23andMe made a significant change about the time I was recording my RootsTech presentation about triangulation which provided examples at each vendor. Unfortunately, there was no notification to customers, so most people still aren’t aware.
In the fall and winter of 2020, 23andMe made several changes that resulted in losses to the genealogy community.
At first glance, it looks like this particular change is cosmetic – simply a column heading title change – but there are modifications behind the scenes that negate triangulation at 23andMe. At least in the way triangulation previously worked with the functionality genealogists have long understood to be triangulation at 23andMe.
This article explains the changes, what they mean, and how to work around the issues.
Update
Please note that as of March 12, 2021, some of the changes seem to have reverted, but it’s unclear if all changes have reverted to the original status. It’s virtually impossible to confirm because testers cannot search for “Relatives in Common” by surname. Therefore, proceed by confirming that people who are marked as “Yes” for “DNA Overlap” do in fact triangulate on each overlapping segment using the techniques I’ve described below.
Triangulation
If you need a refresher about what triangulation means, how it works, and why it’s important, I’ve compiled triangulation resources into one article, Triangulation Resources in One Place.
Let’s look at what happened at 23andMe.
Before the Changes
Before the changes, it was possible to quickly determine if you triangulated with two other people on at least one segment by looking at the “Shared DNA” column. Now, it isn’t.
This change has HUGE ramifications.
Unfortunately, it’s easy to simply not notice the change or interpret the column heading change from “Shared DNA” to “DNA Overlap,” as unimportant, but that’s not at all the case.
A “Yes” in this column NO LONGER MEANS triangulation.
This change makes the 23andMe slides of my RootsTech session, DNA Triangulation: What, Why, and How, obsolete.
I’m rewriting that section, step by step, in this article.
Previous Information
Click any slide to enlarge
On slide 24 of my presentation, available here, I talked about clicking on a match, then scrolling down to the “Find Relatives in Common” link. If you click on that link, you see a list of who you and that match both match in common.
In this case, Everett Harold (not his surname) and I both match with my V4 kit, DH and Stacy.
That page, back then, had a column titled ‘Shared DNA.”
At that time, a “Yes” in “Shared DNA” meant that the three people triangulate on at least one segment. That’s not what it means now, and the column header has changed too.
What I said in the presentation was this:
“Looking under the Shared DNA column, the people with a Yes triangulate, and the people with a No, do not.
This means that Everett Harold, me, and DH triangulate. It also means that Everett Harold, Stacy, and I do NOT triangulate.”
Please ignore this and the next slide, #25, too, because the 23andMe page has changed – along with the meaning.
Just put what I said and what you think you know about how triangulation works at 23andMe out of your mind. If you haven’t yet watched my Triangulation session at RootsTech, please just simply skip those two slides (24 and 25) so you don’t confuse yourself with old and now irrelevant information.
We’re starting over here with triangulation at 23andMe.
Current 23andMe Information
Here’s the same 23andMe “Relatives in Common” page, today:
Click to enlarge
You can see that while Stacy was marked “No,” on the previous “Shared DNA” page, the column is now titled “DNA Overlap” and she is now marked “Yes.”
The new infographic says this:
Here’s what this change means:
- Previously, if someone was marked as “Yes,” it meant that in fact all three people did share a common segment of DNA AND matched each other on at least one segment. That meant they triangulated on at least one segment.
- Currently, this field only means that they share an overlapping piece of DNA with the tester. It DOES NOT mean that they all 3 match each other on that segment.
- They may or may not triangulate.
You might be wondering how that’s different. It’s very different and quite important.
Overlap Versus Triangulation
Here’s an example of two people who both match me on chromosome 15 and are marked “Yes” in DNA Overlap. Based on this graphic alone, or that “yes,” you can’t determine if this overlapping segment means triangulation, where the orange and purple person also match each other, or not.
- BOTH of these people match ME on chromosome 15.
- If they also match each other on a reasonable portion of chromosome 15 where they both match me, then we all triangulate. A reasonable amount of matching DNA at 23andMe is 6 cM, their match threshold.
