More Ethnicity Updates from AncestryDNA

AncestryDNA is maintaining their annual ethnicity updates, and it's a little early this year. But it's a new kind of update - rather than the usual changes to either the reference panel, or algorithms, or both, this one introduces a new feature called SideView. It is essentially phasing our DNA with our DNA matches to determine which ethnicities come from one parent or the other. It also means adjustments to our individual percentages, which should theoretically be an improvement. Phasing is usually done with parents or other very close family members, so I was skeptical about AncestryDNA doing it with our more distant matches. Your parents don't have to have tested for this new feature to work, but I was hopeful that my parents having tested would make it more accurate.

I find the parental breakdown (shown above) is very reliable - at least, it's as reliable as it can be given how accurate (or not) each of my kits are to begin with. For example, it correctly identified that my Norwegian and Italian ancestry are from opposite sides of my tree, and that is true: Norwegian is on my mom's side, Italian is on my dad's side. But it puts all of my Germanic ancestry on my dad's side because my mom's results still don't include Germanic despite having a great grandfather of full German descent (dozens of DNA matches on this branch confirm there's no NPE) and several other German branches further back. 

Looking at my mom's parental breakdown, shown above, (neither of her parents having tested), there is less reliability, that's partly due to the fact that her Norwegian ancestry is grossly exaggerated. She now gets a whopping 47% in Norway despite only having had one Norwegian (or Scandinavian) grandparent (so she should be about 25%, although it may vary, it shouldn't be more than about 36%). The majority of her Norwegian results does get put on one side, but that means there's not much room left for the other 25% on her mom's side that should be mostly English. Most of her English results get put on her other side, which isn't exactly wrong, she does have some English ancestry on that side too. But her dad's side should be mostly Germanic, and again, she gets no results in Germanic. If the percentages were more reliable to begin with, the split up would be more reliable too.

My dad's parental breakdown is very accurate, probably partly because his father tested but also because there is more genetic distinction between his mom and dad's sides - his mom was Italian, his dad mostly German and some Scottish and English. The split up (shown above) correctly shows all his Italian (Southern and Northern even though his ancestry is only Southern) plus trace amounts in Cyprus and Levant (obviously coming from his Italian ancestry) on one side, equaling exactly 50%. On the other side it correctly places all the rest of his ethnicities, although they are not all accurate - he wrongly gets results in Scandinavia where he has no known ancestry.

My paternal grandfather's parental breakdown is surprisingly very consistent with his tree, considering neither of his parents tested. On his paternal side, he is German with some English. On his maternal side, he's German and Scottish, with some English. Although his percentages are overall off (too much English, not enough German), the split up is accurately reflected here. English on both side, German on both sides (though barely), and Scottish on only one side.

My husband's parental breakdown (shown above) is also as accurate as possible given his percentage results and the fact that neither parent tested. It correctly identifies the majority of his Irish ancestry on one side and all of his English ancestry on the other side. His father was Irish, his mother was mostly English. He overall gets 40% in Ireland (a decrease from previous 47% which was much more accurate), and 36% is assigned to one side, his dad's side (shown below). His mother does have one Irish branch from much further back, which would amount to about 3%, and interestingly it puts 4% Ireland on his mom's side. Not bad. It then splits his Scottish results up more evenly on both sides - he does indeed have one Scottish 2nd great grandparent on his mother's side, so the Scottish portion being assigned to his father's side is obviously just due to the genetic overlap between Ireland and Scotland. His Scottish percentage is exaggerated to begin with: 22% when it should be more like 6% and probably no more than 12%, but interestingly the amount that is put on his mom's side is 9%, which is consistent with the Scottish 2nd great grandfather on his mom's side. Again, not bad, AncestryDNA, not bad. However, he has no Welsh or Norwegian ancestry, so those are obviously coming from genetic overlap with England.

So overall, the split ups among most of my kits were very reliable, but I can't say the percentages have benefited from the phasing. For example, my Scottish results wrongly shot up from 12% to 29% - based on my tree, the former is more accurate. And as mentioned, my mom is still lacking any Germanic results at all when she should be at least 12%, while her Norwegian results were already too high to begin with (43%) and just went up even more (47%). My dad's results didn't change by much, but he's now getting small percentages in incorrect regions that he didn't get before. In fact, most of my kits have seen this too - most of them now have small percentages in Ireland which they didn't have before. To my knowledge, all of my so-called "Irish" ancestors were actually Scots-Irish. So previous results were more accurate and the sudden appearance of Irish in results is disappointing (only because it's not accurate, not because there's anything wrong with being Irish, lol - obviously, my husband is half Irish).

Add Specific Relationship, AncestryDNA's Latest Feature

It sounds like it hasn't been rolled out to everyone yet, but it should be coming soon - AncestryDNA is (finally) adding the ability to change the estimated relationship range with a DNA match to a specific, known relationship instead. They're a few years behind 23andMe and FTDNA (although 23andMe still don't have shareable family trees so 23andMe is no better overall), but better late than never.

