You are right that with the recent crop of international adoptees, there are many who come from countries where DNA testing is unheard of (especially via a USA-centered website) and so will not get as many matches.

Here’s a recent Washington Post article which does not get into any technical details at all, but gives you an example of how even a young Vietnamese adoptee can learn valuable info from this testing:

DNA Testing For Adopted Children.

Actually, it’s the opposite.  I can’t tell you the number of people who have come to me for help because DNA testing proved that their grandfather wasn’t biologically related to their father (or some similar situation).  All the paperwork is in place….everything looks perfect….but no, he’s not your daddy/grand-daddy.  People think this is a recent phenomenon, but I’ve helped in a number of cases where the NPE (non-parental event) was back in the 1800s.  It means everything on that branch of their family tree has to be cut out and re-done from scratch.

Any genealogist worth their salt will tell you that every tree has its NPEs and other secrets.  DNA testing is much more accurate than paper records.

There are different types of consumer DNA testing.  Sibling tests, for example, are a waste of time and money.  But autosomal DNA testing is actually quite good.  Lots of room for improvement, sure, but already rather miraculous.

The currently accepted threshold between IBS (identical by state) and IBD (identical by descent) is around 7 cMs.  At that point, the statistical odds are extremely high that the segment match is due to a common ancestor and not just chance.  If you bump that up to about 15 cMs, you can be absolutely certain you’re dealing with a true “cousin” match.  Smaller segments can be useful for filling in the gaps when you have enough matches to triangulate your results.

The better companies (23andMe, FamilyTreeDNA) provide a lot of info to explain how their testing matches you with others, and gives you your own raw data to download so that you can check it yourself (or give to someone else who has the computational power to do so).  Each generation out is less accurate, in that the computer algorithms will miss quite a few matches because the segments have broken down to less than 7 (or 5) cMs.  Doesn’t mean the two customers aren’t related IRL, just that they inherited such different segments from their MRCA (most recent common ancestor) that they have almost no genetic commonalities between them anymore.

There are certain endogamous populations which have to be more conservative in interpreting their results because they are more inter-related than other groups.  This does not mean the results aren’t true, just that the MRCA might be a few generations further back than it seems by the level of segment matching.

Ancestry has recently gotten into the consumer DNA testing game, and is still working out the details.  They state that they will release raw data to customers in future, but for now, you get nothing but their word about who you match.  And their ethnicity breakdown is wrong more than it is right.  They need to work a lot of kinks out before they can be considered in the same category as the above two companies.

DNA testing is supported by old-fashioned paper trail genealogy to compliment and confirm.  Ultimately, it is breaking down a lot of brick walls in family trees, thanks to proving (or disproving) family lore and incomplete record-keeping.

And yes, it is helping to find additional relatives that families didn’t know they had.

lhdna,dr

I’m pretty positive I know a *lot* about my ancestors from my DNA alone. I’m absolutely certain that I come from a metazoan lineage, for example. Molecular phylogeny is a wonderful thing. 

To the original issue… I see this in the adoptive community, where parents get all excited because they have found a biological sibling for their child (with unknown biological parents). Well… probably not, unless it is an identical twin. What they get as a result is that the children are related with some probablility… but that is still not a yes/no, and without the biological parents’ DNA to match, there is absolutely no way that they could give anything else as a result. And add to this that the labs most probably don’t have any DNA samples from the area the children are from to build up a reliable database on how common different markers are in the area, even the given probability is pretty meaningless. But… people see the 90% probability they are related (which could mean second cousin, or just that they used very common markers for everybody from that area), or whatever they get from the lab, and read it as “Yes! We found a sibling!”.

But sending off your own DNA to be analyzed to try and figure out your family tree from scratch? Useless.

Why? Because DNA is really only a confirmation tool. You can’t extrapolate much about your ancestors from your own DNA alone. You can, of course, make some educated guesses about things like phenotype and ethnicity, as well as what region of the world your ancestors most likely came from, but beyond that, it’s up to historical investigation. An old photograph with a name and date written on the back can tell you far more about your family history than any isolated DNA test.