- If those two people do not also match each other on a reasonably sized segment (6 cM) of chromosome 15, then we do not triangulate. This would indicate that one match is from my mother’s side, and one from my father’s side, or that perhaps one is identical by chance. In other words, we do not share a common ancestor on this segment which is the purpose of identifying triangulated segments.
Based on other comparisons which I’ll show you how to perform in a minute – the purple and orange people don’t match each other on this segment. Therefore, this segment is not triangulated between me and the purple and orange people.
Previously, for this match, the “Shared DNA” column was marked “No,” and now the “DNA Overlap” column is marked “Yes.”
The three of us don’t triangulate, and “DNA Overlap” now only means that the three people share some DNA on the same portion of a chromosome with me, NOT that they match each other, which would mean that we triangulate.
It’s a hugely important distinction.
Before, “Yes” meant triangulation and now “Yes” just means an overlap, but NOT necessarily triangulation. You have to figure that out for yourself.
Overlap at 23andMe
An overlap simply means that two people match you on the same portion of DNA.
Someone from your Mom’s side and someone else from your Dad’s side will both match you on a segment of DNA in the same location on a chromosome, shown above. However, they won’t match each other because one is from your Mom’s side and one is from your Dad’s side. Your Mom’s DNA is different from your Dad’s.
To prove that you all three share a common ancestor, you all three need to match each other on the SAME reasonably sized overlapping chromosome segment.
However, things are even more confusing now at 23and Me.
An Additional Complication
23andMe now indicates that Everett and Stacy have a DNA overlap with me, but the chromosome browser shows NO overlap on any chromosome when I compare both Everett and Stacy to me on my chromosome browser.
How is no overlap even possible when Stacy is listed on the Shared Relatives list with me and Everett, AND 23andMe shows a yes for DNA Overlap?
I eventually found the answer, which makes match analysis much more cumbersome for genealogists. What used to be one step now takes several, not to mention the “yes” answer is now unreliable.
Essentially, all that “Yes” in the DNA Overlap field means is a hint for you to dig further.
Determining 23andMe Triangulation
It appears that the only way to tell if your two matches match each other on the same chromosome as you is to “Select different relatives or friends to compare” at the top of the chromosome browser page.
You’ll see your name plus the two people you were comparing against your DNA in the chromosome browser.
You’ve already seen how they match you on the chromosome browser. What you now need to view is how they match each other.
You can remove yourself, and replace your name with one of your two matches, as shown below.
This will show Everett’s chromosome with Stacy compared to him.
Everett and Stacy do match each other on two smallish segments, but not in the same locations as shown on their match with me.
This is Everett’s match with Stacy (purple).
I match Everett on chromosome 18, but not Stacy.
I match Stacy on chromosome 7, but not Everett.
There is no overlap shown.
Ok, I’m adding myself to Everett’s matches, just to double-check.
Next, we’re looking at Everett’s chromosomes in grey. Stacy is purple and I’m orange.
Overlap Issue
I’ve found the confusing overlap issue, but it only makes the situation worse.
Everett matches both me and Stacy on adjacent and very slightly overlapping portions of chromosome 18. However, the amount of DNA where I match Stacy on chromosome 18 is too small to be considered a match when compared to Stacy directly, meaning it’s less than 6 cM – the smallest 23andMe segment to show as a match. This tiny sliver of overlap only shows when comparing from Everett’s perspective where we can see his match to me and Stacy both on the same chromosome.
A secondary change is that now it appears that 23andMe is showing any small piece of overlapping DNA with a “Yes.” Any segment of DNA smaller than 6 cM, their match threshold, should not be listed as overlapping if we all three don’t match each other on at least 6 cM of DNA.
You can work around the changes 23andMe made, but it has made a one or two-step easy process into a more complicated, cumbersome multi-step procedure involving comparing multiple people to each other separately.
Summary
| Previous | Now | |
| Column Title | Shared DNA | DNA Overlap |
| Triangulation Status | Triangulation if “Yes” in the “Shared DNA” column | Not an indication of triangulation, even if “Yes” in the “DNA Overlap” column |
| Triangulation Indicator | “Yes” in the “Shared DNA” column | None, triangulation not flagged |
In summary, for triangulation now at 23andMe:
- The DNA Overlap status of “yes” DOES NOT indicate triangulation.