In the process of adding the specific relationship, it asks you which side of your tree the match is from, your mother's side, father's side, or both. And for matches you're unsure of the specific relationship, but you know which side of your tree they come from, there's an option to select which side and then instead of choosing a specific relationship, you can click "I'm not sure". It will then display "Mother's Side" or "Father's Side" (or both) without a exact relationship (the original estimated range will remain). 

Unfortunately, it does have some limitations. The main one is that it only goes out to 5th cousins, and any more distant relationships only have an option for lumping them all into a general "Distant Relationship" label. Not only does this rather defeat the purpose of being able to add a specific relationship if it's not actually a specific relationship, but it's also inconsistent with ThruLines, which at least goes out to 6th cousins (though that too is arguably a little limited). So essentially, ThruLines is going to show us our exact relationship with many 5th cousins once removed and 6th cousins, yet the new feature offers no way to add those specific relationships. The least they could do is expand it to the 6th cousins so it's consistent with ThruLines.

The other limitation is that it doesn't let you select more than one relationship, which is a complete oversight when it comes to lots of people who have endogamous branches of their tree, and identifiable endogamy (more than one set of most recent common ancestors) with many matches. Even when you select "Both Sides", it doesn't give you the option for more than one relationship. If it's a close match, assumes you've selected both sides because the person is someone like a niece or nephew, or full sibling, etc. Someone who shares your whole ancestry. If they aren't a close match, it seems to assume that although you may have two different relationships, they must be more distant than 5th cousins and only gives you the option to select "Distant Relationship". I suppose they're trying not to over complicate it for newcomers, but for people who use this for heavy research and breaking down brick walls in their tree, noting multiple relationships is vital.

It should also be noted that if one or more of your parents have tested, the system will automatically assign a match to your mother's side or father side depending on who they match. If for some reason, the system got it wrong, or only selected one when they actually match both, you can edit this by simply clicked the back button in the upper left corner of the side window (highlighted in yellow in the screenshot below).

That pretty much sums it up. In general, it's great they finally added this option, I know lots of people have been asking for it for a while. And I have gone through and selected known relationships for all the matches I've identified. But you may notice I have, for a very long time now, always noted the relationship and shared ancestor(s) in the notes field (along with emojis I used before groups were available). Unfortunately, due to the limitations of the new feature, I will have to continue noting the relationship myself in the notes field instead of relying solely on Ancestry's tool.

Small but Significant Changes at Ancestry.com

Ancestry is rolling out some new tweaks to their website that has everyone in a tizzy and I don't really know why, because in many ways they seem like an improvement to me. 

One of the changes was the removal of the clickable alphabet at the top of our List of All People. It allowed us to jump to surnames that start with any letter by clicking on the letter. I know the removal of this seems like a negative, but it's really not. What remains are two name search fields, one for first name, the other for last name. They were always there, and they always offered the ability to do what the alphabet list offered too, which I imagine is why the alphabet was removed - there is no point in having two different tools that do the exact same thing. You can still jump to surnames starting with any letter by simply putting that letter into the surname search field. But even better than that, the search fields offers way more versatility than the alphabet did, because you can also do the same for the first name field (shown above), and you can use more than one letter, so you can quickly bring up all "Mc" or "Mac" surnames, for example (shown below). I believe this was always an option, a lot of people just apparently didn't realize it.

The bigger change is in hints, clicking on a hint now brings up a side bar to preview the hint (shown below) instead of it loading a whole new page. People are complaining that it requires more clicks to confirm and attach the hint now, but that's just not true. In the past, you had to click on the hint, it loaded a new page where you could review the record, then you clicked "yes" to the hint being correct and it loaded the page that allowed you to edit the data you're adding, then you click "save to your tree" and you were done. That's 3 clicks.

It's the same now. You click on the hint, but instead of it loading a new page, the side bar pops up. Instantly, I felt this was an improvement because you haven't left the person's profile page, so you can still fully view and compare all their data, and sources, etc. You can click on Facts, Gallery, etc and the side bar remains up, allowing for a full comparison (shown below). In the past, the only way to do this was to right click the hint and open it in a new tab, which you can still do, but now there's no need. I always did this, because I generally want to refer back to the profile while checking a hint. Now, finally, I don't have to open a new tab, which is going to make my workflow much more efficient. 

Even if you didn't open the hint in a new tab in the past like me, the number of clicks is still the same. After clicking on the hint and the side bar popping up, you click "yes" to the hint being correct and it loads the page where you can edit the data you're adding, and then you click "save to your tree" just like before. That's 3 clicks.

So I'm really not sure what the fuss is all about. The changes either won't slow your workflow, or they will actually improve it. Give it a chance, you might find it works better.

23andMe: Worse and Worse

It's never been a secret that I feel 23andMe is the worst DNA option of the 4 main companies when it comes to using it for genealogical purposes. While they do seem to still have the most reliable ethnicity percentages, and they offer the easiest way to get health reports that may actually be useful, when it comes to using our DNA matches for genealogy research, 23andMe are an epic fail, and over the years it has just become worse and worse. Between not hosting family trees/gedcom uploads, and capping our match list more and more, it's hardly surprising I've gotten very little use out of it and now it's only gotten worse. 

