I, similarly to many on JewishGen, have attempted to wrap my headWe don't know the exact algorithm, but many of us are aware that
certain key factors are weighted highly: longest block, number of
shared blocks above a certain threshhold, and total cM of blocks shared.
You might even be able to reverse engineer the algorithm by looking at
these data for everyone you match. Its doubtful you'll get something
exact, but it might be close enough to see what goes into the
This weekend I decided to use FtDNA's chromosome browser to compareI've seen many people lately looking into these shared blocks (that
is, seeing if everybody who shares the same block on Chromosome 5 is
related somehow, etc.). I know its becoming more common to do this
in some surname projects, especially when it seem they don't relate
by YDNA (so, then, the question becomes, do they relate at all?
Maybe FF and chromosome browser can add insight).
My gut feeling is that you are right to have serious concerns! I
think all sorts of bias, and, in addition, noise (for small blocks)
can be problematic. But...16.8 cM is a moderate sized block. You
probably really do have a relatively recent MRCA with these folks.
Unfortunately, "relatively recent" when it comes to DNA can still be
hundreds of years. I certainly think it would be well worth your while
(and the while of yoru matches) to explore this more via standard
genealogy. That is, push the time limit of records, or whatever you
might find for these families.
I've had a lot of doubts over this method, but probably when the
blocks are big, and the "matchees" have something that indicates a
connection (name, geography, whatever) , it might be worth exploring.
You might also try to get more of your close and far relatives to
test. That can help in determing whether these relatives are maternal
or paternal, etc..
Steven D. Bloom