- The DNA Overlap status of “yes” indicates overlap on the same chromosome, not triangulation, meaning all three people do not necessarily match each other.
- DNA Overlap status of “yes” MAY mean the three people triangulate, but further comparisons are needed.
- DNA Overlap status of “yes” may refer to overlap smaller than 6 shared cM which is not reflected in individual one-to-one matches.
- The DNA Overlap status of “yes” may therefore not be technically accurate in terms of genealogical matching and triangulation.
- A DNA Overlap status of “no” means you do not overlap which means you cannot triangulate.
- To determine triangulation, meaning if you and two other people all match each other if you share an overlapping segment of DNA on the same chromosome, compare each pair of people one-to-one in the chromosome browser.
- If you do not find overlapping DNA when comparing three people one-to-one, try the same comparison to the other two people from the perspective of one of the other people in the group, as I did with Everett. This may reveal a small overlapping segment, as illustrated in this article on chromosome 18 when I showed me and Stacy on Everett’s chromosomes.
It’s worth noting here that every segment is different. Triangulation on any individual segment should not be extrapolated to mean triangulation on every common segment, even between the same three people, is valid for all overlapping segments. Evaluate each overlap separately.
This fundamental change makes triangulation at 23andMe much more difficult for the genealogist. Fortunately, there is a work-around.
Please feel free to share this article with anyone who may have tested at 23andMe and is using their tools for genealogical purposes.
_____________________________________________________________
Disclosure
I receive a small contribution when you click on some of the links to vendors in my articles. This does NOT increase the price you pay but helps me to keep the lights on and this informational blog free for everyone. Please click on the links in the articles or to the vendors below if you are purchasing products or DNA testing.
Thank you so much.
DNA Purchases and Free Transfers
- FamilyTreeDNA – Y, mitochondrial and autosomal DNA testing
- MyHeritage DNA –Autosomal DNA test
- MyHeritage FREE DNA file upload – Transfer your results from other vendors free
- AncestryDNA – Autosomal DNA test
- 23andMe Ancestry – Autosomal DNA only, no Health
- 23andMe Ancestry Plus Health
Genealogy Products and Services
- MyHeritage FREE Tree Builder – Genealogy software for your computer
- MyHeritage Subscription with Free Trial
- Legacy Family Tree Webinars – Genealogy and DNA classes, subscription-based, some free
- Legacy Family Tree Software – Genealogy software for your computer
- Charting Companion – Charts and Reports to use with your genealogy software or FamilySearch
- RootsMagic Software – Genealogy software for your computer
- Newspapers.com
Books
- Genealogical.com – Lots of wonderful genealogy research books
Genealogy Research
- Legacy Tree Genealogists – Professional genealogy research
Why would 23 do this? Less computational resources on their end? Whats their motive?
If you think about the resources and table storage it frees, I expect that’s the motivation. But I don’t know and that’s speculation. Remember, they have been moving away from genealogy for some time now.
I think that might be the reason indeed. Their CEO has clearly indicated in a recent interview that the focus is on Health, not Ancestry. They made significant losses in the last two financial years, around $40 million each. They also burned through the whole $750 million seeding money hence now the step with the SPAC. So far it’s a loss making business, well, if you would exclude the stock compensation then actually it’s not. So I leave it to you to make conclusions here what the real cause is for making a loss 😉
All this juicy info can be found in their investors presentation BTW.
So yeah, as with every new customer they have to calculate at least 10 million comparisons (for their 10 million customer) that’s why automatically calculating triangulation for the whole customer database is so compute and storage intensive (instead of doing it on the fly).
Got the numbers wrong (from memory). The losses were much bigger and even without stock compensation it would be a loss. But I still hate it when top management takes fat stock bonuses whilst in this case a larger number of employees go the kick in the butt in 2018.
Here’s the post I wrote as a reaction of their video conference and investors presentation (to be found on WikiTree https://www.wikitree.com/g2g/1177459/23andme-to-be-acquired-by-richard-bransons-vgac-and-go-public?show=1199754#c1199754):
This is a very telling interview with Sir Richard Branson and Anne Wojcicki:
Why Sir Richard Branson gave 23andMe millions of dollars to go public
If you listen to what Anne says at 5:57 onwards, she said:
“… was really about emphasising more and more of the health opportunity. And Ancestry and Health have always been key components of the product. But more and more, the future is about the health side.” and “… it’s a bigger opportunity on the health side.”