Years ago, back when I originally tested, they hosted uploaded gedcoms (family trees). Anyone who has done DNA based tree research knows this is essential to getting use out of your DNA matches. But not long after, 23andMe obviously decided this was a waste of their server space, but they at least attempted to provide an alternative. They did a deal with MyHeritage (long before MyHeritage got involved in DNA themselves), where gedcoms at 23andMe could be moved to MyHeritage, and a link to your MyHeritage tree would automatically appear in your 23andMe profile. Unfortunately, this didn't last long because at MyHeritage, you have to subscribe to view other people's trees, and probably a lot of 23andMe users weren't going to subscribe just for that reason. So it quickly became apparent that this was rather useless for most people. And of course, MyHeritage eventually began to sell their own DNA test, so they didn't want to be associated with any other DNA company at that point. 

For a while, 23andMe simply didn't host any trees at all. They did offer a spot in your profile to paste a link to an off-site tree. But most people didn't bother, and just like at MyHeritage, viewing trees at Ancestry.com also requires a subscription (though they now have a sharing option, they didn't at the time). So unless your tree was available somewhere for free, this was still useless, which is why most people didn't bother. It seemed like 23andMe had abandoned any pretense they ever had at being genealogically useful.

Recently, they did trial an option where you could link your FamilySearch tree to your 23andMe account. This finally seemed like a great solution - it's free, and it's integrated, not just a link to an off-site tree, but something you could view at 23andMe. Sadly, not many people participated in the beta trial, and after months of beta testing, instead of officially adding it as a feature, it disappeared without a word from the company (something that happens a lot). I don't know if it's because not many people tested it out so they thought it wouldn't get used, or if it was something else, but one day it was just gone, so once again we're left with nothing.

Granted, they have recently added a tree feature that let's you add your ancestors and DNA matches to it, which helps visualize how you are related to some of your closest matches. But it only goes back to 2nd great grandparents (3rd cousins), and more importantly, this is for your own private usage only, no one else can see it. If no one else can see it, no one else can make any use of your tree for genealogical purposes. So this is not really what we actually need.

I did also notice they are advertising a "free quote for a genetic genealogy research package offered by Legacy Tree" which I assume includes a family tree. But not only does that cost a lot of money, it's totally unnecessary if you've already build your own tree. And even if you have a tree built at Legacy Tree, it's not integrated into 23andMe.

If that's not disappointing enough, let's talk about our match list, called "DNA Relatives". 23andMe has always capped our match list. At one point, it was capped at 1,000, then they upped it to 2,000, which was great. And more than that, they offered way to search for and find other people you shared DNA with, that you could connect with and add to your match list. But over time, they gradually removed those features, making it harder and harder to expand your match list. Of course, your match list still expanded as more people tested - it's not like people got bumped off the end of the list as new ones came in. Apparently, 23andMe have decided that these essential matches are taking up too much server space and have quietly reduce our match list to just 1,500 people. 

In comparison, I have over 22,000 matches at AncestryDNA, and that's not just because more people have tested there, it's because AncestryDNA's matching threshold is 8 cM. At 23andMe, capping my list at 1,500 people (actually 1,454 for me, whereas previously I had over 1,800) means my most distant matches share 20 cM with me. I regularly point this out, but shared segments of 15+ cM have a 100% chance of being identical by descent. That means 23andMe are excluding thousands and thousands of matches that have a 100% chance of being identical by descent. It's always been a real bummer, and in some ways I'm not sure that losing a mere 400-500 matches is that big of a deal since I never got much use out of 23andMe's matches anyway, thanks to their lack of hosting shareable trees/gedcoms. But here's the worst part about the new changes at 23andMe...

They are offering an option to expand your match list to 4,500... great, right?! Except it's going to cost you. Firstly, if you haven't tested on the V5 chip and/or haven't paid to include Health reports, you'll have to upgrade your test. The expanded service only applies to people with an Ancestry+Health V5 test (because it includes extra health reports too, not just the extended match list, and that requires the raw data in the V5 chip). If you tested previously on an old chip, you can upgrade to V5 Ancestry+Health for $99 (normally $199). If you're already on V5 but don't have Health reports, the upgrade to Health will cost $125.

And on top of that, you will have to pay a yearly subscription of $29. While that is not a huge amount of money, no other DNA company requires a subscription to access extra DNA matches. Especially when you consider that even the expanded match list you have to pay extra for is only a small fraction of what you'd get at AncestryDNA for no extra cost, this offer seems of poor value, unless of course you're actually after the extra health options that come with it, that AncestryDNA doesn't even offer. 

What that tells us, is that just like always, 23andMe are really more about the health and ethnicity side of DNA testing, whereas AncestryDNA are geared more towards genealogy. That's not surprising, since Ancestry.com are, after all, a genealogy website, whereas 23andMe are not. But it still means that for us genealogists, 23andMe is not the ideal company to test with. 

For more info, see 23andMe's page on their "23andMe+ Experience".

