Ancestry shared matches and 'relatives in common'

[PhilGee]

Perhaps the following may help, starting from a sentiment I agree with.

I guess it is a case of having a range of tools at our disposal to help with those more problematic matches. The shared match feature alone has too many limitations.

The limitations are there for a good reason - there would be so many mistakes made by unwitting that it would bring Ancestry DNA into disrepute.

Hmm. Ancestry don't seem too worried about their reputation when it comes to things like Tree Hints, and all the mistakes that they lead to.

I'm not aware that Tree Hints have damaged Ancestry's reputation.

However, I am at a loss as to why Ancestry providing "accurate" information that two samples have matching segments that are not in their timber/pile-up database should damage their reputation. It is the application of this information that could be erroneous - and probably no where near as "idiotic" as hints.

Having set this up, I note that it could have been helpful if I had not failed to include quote #2 in the post of quote #5.

 

[peter]

I'm still none the wiser as to the relevance of pile-ups to the discussion. Can you please explain?

 

[Bob Spiers]

That's a very cryptic comment. Please explain exactly what you meant by this sentence in your earlier post - I can only guess at what you intended it to mean and none of the interpretations I've come up with so far make sense to me:

I hope ATINZ does not read your 'Cryptic comment' note about PhilG's post Peter, as she recently accused you of same on another matter. However, that aside, I quite agree with you, but then I rarely understand Phil at the best of times, even when not just posting one liners (sorry Phil).

 

[Pauline]

In my experience the limitations of the shared match feature at Ancestry already lead to wrong assumptions being made, so I still don't really understand why it should be thought such a bad thing to provide a wider range of tools.

 

[PhilGee]

I'm still none the wiser as to the relevance of pile-ups to the discussion. Can you please explain?

First I really need to know what you consider to be the subject matter of the discussion, however:-

I consider it is about DNA matches producing shared matches that appear to be erroneous and the possible benefit of Ancestry providing details of (all?) the segment(s) for a match. I therefore added a comment that, in my opinion, matches are to a high level of confidence ("accurate") because of the filtering done by Ancestry's proprietary tools and databases, especially the database of excess IBDs preventing gross matching errors.

I believe I have said previously that Ancestry may not wish to provide segment data to protect their proprietary database. If only to easily group matches, it could be useful. The attached is a table that started with a match between myself and "TS" with added shared matches and shared matches of shared matches! Personally, I would prefer grouping by segments.

 

[peter]

In my experience the limitations of the shared match feature at Ancestry already lead to wrong assumptions being made

Do you have an example? I manage around a dozen Ancestry DNA tests, but can't think of any case in which shared matches led me to make the wrong assumption.

I still don't really understand why it should be thought such a bad thing to provide a wider range of tools

It wouldn't be a bad thing if the tools were used properly and produced reliable results - but sadly that's rarely what happens.

 

[Pauline]

Do you have an example?

Any time we make an assumption about an unidentified match based on a shared match with a documented cousin we may potentially be making a wrong assumption.

Much of the time, we may be reasonably safe in assuming that a shared match indicates that all 3 of us share a common ancestor, but in a significant minority of cases a shared match indicates 'relatives in common', where the 3 do not all share a common ancestor.

I colour code all my documented cousin matches, and if I have an unidentified match who has, say. ten independent shared matches with cousins in my dark blue group and one in my yellow group, then it is likely that the shared matches with the dark blues indicate a shared common ancestor, while the shared match with the yellow cousin probably indicates a 'relative in common'.

However, if an unidentified match has just one shared match, then I have no way of knowing if the 3 of us share a common ancestor or if this is the result of 'relatives in common'. And similarly if an unidentified match has shared matches with, say, one dark blue and one yellow. These latter scenarios may well occur in areas of our tree where we are trying to break down a brick wall, and have the potential to really muddy the waters.

 

[peter]

Pauline, you haven't given an example of a wrong assumption being made, you've given an example of how a wrong assumption might be made.