That’s obviously a very disappointing direction that 23andMe is taking for all genetic genealogists. And it comes despite AncestryDNA pulling out of that same DNA health sector.
It’s clear that they think that the market for the ancestry product is saturated or rather “belongs” to AncestryDNA and maybe newcomer MyHeritage.
I’m not going so far to say that 23andMe will be pulling out of that, with 10 million existing customer.
But their focus for the future is clear and investment in new genetic genealogy features or improvements of their consumer facing genetic genealogy features will be little.
BTW, there is a ton of interesting numbers in their investors deck: 23andMe and VGAC Investors Presentation
But there’s also a telling when numbers only start in 2020 and nothing is being shown especially for the “growth” period from 2018 to 2020 (which probably was pretty flat).
Oh and I love her not answering about profit. Because 23andMe has made a loss as per the investor paper. Adjusted EBITDA for FY19A: -$141 millions, FY20A: -$147 millions
Lastly, both years their top management was so successful that the stock based compensation was $37 and $44 million respectively (and remember that they spend $45 in 2020 on restructuring and exit costs). To me, that’s anything than successful but it’s the typical corporate greed of today, fire low level people whilst filling your own pockets with millions from of stock compensation.
Thanks for this information, Roberta. Do you have any knowledge why 23andMe didn’t announce this change in their blog or at the RootsTech conference? As you wrote, this isn’t a small change (and they are silently doing others as well, eg. for my v3 they removed all regions and now basically show me only as 100% Northwestern European when before they had my Italian ancestry down to the Veneto region – where my Asiago ancestors are from).
As for doing manual triangulation, I still find it better that they provide us with such a great tool when AncestryDNA and FTDNA (it won’t show us B vs C comparison before some shouts “wrong”) aren’t giving us any chance to perform triangulation at all on their website.
But for those that find all this triangulation too complicated or time consuming, I suggest you try out my “Your DNA family” app which does the triangulation automatically for you.
The app does hundreds of comparisons in the same time, so large potential groups of matches are no longer a problem to identify who triangulates and who’s not.
For more info: https://help.yourdna.family/hc/en-us/articles/360002562395-Who-are-we-and-why-are-we-doing-this-
I don’t know why. They have been moving away from their genealogy customers for a long time.
I also just discovered that they no longer let you upload Raw DNA from other companies. They say it is for our privacy, but really it seems they either want you to buy their kits and/or don’t want to be bothered with the genealogy side. They never were very interested in it to begin with!
They have never allowed uploads, except for one time for a very short time and then those people didn’t receive the genealogy tools.
I’m afraid I have to ask a very basic stupid question. I’m not seeing how to get to the browser page you refer to. Can you explain that starting from what shows up on the Relatives in Common page?
1. Click on the Ancestry tab. 2. Click on DNA Relatives list. 3. Click on one of your matches. 4. Scroll down the page and click on the blue “fine relatives in common.” You will only have this if the other person has agreed to share. 5. You will see the people you share in common with the person you just clicked on. 6. Click on the “yes” or “no” in the “DNA Overlap” column, which shows you the browser. 7. To view other people compared to each other, at the top of the browser page, slick on “select different relatives or friends to compare.”
WHY?? JUST, WHY???
Thanks for this explanation.
It is an annoying turn of events especially as there has been no communication from the company about the change as far as I have seen. Doubly annoying for you having just completed your RootsTech talk before the change happened!!
I have shared your blog post at http://www.ancestorian.com
I feel really bad about the info in the talk because I know it will inadvertently mislead people.
Don’t beat yourself up Roberta…23andMe were quite aware of the timing and it just illustrates a total lack of consideration for existing consumers. If they’re going to make backward changes like this, out of the blue, then they’ll also lose customers like me who paid the extra fee for more info. To be honest they were my first DNA testing site, but are my least used now, and I never recommend them for genealogical purposes.
I’m curious as to what company you now prefer?