AncestryDNA Ethnicity Update in the Works

Recently, AncestryDNA have announced that in coming weeks, they will be rolling out a new ethnicity update with a banner at the top of the DNA homepage. Clicking on the banner takes you to a page that promises new regions and their most precise breakdown yet. It includes a map showing the different coming regions, but we don't get any details beyond the map. It also claims to have over 40,000 samples in their reference panel, and looking at the new white paper shows it's actually over 44,000 which is only a slight increase from the last update which used just over 40,000. With only a minor increase in the samples, that suggests much of the update might be in a change to the algorithm. The FAQ on the announcement page provides a little more info, but it doesn't actually detail what the new breakdowns will be.

Left: European regions before the upcoming 2020 ethnicity update. Right: European regions after the upcoming ethnicity update.

However, we can get a little bit of a preview by looking at our newest DNA matches, who are obviously already receiving the update. If you go to your DNA match list and click "groups" and select "new matches", then look at the ethnicity comparison with them.

From that, you'll be able to see some of the new regions and how they will be broken down. For example, Wales will now be a separate category, no longer lumped in with England/NW Europe, and Ireland and Scotland are now separate categories too.

In southern Europe, Italy will be split up into Northern and Southern Italy (see below). This isn't shown in the before/after map Ancestry's announcement page provides, which is why I say it's not very detailed and I don't think it's giving us the full picture. Additionally, Cyprus will be getting it's own category, no longer a part of Turkey/Caucasus or the Middle East.

It doesn't look like there's much, if any, changes to Africa, Native America, or Asia, but that's because the before/after map on the announcement page isn't reliable. The "before" map seems to actually be using the regions from two updates ago, not what it is now. That's misleading, and if you compare the "after" map to what it is now, there's no difference in Africa, the Americas, or Eastern Asia, only to Europe and West Asia. But the new "after" map doesn't include some new regions we know are going to exist (like Wales). So that map really isn't reliable and doesn't really tell us much. However, the map in the new white paper looks like it does include new areas. It's not interactive and doesn't let us zoom in to see details, but it does appear that there are indeed new regions in other parts of the world too, not just Europe and West Asia. I am not sure why the before/after map on the announcement page is not actually showing the new regions/breakdown when that is supposed to be it's sole purpose.

Left: Africa on the announcement page, supposedly what the update will look like but its exactly the same as it is now. Right: the updated Africa map from the new white paper - what the new regions will actually look like after the update.

They've also already updated their white paper with the European PCA chart.

Here we see quite the breakdown into individual countries, but these are just where their samples come from and don't necessarily reflect how they might group the populations in our results. Like the last one this PCA chart doesn't really show much difference between Portugal and Spain, so attempts to split them up might not be accurate. And of course, we are still seeing massive overlap among all of Northwest Europe. The British Isles, Germanic/France, and Scandinavia all share a significant genetic overlap that still makes them difficult to tell apart in many cases. There are some German and French samples not a part of that group, but there are also many which are. This is understandable since France also shares some overlap with it's neighboring Spain, while Germanic Europe shares DNA with it's neighboring Eastern Europe.

But particularly in regards to the new results splitting up regions like Ireland and Scotland, or England and Wales, I'm skeptical about the reliability of that since the PCA chart shows no new genetic distinction between them.

Additionally, I noticed that European Jewish is missing from the PCA chart, which is a shame because it's always interesting to see how genetic unique they are. And as ever, the PCA chart only includes Europe for some reason, we never get to see ones for other areas, which might be enlightening.

This will be AncestryDNA's third update in three years - does this mean we can expect the norm to now be an update every year, even if it's only some tweaking to the algorithm? We can only wait and see.

MyHeritage's Photo Enhancements

Previously, I did an analysis of MyHeritage's photo colorizing tool. Since then, they have also added an enhancement tool, which identifies faces in photos and "enhances" them by making them sharper and clearer with more details. They also appear to have improved the colorizing. I noticed how the same photo I had colorized before, which left some areas like a hand here or there uncolored, are now magically colored (see bottom of article for example). It's still not perfect, but it's improving.

Normally, both tools are limited to 10 photos with a free account, so you have to subscribe with the Complete Plan (the most expensive one) to use unlimited photos with these tools. But right now, MyHeritage are trying to entice people to subscribe by making the tools unlimited for free for one month. They are obviously hoping people will be so impressed with the tools and get used to using them on unlimited photos that when the month is up, some will subscribe to continue having access to them. But personally, I'm not about to spend another $300 a year just for access to these fun little tools so I'm making the most of the free access while I can.

Much like the colorizing tool, the enhancement works best on images that don't have too much degradation or blurring. If you click the above photo of my Nan to enlarge it and see details, you'll see it had only some minimal blurring and the enhancement tool made it very sharp and clear, a much better picture of my Nan. You'll note that it does not remove scratches, spots, or other surface damage to the photo though, even when they appear on the face. And yes, you can enhance it and colorize it at the same time (see examples below), I just chose not to on this one of my Nan to show you the enhancement alone.

