Ancestry Matches 7cM and less

[peter]

4 of the spurious matches are in the 6-8 cM range, so deleting all matches in that range would remove over 6 times as many genuine (or probably genuine) cousins as spurious ones from the common ancestor list.

ALL of my Common Ancestors are correct, and I've added Notes to all of them recording how we are connected, what the relationship is, and if and when I messaged them. It's surely not going to take very long for those who haven't already checked out their Common Ancestors to do so, or at least to add a coloured dot?

 

[peter]

But I agree with your reasoning why Ancestry have taken this decision, that their system cannot cope rather than improving match quality.

I doubt that's the reason, but we'll have to wait for the White Paper to be sure.

 

[Bob Spiers]

For researchers with ancestors who mostly originated from the British Isles Ancestry is streets ahead of every other site.

Totally agree on this point especially.

 

[Pauline]

ALL of my Common Ancestors are correct

I wonder how common that is, or if you are just lucky?

It's surely not going to take very long for those who haven't already checked out their Common Ancestors to do so, or at least to add a coloured dot?

Using the methods suggested here to speed things up, then unless someone has a huge number of Common Ancestor matches, just adding a coloured dot shouldn't take more than a few minutes.

 

[Bryman]

You can speed things up by creating custom groups based on (say) major surnames

It's surely not going to take very long for those who haven't already checked out their Common Ancestors to do so, or at least to add a coloured dot?

I already use the coloured dots to indicate which eighth of my tree (great grandparents) the DNA matches apply to when I have determined such. I am worried that 6cM and 7cM matches will disappear before I have been able to investigate more than a small percentage.

ALL of my Common Ancestors are correct

I think that you must be lucky. Two or three of mine have strange properties, like apparent fit in tree as sibling of my direct ancestor but individual not known to me or any other researchers that I know of. One of my latest notifications looks good back to my earliest ancestor in that branch (my tree) but the suggested MRCA is shown with a birth date a couple of years AFTER her daughter, presumably from the other person's tree!!

 

[Winksetter]

I think that Ancestry has been experiencing more and more technical issues because of volumes of data and of system usage. I imagine that they will try to justify the change as for the benefit of users when really this is due to their inability to cope. All of this will be hidden by the smoke and mirrors of DNA gobbledygook. Their system architecture seems to be flawed necessitating the volume reduction resulting in user data being suppressed. They do not know whether anyone is a relation to anyone else, even when apparently close. Users must be the arbiters of this, especially when we get to the distant relationships. In my case 6 to 7cM relations are my most key ones to me, it takes a long time to work on these. Names are not helpful always when determining relationships. I continue to recommend everyone to spread research outside Ancestry, this is becoming vital when they take arbitrary decisions such as this. Contrary to silly comments made on this thread Myheritage can be great, but I would never recommend its exclusive use.

 

[Andrew Lloyd]

But I agree with your reasoning why Ancestry have taken this decision, that their system cannot cope rather than improving match quality

As I mentioned previously, the reason why I know the distribution of my matches by cM strength is because in the past I have used a 3rd party tool to obtain the data. I know that some of these 3rd party tool providers have been issued 'Cease and Desist' legal notices by Ancestry in the past month which is why they are no longer available. So looks like Ancestry are certainly looking at improving performance at the moment. Cant say I have noticed any less 'Out backend services are overtaxed at the moment' since the edict went out though.

 

[Andrew Lloyd]

I haven't tried to work out what proportion of my matches fall into the 6-8 cM range but around 20% of my common ancestor matches are in that range

Not sure of the science behind finding a common ancestor are, but it certainly seems top heavy with regard to the strength of the match. For example 13% of my common ancestors are in the 6-8cM range whereas 73% of my matches are in this range. Compare this with 37% of my common ancestors in the 20+cM range which accounts for only 0.59% of my total matches.

Breakdown of matches as follows, all other tests I manage are within 2% of these figures:

cM % all matches

> 19 0.59%
19 0.19%
18 0.26%
17 0.35%
16 0.51%
15 0.76%
14 1.27%
13 1.72%
12 2.47%
11 3.61%
10 5.62%
9 9.30%
8 17.09% (of which 7.13% > 8.0, 9.96% < 8.0)
7 32.27%
6 23.99%

 

[Bob Spiers]

All of this will be hidden by the smoke and mirrors of DNA gobbledygook

...

% all matches> 19 0.59%-19 0.19%-18 0.26%-17 0.35%-16 0.51%

...etc.etc.etc

As good an example of DNA gobbledygook I have seen and certainly well beyond anything Ancestry produces.

Contrary to silly comments made on this thread Myheritage can be great,

but not quite so 'silly' as anyone suggesting MyHeritage is (or ever could be in the foreseeable future) 'great'... (and I have experience to back this up as many in the Forum will know )

 

[canadianbeth]

I have made several groups - Barratt, Bowyer, Joyce, Riches, Probably Joyce, Probably not paternal, No Shared Matches and Unknown. The 97 common ancestors are in the four first-named lists. They all have notes listing to whom they are connected. I started putting names into the other four groups originally but then just did not bother; there are just too many of them.