Ancestry is even worse, since they have no kind of chromosome browser at all and also use the presumptuous (and often incorrect!) algorithm Timber.
FTDNA might be a somewhat better choice, since they at least do have a chromosome browser.
MyHeritage also has a chromosome browser, but I consider their failure to utilize the X chromosome to be a major drawback.
I use both FTDNA and MH daily.
Ditto for me Gary, although Ancestry is invaluable for family trees. Plus Gedmatch, DNA Painter etc.
I’m actually replying to both Roberta and Verity.
Please don’t misunderstand me. I actually do use both Ancestry and MyHeritage, and there are a number of things to like about each company.
However, Ancestry’s continued refusal to provide any sort of chromosome browser makes triangulation even harder there than at 23andMe. At 23andMe, there are still ways to do triangulation in some cases.
For example, I have a tested father at 23andMe, plus five tested full siblings, a tested daughter, and four tested nieces/nephews. Nevertheless, I actually have the maximum subscription to ancestry.
I also have a full subscription to MyHeritage. Currently, I believe their “ethnicity estimate” to be the one closest to my paper trail. It’s just that I regard leaving out the X chromosome to be a significant matter.
All but 20 cM of my X chromosome came from my mother’s father — 20 cM surrounding the centromere. This means it’s very easy for me to see not only that someone sharing an X-chromosome segment with me is on my mother’s side, I can tell which grandparent I’m related to them through.
My maternal grandparents, like may parents, do not appear to have any shared ancestry, and I have not yet found a situation in which I share with one grandparent on the X chromosome but the other grandparent on autosomes — with the exception of my siblings and their descendants.
So I would recommend both Ancestry and MyHeritage, and I use both. But they have their own issues. As to FTDNA, I think it’s unsurpassed when it comes to mtDNA and Y DNA haplogroups.
Thank you for this workaround. In fact, this may force me look more closely at the matches’ relationships.
I’ve sent multiple messages to 23andMe since the first day I noticed I was seeing people on my father’s side now matching on my mother’s side on the same chromosome when that had never happened before. They have their own unique segments and none of Dad’s matches were on the same chromosome Mom’s segment was located, ever.
Yes, there are several at 23andMe and other sites that match both parents, despite the fact they are not related, but they had family living very close since the mid 1800s, different counties but not that far apart. Some are still in those counties so a cousin from one parent marries a cousin of the other parent and their children or grandchildren tested DNA and they match both of our parents sometimes.
I told 23andMe there was nothing worthwhile left for me there except to look at matches and if I had to do the triangulation manually, well let’s just say I’ve spent less than a hour a day there since then. I’m only looking at the closer matches these days. I was outraged they took away the original triangulation and left us the manual job of doing triangulation. It is a horrible thing to do to their customers. I must admit what I said was not stated as nicely as I probably should have done but I was extremely upset with what they did. As far as I’m concerned they can go out of business for all I care after that stunt, they deserve to fail.
So they have nothing to recommend them anymore. I won’t even renew that 23andMe+ I signed up for all 3 kits I have there. I’ve had enough of their failures to fix things they broke years ago and the worst customer service ever. Then they pull this stunt. I have to stop writing as my blood pressure is climbing.
Very grateful that you provided this new and detailed explanation…disappointed that 23 And Me has made this significant change. Sigh.
I noticed this over the weekend and thought I was losing my mind!!!
Luckily for me I’ve been habitually saving the all relatives file about monthly if not quarterly at least for all my people I manage for the past couple years due to these insidious changes!!
It is interesting that just the other I tried working with the DNA Overlap, opening a second browser and using the select different relatives or friends to compare and loading in all those DNA overlap people into it, to see how they were on the browser. Funnily enough, there were a lot that were around the same segment on a single chromosome around the same position on the browser and a few that were on were on another chromosome around the same segment so I realized that they probably there were two different possible triangulations. I only had my DNA done a few months ago on 23andMe and have not found it at all intuitive or their tools very abundant or easy to use. I was very disappointed. I agree they seem to be pushing the health aspects which in my opinion are a total waste of time and laughable. Anyhow, I am still having a really hard time figuring out this triangulation thing, no matter how much i read up on it and how it can help me find out how these triangulated matches fit into my family. In the example I used above I found about 12 of these possible triangualated matches but don’t know what to do with them next. I feel so dumb. lol.