It also works best on faces that are closer to the camera - the smaller/further away the faces are, the less effective the enhancement is, and sometimes it's not effective at all. A few photos I've tested so far (see below), the faces were so far away and so blurry that the tool didn't even attempt to enhance it (though it did seem to identify it as a face since it colorized it correctly as skin colored). The ones that were enhanced were minimally done. 

The enhancement tool only worked on 3 out of 5 faces here, because they
were too small and blurry. The 3 enhancements were minimal too (see below).

A close up comparison of one of the small faces in the above photo with the minimal enhancements

Additionally, sometimes the enhancement leaves the face looking a little plastic and weird, like the person is wearing a mask (see below). This is more likely to happen the more blurring there is to the photo and faces, partly because it's difficult to enhance something so small and so unclear, but also because the tool only enhances faces and nothing else. A sharp, clear face next to blurry hair and clothing just looks weird. But in some cases, it's better than nothing, and it does give us somewhat of a better idea of what someone looked like.

Note how the one on the left looks fairly normal but the other two appear mask-like

Keep in mind though, that this tool is attempting to create data where it doesn't exist, so there comes a point on a heavily doctored photo where it may not be an accurate representation of someone's face. It's fun to explore, but for example, I would avoid using it while comparing people in two different photos to determine if they are the same person, or related. While it's tempting to use enhancements to do such comparisons because they seem clearer and sharper, making it easier to compare, it could actually be wrongly altering someone's appearance and leading you to the wrong conclusion about their identity.

So just like with the colorizing tool, the effectiveness of the enhancement tool can be a little hit and miss. It handles some images better than others, and there does come a point where certain photos and faces are too far gone to recover. Have fun with it, but don't expect too much from it with all your photos.

A comparison of the initial coloring of a photo (left) with the updated coloring of it (right), note
some of the hands that were previously uncolored are now a correct skin tone. There's also some
minor difference in the the darker skirts. Click to enlarge.

Changes at Ancestry.com

Recently in the news, there's been more than one announcement about changes happening at Ancestry.com and AncestryDNA. First, AncestryDNA announced they were raising the threshold for how much DNA you have to share with someone in order to match them, and then came the news that Ancestry's primary owner, Permira, are selling to another company called Blackstone.

Cue the freak outs.

But really, everyone, it's going to be okay. Here's why.

AncestryDNA are raising the threshold for matching from 6 cM segments to 8 cM. I know this seems extreme, but remember that the vast majority (thousands) of your matches under 8 cM are identical by state, which means they do not share a common ancestor, at least not within any kind of genealogical time frame. And even those matches which you can identify a most recent common ancestor with, at this level of DNA, it's so unreliable, you can't be sure the shared DNA is actually coming from that ancestor after all. Because of this, these low matches can actually be leading you astray, not breaking down the brick walls you thought they would. By raising the threshold to 8 cM, AncestryDNA are assuring that the majority of your matches will be identical by descent and therefore more likely to be most useful to your research.

In fact, in the past, AncestryDNA was often criticized for their threshold being as low as 6 cM, as at least 7 cM was considered more the standard. And if you think 8 cM is extreme, don't even bother testing at 23andMe, where rather than have the same cM cut off point for everyone, instead they cap your match list at 2,000 matches (or if you tested long enough ago, it may be 1,000). Whatever cM your matches share with you at that 2,000 cut off is where they draw the line. So, for example, let's say your 2,000th match shares 12 cM with you - they will include everyone that shares 12 cM with you even if it goes above 2,000, but that's it, that's your cut off point, you won't have any matches sharing less than 12 cM with you. Personally, I was unfortunate enough to test when the cut off was 1,000, and they didn't raise it. My cut off point at 23andMe? 20 cM. That would basically only be my estimated 4th cousins or closer at AncestryDNA. Imagine your match list being cut off at only estimated 4th cousins or closer.

Additionally, AncestryDNA have promised not to remove any matches that you have marked in any kind of way. If you starred them, put them in a group, left a note, or messaged the individual, those matches, even under 8 cM, will remain on your match list. So you're not going to lose any info that you've already established, only the ones that you haven't gotten anywhere with yet, which are the ones most likely to be identical by state anyway. 

So AncestryDNA are not being unreasonable by raising their cut off to 8 cM. It's more in line with what is considered standard, it's still way better than the crappy deal you get at 23andMe, and you won't lose any established data. What I recommend doing is going through all your ThruLines or matches under 8 cM with Common Ancestors and star or group them somehow, so you at least don't lose those. For me, at least, this is worthwhile because most of my ThruLines wind up being accurate. You may also want to star ones that have trees - the ones that don't, you probably wouldn't get very far with anyway. You can easily do all this by going to your match list, selecting "Common Ancestors" and then under "Shared DNA" select "custom centimorgan range" and put in 6 to 7. For those with trees, click "Trees" and alternately select each of the options. This will help prevent you from losing those low matches which have some chance of actually being useful.