Yes good luck with MyHeritage, you are likely to find a match to the Queen of Sheba and other biblical and non-biblical characters from History,

Following Bob's comment, I think I shall just ignore the apparent DNA matches to my Riches/Moss family that MH has given me and just go with the Ancestry one, especially since I do have documented information as to that marriage. I still do not understand, however, how it can be so different.

 

[peter]

I wonder how common that is, or if you are just lucky?

Lucky, perhaps. Working through my brother's Common Ancestors today I came up with a probable error.

 

[peter]

Not sure of the science behind finding a common ancestor are, but it certainly seems top heavy with regard to the strength of the match. For example 13% of my common ancestors are in the 6-8cM range whereas 73% of my matches are in this range. Compare this with 37% of my common ancestors in the 20+cM range which accounts for only 0.59% of my total matches.

It's inevitable that closer cousins are more likely to show up as having Common Ancestors, in the same way that it's easier to work out how we're related to our close matches. Nevertheless it's interesting to see the statistics. I can't wait to see the White {Paper.

 

[peter]

Thank you for the heads up Peter. My first premise is that DNA connectivity is so very random. This is why I feel that ancestry.com’s decision you outlined is not in the best interests of customers. Their refusal to provide a chromosome browser and other tools does not allow us to investigate properly potential relationships we find independently, but tie us completely to the formulated pathways they determine. As users and customers I feel we should have the choice to determine which match is relevant and which match is not, because we are the ones that know our ancestors and cousins far more intimately than the computer model they use.
I have numerically fewer matches than many other people tested. Some lines on my tree are mainly devoid of matches. This I feel is mainly due to geographical bias and hence numbers tested. Thus I am regularly working in the low cM area of matching and have found connections with proper documented trees many times at the 6 to 8 cM range. My research would be so much weaker in effect without having access to matches they now want to withhold. Thrulines is a good tool but does not take into account the randomness and extension that can be achieved in our matches. I found connection back to 7 x g grandparents lately with several matches being involved, not just the one.
Ancestry may tell us there are work arounds to enable us to retain some matches, but that is only in the now and doesn’t take into effect any new discoveries in the future. A Red Herring entirely!
This proposal is not in the best interests of Users and as such we need to be able to make the relevant decisions that effect the research in our trees not the Monolith that is Ancestry.

My experience suggests you're wrong, but let's wait and see what the White Paper says.

 

[Pauline]

Lucky, perhaps. Working through my brother's Common Ancestors today I came up with a probable error.

Since many Common Ancestor matches are found by Ancestry extrapolating a match's own tree, then bearing in mind the number of distinctly dodgy trees at Ancestry, it is perhaps surprising how many of our Common Ancestor matches are not spurious.

However, it is perhaps worth saying, that although I have identified only a relatively small number of Common Ancestor matches as definitely spurious, there are quite a few others that I can't confirm either way.

Incidentally, I see this morning that Ancestry are now announcing the forthcoming changes in a banner on the DNA pages.

 

[peter]

Pauline, thanks for the tip - I'd been expecting notification on the main DNA page, not the matches page. See the White Paper for full details (I'm just about to read it myself).

 

[peter]

I've now read the White Paper, as I imagine many of you have. The bad news is that the 8cM cut-off seems to apply even when there are other larger segments, unless the total of the over 8cM segments exceeds 90cM.

You will note from the White Paper that 50% of 6cM segments and about 30% of 7cM segments are false matches. Fortunately we can save any of our existing matches which are potentially useful; what we won't know about are new matches.

 

[Winksetter]

I have online conversations with Ancestry and conclude that they have no idea of the implications of this change, what is worse they don't care really. They had no real answers at all, at least none that were not restatements of the Ancestry standard “line”. They kept saying the results would be more accurate, but could not answer the basic fact that lots of vital matches would be lost altogether, both now and any potential future ones.
I am amazed that people are not more angry about this, and also surprised that Peter is apparently so supportive of this change.

 

[peter]

I have online conversations with Ancestry and conclude that they have no idhopea of the implications of this change, what is worse they don't care really. They had no real answers at all, at least none that were not restatements of the Ancestry standard “line”. They kept saying the results would be more accurate, but could not answer the basic fact that lots of vital matches would be lost altogether, both now and any potential future ones.
I am amazed that people are not more angry about this, and also surprised that Peter is apparently so supportive of this change.

You're probably talking to someone in Customer Services - I was talking on Tuesday to one of the key people involved in the update.

I wouldn't worry about the existing matches you are going to lose - the chance of ever figuring out how you are related to distant cousins who don't have a tree is minimal. Similarly if they do have a tree but don't show up when you carry out the searches recommended in the Masterclass it's highly unlikely the match will prove of any value.

Future matches are more of a concern, but realistically there won't be many with large trees who haven't already tested. And there is always the option of GEDmatch for those who aren't satisfied with 10,000 plus matches.

 

[Andrew Lloyd]

The bad news is that the 8cM cut-off seems to apply even when there are other larger segments

That's interesting, I'm glad you were able to interpret that from the Algorithm. I have a common ancestor match who I know for sure is a 3rd cousin one removed. He comes in at 13cM across 2 segments, so using this algorithm would I be correct in thinking that even if he doesn't get the heave-ho he is in for demotion? I shall be watching to see what happens to him after the changes.

 

[peter]

If you want another opinion on the changes see this blog article by Blaine Bettinger, which explains things a lot better than I could.