Look for a common ancestor. I paint mine. Look at the DNAPainter articles.
Thanks Roberta for your suggestion. I thought that was probably the route to go and I do have a couple of DNA Painter maps of that I can toy with but I struggle with figuring out anything with that as well. I guess it doesn’t help when I don’t have a lot of family who have tested or are willing to test and the matches I have at various sites that I can identify are just a handful.. I do love your posts though. I have had them delivered to my unbox for years now. Just wish I was better at this. I seem to be losing a lot of my processing capabilities as time goes on. Thanks for being there for all of us and our willingness to help.
Thanks for posting about this and showing the work-around, Roberta. I appreciate all the effort that you made to break this down and explain how to make the best of it. I am glad we at least still have the tools available to do the work-around. I like that there are a lot of testers on 23&Me that have not tested anywhere else. I did opt for the chip upgrade and the new 23&Me+ subscription. Being able to access the segment data for 5,000 matches has been very beneficial. I’ve also taken the opportunity to hit the Connect button for hundreds of my matches and it’s unbelievable to me how many have responded with an actual reply, not just an acceptance. I appreciate this Connect feature. So while I’m not happy to see this change, I still find 23&Me very valuable for genealogy.
So that explains the confusion I experienced the other day. I was wondering what I had done wrong, but now I know it’s 23andMe showing their lack of concern for those of us using their tools.
I was looking at a new DNA match and saw the “Yes” and thought it was triangulation. When I looked at the matches, it didn’t show what I was expecting. Fortunately both of my parents tested there, so I can double check for things like that easily. Now I’ll have to rely more on that. The “Yes” seems rather useless now. I was wondering why there were so many “Yes” comments in that column.
I used to be a great fan of 23andMe.
Roberta,
Thanks for the explanation.
Roberta, I found a reference to your writing up something about changes at 23andMe and glanced over what you have above. Have you been able to have a discussion with them about this? I opened a problem report on this issue last Tuesday, a few days after I started noticing that I was getting shared matches with “Yes” under the “DNA Overlap” column, and even though the writeup that the little icon next to “DNA Overlap” presents when you click on it gives you a write-up indicating that it means the same thing, it doesn’t.
I reported the problem as giving some 10-20% of the “Yes” matches not bearing any DNA Overlap at all. I got a first response back today and they sent me to the tutorials that still show the old display with “Shared DNA.”
Since they have all these tutorials with old info, it seems to me that no one’s in charge right now, and someone has made active a part of a change that hasn’t been carried out very well.
They have been moving away from genealogy for a long time. They may actually have no idea what they’ve done, or they may simply not care.
I hope it’s the former rather than the latter. Since they’ve escalated my problem report and responded that they were aware of and working the issue that tends to make me believe they do care.
Since they’ve not changed any of their tutorials, my gut feel (based on 40+ years at IBM in product support, testing, and development) undertook some “simple” cosmetic changes in the user interface without a clue that doing so could affect the results beyond cosmetic changes. And then, not being genealogists using DNA for that purpose, they brought online those “cosmetic” changes. without the needed testing. So, the first significant testing the changes received was when us genealogists began putting them in use, as in you preparing for RootsTech, when we realized, “Houston, we have a problem,” and I started looking for shared segments and realized that the “matches in common” I was looking at didn’t even share segments on the same chromosome, much less overlapping on the same chromosome.
I also meant to say, I’m holding off trying to do any kind of matching / triangulation until I get some kind of resolution on my problem report.
Maybe if several of the more well known genealogist would contact them and explain the problem they might correct it it. They may have less interest in this part of their business but if it becomes less helpful to genealogist they will get less people to test and thus less data for health research. Maybe their staff didn’t realize the impact when the change was made.
They dissolved their genealogy advisory group maybe 3 years ago. We have no avenue, other than articles and individual support interaction.
Hi Roberta – thank you so much for this article. I had also been totally confused with the “overlap” column but no triangulation. That column is redundant now since ICW already means you share segment(s) with each match and the inclusion of less than 6cM segments is meaningless and misleading.