Now let's look at the second news: a change in ownership. When the news first broke, there was a lot of shouting about how the company taking over is a private equity firm, as though this is some of dirty word that we should all be terrified of. Well, guess what? The main company that's selling, Permira, who have been the primary owners of Ancestry for the last 8 years, are a private equity company too. The company is merely changing hands from one private equity company to another. (For the record, there are a few other companies that also held shares in Ancestry. and two have also sold to Blackstone, one is retaining a minority ownership along with Blackstone. They are also private equity companies, but Permira was the primary share holder, as Blackstone will now be).

Of course, any time ownership changes, it could mean changes for the company. People mainly seem concerned with their DNA and who has access to it. Blackstone responded to this by making a statement reassuring users that they won't actually have access to DNA data:

"To be crystal clear, Blackstone will not have access to user data and we are deeply committed to ensuring strong consumer privacy protections at the company," a spokesperson for Blackstone told Motherboard in an email. "We will not be sharing user DNA and family tree records with our portfolio companies."

Of course, companies can lie or change their mind, but there's no reason to think that's going to be the case. Remember, Ancestry was in the hands of another private equity company for 8 years up to now and no one seemed to have a problem with that. 

But wait, I hear people saying, the concern is that Blackstone have their fingers in the health industry, which surely means they acquired AncestryDNA to mine genetic data for the health companies they own! Our DNA is going to be sold or given to other companies without our consent!

It's true that Blackstone also has investments in the healthcare industry, that's what private equity companies do - they invest in other companies and frequently have investments in multiple different industries. But guess what? So does Permira. Yep, Permira also has investments in healthcare, yet that was never a problem for the 8 years that Permira were the primary owners of Ancestry.com and their massive DNA database.

Look, I'm not going to say nothing could ever happen to the security and privacy of your genetic data. Hacks happen. Illegal deals happen. Companies can violate their own TOS. But that's always going to be a risk, no matter the company you tested with, and no matter who owns it. This particular sale doesn't mean it's anymore likely than before or with any other company. If it really, deeply concerns you, then you shouldn't ever have your DNA tested anywhere, and if you already have, you should delete it immediately. If you understood the potential risks (which are low, in my opinion) and were okay with them before, there's no reason to suddenly be concerned now.

So before you buy into all this fear mongering, just do what genealogists always do: research! Before you freak out, do some research about these companies buying and selling Ancestry. Do some research about the validity of low DNA matches. You might come to the same conclusion I did. And if you don't, fair enough, at least you did your research and came to an informed decision, which is more than I can say about most of the fear mongering going on.

FamilySearch's Unindexed Images

Recently, FamilySearch made an update to their website in attempts to draw more attention to the wealth of unindexed records in their catalog, all available for free. The records available by using the search or even the collections list are a drop in the water compared to their vast catalog. You have always been able to access the catalog by click on "Search" and then "Catalog" from the drop down menu. Although it's readily available, it generally does not get used by people who don't know what it is or how to use it. Due to the fact that the images are not indexed, you can't search them by name or other details, you have to manually browse the images. To find the right collection, you have to search by location, collection title or author, keyword, subject, or, if you know it, film number (because the catalog used to be for looking up film rolls you could order). It's usually best to search by location, but this also requires knowing what jurisdictional "level" records are held at. For example, probate records are usually held at county level, so if you're searching for probates in Bucks County, Pennsylvania, you have to search the location field for Bucks County, Pennsylvania. Looking under just Pennsylvania will not find collections cataloged at lower levels, like county or city.

FamilySearch Catalog

FamilySearch's answer to this was to create a new option under the "Search" menu at the top of the site called "Images". Here, they have tried to simplify a way to find unindexed collections by making the location search field the only option unless you click on "more" and again on "advanced", which allows you to also search by time period, record/collection type, film number, etc. But unfortunately, the results seem to be lacking a lot of existing collections and the ones it does include are organized in a very convoluted way.

In the catalog, if I search for Philadelphia, Pennsylvania, I get a list of record/collection types, which I can click on to see the individual collections and select any of them. Fairly straightforward. In the new "Images" search, I get a huge list of over 8,000 results, many of which seem to be from the same collection but for some inexplicable reason, are broken down into multiple results (it appears they are broken down by individual film roll number, even though the film number isn't included in the results list). This means, for example, there's dozens of listings of probate collections, sometimes even multiple listings for probates from the same year! How am I supposed to know which one to use? In the screenshot below, it shows how if I'm looking for a Philadelphia probate record from 1913, there's multiple listings for it, and they aren't duplicates, they're different records. This is going to be far more confusing for people than the catalog ever was.

Of course, I can narrow down the results by using those more advanced search options, like adding a year and record type (1913, Probate), but that doesn't solve the problem of there being multiple results just for 1913 Philadelphia Probate records. In fact, there's 115 results! How on earth am I supposed to know which one to use? There is literally nothing distinguishing them from each other except sometimes the image count.

Maybe I just haven't gotten the hang of it yet, but so far, I haven't had any luck finding actual records or collections I know exist in the catalog with this new "Images" search option. As far as I can tell, it looks like they are not including collections that are only visible at a Family History Center or affiliate library, which is a huge portion of their catalog.

