Adam,

You're correct, but as with everything else in science, accuracy will
increase with the number of data points. If your triangulation consists
of a number of known points, then you can give the results more weight.

For example, if I want to see if someone is related to me on my paternal
side or my maternal side, I'll compare them against at least 10 people
who share my paternal DNA but not my maternal side, then do the same for
my maternal size. If that person matches at least half of the people on
the paternal size and none on the maternal side, I'll feel confident
that the match is through my paternal side. I'll then repeat the
process against various paternal branches in an attempt to narrow things
down to a specific part of the paternal tree. It doesn't always work,
but I've had enough successes to continue using this method.

If a kit matches the several people on both the paternal and maternal
sides, then I know without a doubt that endogamy is at play. They're
related through both sides, probably considerably farther back than the
estimates predict.

This process can become insanely expensive, but I believe it works
fairly well (as does Lara Diamond and Israel Pickholtz). I've used it
with success to identify 3rd cousins, two of which have more
traditional genealogical evidence to corroborate the relationship. I'm
still looking for traditional sources to verify the 3rd.

Thanks,
Mindie Kaplan