Would you recommend writing “error reports” as mentioned above or sending directly to someone else? If we all did that (more or less politely) and suggested that 1) they bring it back 2) and that we no longer could recommend 23&Me for genealogy without this feature, would they even listen? I too find the “health reports” laughable and won’t recommend them to anyone.
I would suggest submitting a report, if given their health focus, I’m not hopeful. But if we don’t try, we’ll never know.
Roberta, thank you for this info. It’s disheartening but good to know.
It is definitely disheartening when any company takes a step backwards, as 23andMe is doing. I have to say that Ancestry, while refusing to budge on a chromosome browser, is at least trying to improve functionality in other ways.
However, to my knowledge 23andMe is still just about the only company that shows when matching is just to one side and when it is to both. That’s important for me because of my five full brothers and sisters, all of whose results are part of a single account with me.
There are very few times when any relative matches only me or only one of my siblings. And since our father’s results are also in the account, it’s a snap to determine which side a match is on.
In addition, I can look at where my siblings share with my daughter, or where I share with my siblings’ sons and daughters, and compare that to where my father shares with each of his grandchildren. By inference, segments my father does not share with a grandchild must be from my mother.
If I also do not share the segment with a given niece or nephew at that location, the segment is likely to be from my mother but from the strand she did not pass on to me. (In that case, I also should have no more than half sharing with whichever of my siblings is the niece or nephew’s parent).
This is why I encourage people to get even collateral relatives to test whenever possible. In addition to my siblings, I still have nine living aunts and uncles on my father’s side that I would love to see get tested, and I still think it would be worthwhile to use 23andMe — though less so than before. Eight of these aunts and uncles are full siblings to my father, while one is a maternal half sibling.
On my mother’s side, both of her brothers passed away before the advent of DTC DNA testing, and I have been unable to convince either of my two 1st cousins on this side to test (even when I offered to pay!).
Like everybody else here, I *am* disappointed in 23andMe’s decision. But I personally have too much invested in the parts that still work (for now) to completely abandon the company … yet.
The same thing is true for Ancestry, even though I believe that Timber is one of the worst decisions they’ve ever made — perhaps even worse than not offering a chromosome browser. I’ve seen too many instances of folks that I share over 80 cM with *unweighted”, but as little as 30 cM after Timber. Including people as close as 2nd cousins once removed, or half 2nd cousins.
Well the good news is that it looks like 23andme fixed it and the tool is acting the way it used to.
Hey, maybe the article did some good. I know several people reported it. I’ll have to go and check.
I have learned to make a post on the 23andme forum and to tag Customer Care and the Forums Moderator. They do seem to pay some kind of attention there when issues arise.
Yes, that’s what I did several days ago, and customer care has now responded.
Great news! And, yes, I do think you should take the credit, Roberta. 🙂
Roberta, re: my post earlier about the “changes” in Relatives in Common at 23andMe. I
received the following update from 23andMe today:
Hello Doug,
Thank you for your reply. Your request has been escalated to me for review. I understand that you have noticed changes as to what “Yes” means for DNA overlap. We are aware of this change, and are looking to resolve this now, so that “Yes” indicates the presence of the same shared DNA segment between all three relatives. The reversion to our previous segment matching logic should be reflected within your account soon – I appreciate your patience in the meantime.
Best,
L
So, from this note, it looks like the results we have been seeing recently are not the intended results, but it also looks like they aren’t sure they have a resolution yet.
Regards, Doug.
p.s. I’ll follow-up with any update that I feel warrants it.
Thank you.
Roberta, triangulation is now working the way it used to, at least for the test case I used. Could you check your results and post an update at the top of your blog if they look satisfactory.
Followup
After checking all the 23nMe shared matches with Match A that are “yes”s, I find over and over that “tri” sets X matches are all overlapping – UNTIL. I finally found one “yes” tri where the X match in the “yes” group did not overlap.
And I got lucky. Match B in the Yes group passes the “is it really triangulated test”. Even luckier, I know the MRCA for Match B. The MRCA lines for Match B is “me – my mother – my grandmother – my great grandfather” matched to “my ggfather’s brother – his brother’s daughter – Match B (female)”. Match B is my 2nd Cousin once removed.