I do not understand the purpose or function of this new Images search. They now have 3 different ways to find records on their website (for some, it was confusing enough as it was to have 2 different ways), and none of them include their entire database of records. Honestly, I suggest you skip this and just use the catalog or search engine as usual.

MyHeritage's New Colorizing Photos Tool

Ever wanted to have your old black and white family photos colorized, but don't know how to do it yourself, and don't want to pay a professional an arm and a leg for it? Well, MyHeritage just launched a new free feature from DeOldify that will instantly colorize black and white photos. But how well does it work? I was a little skeptical and couldn't wait to test it out.

The photo I tested was just a simple portrait from about the 1880s. I was surprised how quickly it colorized, and I was pleased with how nice it looked but I realized that it actually just looked like a sepia tone had been added to it. I don't think that was the intention, and the skin tones did have a more fleshy color, but everything else looked like it'd just been sepia toned. A little disappointing.



Additionally, you may not be able to see it very well but this man's eyes were clearly light colored - blue, grey, hazel, etc. Something like that. But zooming in on his eyes shows the sepia/fleshy colors of the skin seems to have just been overlaid on his eyes, making them look brown, as if there was no attempt whatsoever to even color the eyes at all.

And while we're on the subject, they are beautiful eyes, aren't they? I've always thought this guy looks a bit like Leonardi DiCaprio.

To show you the difference between what a computer can do and what a human can do, here is my colorization of the same photo (including spot cleaning/restoration):

Back to MyHeritage. I then tried it with a group photo, thinking the multiple faces, garments, etc would add some variation to the possible colors. This was much more impressive:

Not bad for an automated system! Granted, the photo's highlights are a little blown out in places and some of the faces are blurred from too much movement, but colorizing system handled it pretty well in spite of that.

What's even better is that this is a high resolution image I used. I was a little worried that such an advanced tool available for free would only accept low resolution images (maybe charging for high resolution), but this was a fairly high resolution image and it not only accepted it, it still only took a few seconds to generate a color version. Unfortunately, although it will accept high resolution images, there is a limit to how many photos you can colorize if you have a free account. They don't tell you this anywhere but choose your photos carefully because you only get 10 of them, and deleting previous ones doesn't allow you anymore.

And the colorization still isn't perfect.

You may notice how it doesn't exactly take much risk or leaps with the colors it chooses. The men are in black suits, the women all seem to be in black and dark navy dresses, and the kids are all in white or neutral colors. You can probably understand why - I suppose they don't want a man's suit turning up bright red or something equally unrealistic for the era and gender. That's the downside to using a computer instead of a human who can distinguish these things and safely choose a greater variety of colors to apply.

Additionally, when I zoom in, there are areas that look like something almost resembling purple fringing except not along high contrast edges. You can see these sort of random purple splotches in the zoom-in below, particularly in her hair (pretty sure purple wasn't a trending hair color in 1880 Wisconsin), and sleeves. This is just a small area of the photo but these purple spots turn up everywhere if you look closely enough.

There's also some areas of the image that the computer seems to have some difficulty coloring. You'll note above how her one shoulder does not appear colored, or at least seems to be a different color from the rest of her dress - more of a sepia tone again. You see it most prominently in the skirt behind this child below:

At first, I thought maybe it was due to a shading variation in the original that may have fooled the system into thinking the difference in the shading meant a difference in color, but that is not the case. You can see in the original, there is no shading variation.

I guess the tool just sometimes has difficulty identifying edges and items so when it's unsure, it seems to do this. It's understandable, I suppose - after all, what is required to accomplish this in mere seconds must be an incredibly complex algorithm and coding, and it's provided for free, so I can forgive it for not being perfect.

Lastly, you may have noticed MyHeritage put their logo in the bottom right corner of the colorized image, and a little paint palette icon in the lower left. To avoid these, I'd recommend adding a superficial border to your image where the logo and icon will show up, which you can then crop off later.

I decided to try another photo (above), this time with more elements in it - horses, a house, etc to see if the same problems occurred, and they did. Once again, you can see all the clothing colors are very neutral. And again, you can see some weird rainbow-like discoloration at the top of the house.

And again, there were obviously some spots where the computer had difficulty colorizing or distinguishing between items - as you can see below, the hand on the shoulder looks like it either hadn't been colorized at all or it's blending in with the color of the other boy's jacket. Conveniently for the computer, it chose to "color" this boy's jacket grey!

So if you don't want any creepy dead hands like this, or your ancestors had blue eyes instead of brown, it's best to hire someone to do this for you instead of relying on an automated system. There are also Facebook groups with generous people who will colorize your photos for free, but be aware that Facebook doesn't easily support high resolution images like this does. This option from MyHeritage is still pretty impressive for what it is though, and if you're not bothered by the small problems that you might not even see very well when zoomed out, this will be amazing tool for many people. At the very least, I enjoyed seeing some color in the faces of my ancestors and relatives, as it seems to make them come alive a little more.

I haven't checked it out yet but there's an alternate colorizing option found at ColouriseSG. It appears to be free.

