Then I would know whether or not this new match could be related to me on my Y-chromosome line.

Instead, I have no idea, so my message to him was a shot in the dark. If I knew his Y-DNA haplogroup, I would have some info to use to persuade him to test with FTDNA, or I would know not to waste my time.

Absolutely would agree with this Big Grin! I have so many times tried to contact Ancestry matches, and asked if they any knew their Y-DNA haplogroup, even giving a link to the Morley DNA or YSEQ tools, but no success whatsoever!

Same here. I feel if I had the advantage of being able to tell them that we are probably related on the Y-chromosome line, they might be persuaded to test with FTDNA. Right now, it's all just a shot in the dark.
Let us now praise famous men, and our fathers that begat us.

- Wisdom of Sirach 44:1
Reply
#17
Ancestry should stop using Timber.  Either altogether, or at the very least they should use it in an informative way.  Currently, they use it prescriptively instead -- to replace the sharing amount actually found through conventional matching techniques with one they hypothesize is more reflective of the relationship of the matches.

Used informatively, Timber may warn customers of the possibility that some of the sharing in a given instance might be due to "excess sharing".  Used prescriptively, customers may often be unable to tell that the relationship either is -- or may be -- closer than Ancestry implies.  For example, if you're using ProTools sharing and you see that two people share, say, 47 cM, you have no way of telling that it might be 87 cM that they actually share.

The rules under which Timber operates make safe only matches with unweighted sharing of at least 90.0 cM -- but the instance unweighted sharing drops even a fraction of 1 cM below this, Timber is free to chop to it's heart's content.  For example, my daughter shares 47 cM with GK, according to Timber's adjusted sharing.  The relationship between my daughter and GK is predicted on this basis to be 3rd cousin once removed or half 3rd cousin.  In actually, GK is my daughter's 2nd cousin once removed since he's my 2nd cousin.  GK and I share 264 cM in 15 segments, with a longest shared segment of 73 cM.

So how did 40 cM of the matching DNA passed directly from me to my daughter suddenly become "excess"?  The answer is, it didn't.  It's just that this 40 cM seems to be located in a region or regions that may be excess for some individuals who share the DNA.  

And if you think this is something that only occasionally is an issue, out of my 87 "Timber-safe" matches, 48 of them are not "Timber-safe" for my daughter (meaning they're less than 90.0 cM).  Of these, 100% have been reduced by Timber, even though if none of my sharing is excess, none of hers should be either.  The sharing amounts have been reduced by as much as 48 cM!  BH, for example, is the granddaughter of my 1st cousin DS.  DS and I share 1,030 cM; our longest shared segment by itself is 111 cM.  My daughter and DS share 489 cM.

Further, my sharing with BH is a respectable 216 cM -- which makes sense, because BH is my 1st cousin twice removed.  Yet Ancestry reports my daughter's sharing with BH as only 46 cM.  Their predicted relationship is 3rd cousin once removed, or half 3rd cousin.  In actuality, they're 2nd cousins once removed.  Average sharing for this relationship is 123 cM, with a range of 0-316 cM.  So 46 cM is still within that range, but according to Ancestry this relationship only has a frequency of about 7% at this level of sharing.  What you wouldn't know even from ProTools sharing is that the unweighted amount is 81 cM.  I only know that because my daughter's results are in my account.

Interestingly, I know from ProTools sharing that two of my daughter's 1st cousins -- with exactly the same relationship to BH as my daughter has -- share 107 cM and 120 cM with BH.  But these numbers are "Timber-safe", so we're seeing the unweighted amount.  My daughter's third 1st cousin only shows sharing with BH of 30 cM.  For all Ancestry reveals -- thanks to Timber -- the unweighted amount could be 89 cM!

Between the alternatives of having somewhat distant matches look closer than they are, versus having close matches looking more distant than they are, I'd take the former.  Especially if Ancestry simply uses Timber to provide a caution that the match might be more distant than it seems.  Instead, Ancestry chooses to conceal the unweighted sharing of others -- no matter how much it might be, unless it is 90.0 cM or above.
My ancestry is Palatine German - Swiss - Alsatian / British & Irish / Menorcan / French / Indigenous American
Reply
#18
For those people who have ProTools, sorting by your match’s cM has been enabled.
Reply
#19
(11-01-2024, 12:10 AM)Mabrams Wrote: For those people who have ProTools, sorting  by your match’s cM has been enabled.

Thanks for the tip.  I've been using it for a little while and it's been extremely illuminating.  It turns out so far that in a great many instances in which Timber decided to chop off significant amount of my sharing with matches, those matches are fairly closely related to other matches that happen to be above the 90.0 cM "Timber safe" zone for me.

For example, there's SA.  My sharing with SA is reported as 30 cM, but this is after Timber chopped off nearly half of the unweighted sharing amount of 57 cM.  Amusingly enough, the longest segment is 38 cM -- so longer than Timber's adjusted total.  In this case, I also share with SA's father, who shares 154 cM with me; with her aunt, who shares 193 cM with me; with a half aunt, who shares 49 cM with me after Timber, but 84 cM unweighted; and with a half uncle, who also shares 49 cM with me after Timber, but 85 cM unweighted.

What Timber seems not to consider when looking at whether "too many" 4th or more distant cousins sharing given regions of DNA, is how the matches are all related to each other -- not just how they're related to one particular match.  What I mean is, if I share the same segment or segments with a lot of 4th cousins, it makes a difference if they all happen to be, say, siblings to each other.  If they share a significant amount of DNA with each other, that increases the probability of my sharing DNA with the lot of them.  It's definitely not the same as if we were all 4th cousins to each other.