My great grandfather is the son of Mary Gilder. I think this means that my original X match only ( Match A) is either a descendant of Mary James Gilder or one of her ancestors following the “X” path. But how many generations back ?? I think I read something about that. Probably on your blog. I will have to go find it.
Nite
Well, my Followup – was only a partial reply. No matter. I realized this morning I had made an error. For me this change in 23nMe introduces too much complexity. I think I can sort my DNA problem out but it will cost me hours.
It seems such a little thing for 23nMe to just check for 3 way matches. They have the information. They have done it before.
I think they have at least partially reinstated some functions. I have not checked in some time now. Some functions are only available with a subscription and I would have to retest plus subscribe and I’m not going to do that.
Linda Price — I posted a response upthread. Triangulation has been working correctly since at least March.
Also we should be thankful that we not only can see B vs C matches (can’t at FTDNA) or matches at all (can’t at Ancestry).
It’s very simple to do the triangulation yourself as 23andMe provides us with the tools free of charge.
Which other DNA testing company does give you a tool where you can compare 1 vs up to 5 other DNA matches? None!
If this is working now could you put an update at the top of the blog post. I had to read through all the comments to work out that this was no longer a problem.
Thank you so much for teaching me how to use Triangulation on 23 and Me!!! I love the one on My Heritage! I wish Ancestry would include it, too.
You can transfer to FamilyTreeDNA too where they have triangulation through family matching.
How can you do DNA triangulation at FTDNA when they don’t show you where your matches B and C match each other?
You can see if your matches match each other using the matrix. Plus, if someone is bucketed on a parent’s side, they have triangulated with someone. I wrote an article, here: https://dna-explained.com/2019/11/06/triangulation-in-action-at-family-tree-dna/
23andme has changed things again. It now just says “compare” and they have removed the “yes” and “no”s from that column.
Which means you have to click on every single one to determine if they triangulate. Let’s hope it’s just a glitch. I hope they haven’t moved it to the subscription only.
I suspect it’s a glitch — I see the same thing in my subscription account. I sent a message to customer service about it.
Thanks Ann.
Unfortunately I don’t think it’s a glitch because they have updated and removed the ‘yes’ and ‘no’ information when you mouseover DNA Overlap.
This same thing happened a couple of years ago, accidentally, due to some other code change. Took them a at least 4-5 days to resolve it. That was back in March of 2021
Ann,
Did you get a response from 23andMe support on this? Back in March ’21 when this happened I had to contact them multiple times and reference their own documentation on how they supported triangulation and once I did that, they actually opened a case number, which they sent me, and then they restored it within a couple of days.
I’m not Ann, but they said they would pass the feedback on, which strongly suggests it’s not a bug. Plus, a bug wouldn’t insert the word “compare.” Maybe Ann has heard something more.
Update: This change will be the status quo for some indeterminant time. I just received this response when I explained how much I valued and depended on the prior Yes/No/Share choices. “We’re exploring ways to make this type of data available to the people who want it, while not impacting overall performance of our systems.” I will put my thinking cap on and see if I can come up with practical suggestions for them or work-arounds for us.
Is there any chance the real reason for this change had something to do with the data leak that is now being reported on arstechnica?
I have no insider knowledge, but I think the most plausible explanation is the system resources required to serve what is probably a small fraction of the customers.
I agree Ann. In essence because you can still see the data so it doesn’t hide anything. Just makes it more difficult.
Plus, it would have been very unethical if 23andMe positively knew about this situation for that long and didn’t inform customers. I don’t think they would have done that.
They have now completely disabled Relatives in Common, so I personally think there was more to the triangulation change than what they led on. Perhaps the burden on the system was in fact this hacking and they panicked to remove functionality not fully realizing it was an exploit/hack. The timing is just too suspect for me to think it is unrelated.
I agree with your view, Ann. 23andMe has turned off/changed all features that were compute intense (eg in the past it auto calculated the DNA matches all the time, now it’s only created on-demand). This is surely to get the infrastructure cost down, a wise measure for any company.
Given that most users wouldn’t use that feature, it was a wasted resource anyways. But still it’s a pity it’s gone, as it provided me personally with a quick glance if a new DNA match was indeed worth to write to.