Yet Another AncestryDNA Ethnicity Update

AncestryDNA's latest ethnicity update

AncestryDNA had promised another update to their ethnicity report this year, despite there only have been one last year (which was still being rolled out the last people earlier this year too) and it seems to be rolling out already, with little to no beta test like last time.

The reference panel has now jumped up to 40,017 samples, but don't assume that means it's going to be more accurate. As ever, it depends on the individual, and just how admixed you are. Also, unlike last year's update, we're not seeing any new percentage groups in Europe (though I do believe there's new ones in the Americas), just a change to the percentages (though there are some new Genetic Communities).

For me personally, the report just keeps getting worse and worse. They have now removed my Scandinavian results entirely, which is a shame because I did have one Norwegian great grandfather so I should be roughly 12-13% Norwegian. Of course, I may have inherited less than that from this particular great grandparent (and most ethnicity reports do reflect that), but I did inherit something from him (a great grandparent is not distant enough to not inherit anything), so the complete wipe away of any Scandinavian result at all (not even anything in Sweden, let alone Norway) is incredibly inaccurate and a shocking degradation.

Meanwhile, they've increased my French results even more to 21% (was 18%), even though I still don't have any recent French ancestry (I had a few French Huguenot ancestors from back in the 1600s, but we're talking about 2 or 3 out of 1024 or 2048 8th-9th great grandparents - a drop in the water that is generally too diluted to be detectable, and certainly shouldn't be as high as 21%).

They do appear to have removed a bunch of the low percentages that were likely noise, or coming from neighboring regions. That seems like a plus at first, but since my valid Scandinavian results were among them, it doesn't seem like a great improvement.

The other only changes were the swap of Germanic and England/Wales/NW Europe in terms of percentage amount. Previously, I had highest results in England/Wales, which was consistent with my tree, now I have highest results in Germanic.

My Italy results remained my the same, which might seem like a positive (at least they didn't go down), but they are remaining stubbornly at 12% when I should be about 32% (and this one I know for sure because my paternal grandfather tested and I share 18% with him, leaving me to share 32% with my Southern Italian paternal grandmother). I suppose the rest of my Italian DNA is showing up in France. If you look below, the map for France does cover Northern Italy, but of course I don't have ancestry in Northern Italy, only Southern, and AncestryDNA are the only ones who seemingly can't figure this out (well, they are the only ones trying to give France it's own category apart from LivingDNA which more accurately give me 0% in France).

In addition to these changes, we've also seen a redefinition of the maps. Although the categories haven't changed, they've just altered the maps to better reflect the regions each category is supposed to cover. At the same time, they've taken away the page that included more in depth details and background on each category, which is a major regression if you ask me. We now only have a short paragraph with info on the region and it no longer lists the "secondary" areas that each category covers, even though the maps still include them.

As you can see above, there is no longer a link to click for more information. AncestryDNA keep removing more and more details and information, as though they think we are too stupid to understand it. People are simply now going to have more questions and understand the results even less.

Also note how the map for France still shows the surrounding areas that this category includes, but for some inexplicable reason, AncestryDNA has decided to only include the "Primarily located in" information, which only lists France, and no longer lists the "Also included in" areas shown in the map, like Spain, Italy, etc. That won't be confusing for people at all.

I really don't know what they are thinking sometimes, and since they rarely ever explain or communicate their choices to their customers (apart from "we think this is better"), I doubt we'll get an explanation. We still never even saw the return of the Average Admixture chart (shown left) when it was lost in last years update, which was so useful for understanding what regions shared a lot of DNA with what neighboring areas (God forbid they actually admit that is a thing).

Having said all that, the report does seem to have improved for people who are less mixed, just like last years update. My half Italian dad's results went from 44% Italian (which is reasonably close to half) to exactly 50%. The rest of his results didn't change much, which means he still has too much in England/Wales and not enough in Germanic (only 5% Germanic when he should be more like 30%).

My paternal grandfather's results have evened out a little bit - his tree is roughly half German and half Scots-Irish (maybe some English), and his update says 51% England/Wales, and 31% Germanic, a slight improvement from 67% England/Wales and 23% Germanic.

My husband is the least mixed of all. He is a British native, born and raised, but half Irish. Grouping his ancestry the way AncestryDNA does, he'd be about 40% English and 60% Irish/Scottish. Last year's update gave him 38% England/Wales and 61% Ireland/Scotland with a random noise level of 1% in Benin/Togo. Today's update has him at 43% England/Wales and 56% in Ireland/Scotland. Not much of a change and still pretty consistent with his background, apart from that random 1% they've now put in Nigeria.

As for my mom's results, her previous update from last year saw no results whatsoever in Germanic even though she should be around 20%. Meanwhile, her Norwegian results, which should have been around 25% (that Norwegian great grandfather of mine was her grandfather) were inflated at 40% last year. She now had 10% Germanic, which is better than nothing, and 27% in Norway, which is almost spot on. The only thing a little off is the 14% in Sweden.

How have your results changed? For the better or worse?