In this case, though, we're actually more closely related than that.  I'm a 3rd cousin to SA's father and his three siblings.  Also, the fact that two of these siblings are half siblings doesn't affect my relationship with any of them.  We happen to be related on their shared mother's side, and I'm not related to any of the fathers involved. 

What it means is that Timber's "adjustments" with SA and the half aunt and half uncle are almost certainly unwarranted.  All four of those in SA's father's generation are 3rd cousins to me, and SA is my 3rd cousin once removed.  Average sharing with 3rd cousins is 73 cM, but the range is from 0-234 cM.  The Timber-adjusted amounts above would be more expected of 3rd cousins once removed, which would not be the correct relationship except for SA herself.
My ancestry is Palatine German - Swiss - Alsatian / British & Irish / Menorcan / French / Indigenous American
Reply
#20
Another example of Timber's excessively zealous chopping:

I have a 3rd cousin, RW.  At 184 cM in 20 segments, our sharing is actually on the high side for 3rd cousins.  It does, however, fall within the range (0-234 cM).

With CH I only share 36 cM in 3 segments -- after Timber -- with a longest shared segment of 35 cM.  This represents a reduction by Timber of 29 cM from the unweighted sharing of 65 cM.  It turns out, though, that the shared match with CH who shares the most with her is RW.  The two share 1,855 cM.  This is because CH's mother is a full sister to RW.  Obviously, CH didn't get the DNA she shares with me from RW, but from someone whose ancestors are exactly the same as RW's -- including our shared 2nd great grandparents.

What this means is that there is actually no reason to doubt that the full unweighted sharing amount of 65 cM is likely correct.  But there's more.  Timber also chopped off 31 cM from my unweighted sharing with JW of 65 cM.  JW is CH's daughter, so if there is no good reason to doubt CH's unweighted sharing of 65 cM, there's really no good reason to doubt JW's unweighted sharing of that amount.  It's just a case in which the daughter seems to have inherited all the DNA that her mother shares with me.  Finally, there's DH.  DH is likewise a daughter of CH, but in her case Ancestry only shows sharing between us of 16 cM.  This represents a 9 cM reduction, from 25 cM.  In this case, too, there is no reason to regard this 9 cM as somehow reflecting "excess sharing".

I'm seeing countless other examples like this.  Situations in which even if the matches are as distant as 4th cousins to me (and in many cases they're closer), they're much, much more closely related to many of the shared matches.  Yet Timber seems to be unable to consider that even if one match is a 4th cousin to a higher-than-expected number of people, it really doesn't count of they are closely related to each other.

(For example, my full siblings and I share quite a few of the same 4th cousins.  Not because of how much any of us shares with any of those matches, but because of how much we share with each other.)
My ancestry is Palatine German - Swiss - Alsatian / British & Irish / Menorcan / French / Indigenous American
Reply
#21
I can't see matches public trees anymore!?
Wishlist: Get that back.
Reply
#22
(01-12-2025, 05:57 PM)Kale Wrote: I can't see matches public trees anymore!?
Wishlist: Get that back.

I just looked at the public linked trees of a few of of my cousins, as well as a couple of unlinked trees.  They're all still there.  However, I'll note that I'm also at max in membership terms, so I have no way to know if there might be any limitations for folks who may have taken an Ancestry test but are not members.  (Of course, I have no idea whether this could apply to you.)

FYI, it actually bugs me when Ancestry does anything to drive folks away, since I figure that my subscription is enhanced by having more matches who stick around rather than quitting and deleting their info.
My ancestry is Palatine German - Swiss - Alsatian / British & Irish / Menorcan / French / Indigenous American
Reply
#23
For 'non-members'(?) (I only paid for the test, not any subscription) I used to be able to, if a match had a public tree, click on it and see a certain number of generations in their tree. Also come to think of it, Ancestry would suggest which ancestor(s) might be responsible for the genetic match, which was also really cool.

Both of those features are now paywalled. That's just mean, if anything I'm now LESS inclined to give them any more money.
Reply
#24
(01-13-2025, 04:13 AM)Kale Wrote: For 'non-members'(?) (I only paid for the test, not any subscription) I used to be able to, if a match had a public tree, click on it and see a certain number of generations in their tree. Also come to think of it, Ancestry would suggest which ancestor(s) might be responsible for the genetic match, which was also really cool.

Both of those features are now paywalled. That's just mean, if anything I'm now LESS inclined to give them any more money.

I get that.  I've been a subscriber at the highest level for over a decade, but while one might argue that to some extent I'm subsidizing non-subscribers, I don't look at it that way.  I figure that having more people testing is a benefit to me because I end up with more matches -- even if they're only paying for the test itself.  It certainly isn't in my interest to have Ancestry drive those folks away by putting so many features behind a paywall.  Just taking the test ought to at least get a limited-time full subscription.  (And if it's to be for a limited time only, that should be clearly spelled out from the beginning.)
PaulH01 and Kale like this post
My ancestry is Palatine German - Swiss - Alsatian / British & Irish / Menorcan / French / Indigenous American
Reply
#25
That's an important point for them to consider to, it's not just about driving in new subscriptions, but maintaining recurring ones. In this particular case it's backfired on both counts.
I'm not inclined to subscribe now, and to boot if a regular/non-enthusiast person asked me which test to get, I would have been able to gives pros and cons to each Ancestry and 23andme.
Now I see no advantages for Ancestry, and would tell them 23andme.
Reply

Check for new replies

Forum Jump:


Users browsing this thread: 1 Guest(s)