Having more details about the matching segments doesn't mean that wrong assumptions are less likely to be made - it could work the other way. It all depends on the degree of knowledge and expertise of the individual concerned.

 

[Pauline]

Pauline, you haven't given an example of a wrong assumption being made, you've given an example of how a wrong assumption might be made.

I'm not sure that quoting a specific example will add anything useful to what I wrote above. All it would add is names of other researchers, or names from my own matches and tree, and I don't think that's a good idea. But in general, I am aware of mistakes in people's trees due to wrong assumptions made from Ancestry shared matches.

I do fully appreciate that providing information on matching segments could also lead to wrong assumptions, and I also appreciate that showing shared matches for very distant matches is unlikely to be helpful.

However, despite the possible pitfalls of assuming too much from shared matches, and even without providing a chromosome browser, I think Ancestry could make life easier for its DNA users by reducing the shared match limit to, say, 15 cM rather than 20 cM, and by showing how much DNA our shared matches share with each other.

 

[PhilGee]

Just a comment on 'relatives in common':- I have paternal ancestors from the area between Leicester and Loughborough and maternal ancestors in and around Wellington, Somerset. At some point in the 1820s/1830s one relative travelled from Wellington to Leicester and married there (give DNA a pat on the back!). It is not difficult to see the possibility that maternal DNA could be found in a match with an obvious link to the paternal side and no obvious link to the maternal side.

As an aside, I have a few matches linked to this family, but it seems that everyone has a unique version of the family "tree" before 1800 - the time period for 'common ancestors'!

 

[peter]

I do fully appreciate that providing information on matching segments could also lead to wrong assumptions, and I also appreciate that showing shared matches for very distant matches is unlikely to be helpful.

However, despite the possible pitfalls of assuming too much from shared matches, and even without providing a chromosome browser, I think Ancestry could make life easier for its DNA users by reducing the shared match limit to, say, 15 cM rather than 20 cM, and by showing how much DNA our shared matches share with each other.

Whilst I'd personally like Ancestry to reduce the shared match limit to 15cM - and did ask them to do this when I had the opportunity - it is going to substantially increase the number of wrong assumptions made by people whose ancestry is less diverse than ours. Although you or I might have only a couple of hundred matches that are 20cM or more, there are some users who have many times more - over 30,000 in one case.

I too would like to know how much DNA is shared between my shared matches but there is a privacy concern. Where Ancestry could provide more help is for people who manage multiple kits - it is harder than it needs to be to find out which matches I share with the cousins whose tests I manage.

 

[Bob Spiers]

It all depends on the degree of knowledge and expertise of the individual concerned.

I think those few words pack a real punch when trying to understand the complexities of DNA Matches and individuals who appear to Share with me and they to others. It is the classic case all over again as being..." I danced with a man who danced, with the girl who danced with the Prince of Wales" and expecting to find a DNA match within all participants.

Phil clearly loves his forensic type interrogation - depicted in the Screen shot accompanying posting #25 - and both he and Pauline would like to see Ancestry provide additional tools to allow them to further deconstruct Shared Matches. I very much doubt it would be of the slightest use in understanding who begat whom, when and where and how family 'x' features in the scheme of things.

For me the knowledge and expertise I have gained from conventional research - aided now and again by additional DNA information (and paying attention to what I see and read) - is normally sufficient to allow me to appraise a situation without the Herculean Labours others seem to enjoy .

 

[Pauline]

I very much doubt it would be of the slightest use in understanding who begat whom, when and where and how family 'x' features in the scheme of things.

It's mostly in brick wall areas of our trees, where conventional research hasn't provided any answers, that it can really matter to be able to sort the sheep from the goats. Elsewhere in our trees, you may be quite correct in saying these extra tools will make little difference.

When it comes to brick walls you may be looking at unidentified matches who have just one shared match each, or where all the matches likely to be of help in solving the puzzle fall just below the 20 cM shared match threshold. In these circumstances then, as they say, every little helps.